Skip to main content
NCBI home page
Proceedings of the National Academy of Sciences of the United States of America logoLink to Proceedings of the National Academy of Sciences of the United States of America
. 2007 Feb 28;104(10):3925–3930. doi: 10.1073/pnas.0610642104

Systematic genome instability screens in yeast and their potential relevance to cancer

Karen W Y Yuen *,†, Cheryl D Warren ‡, Ou Chen ‡, Teresa Kwok *,†, Phil Hieter *,†,§, Forrest A Spencer ‡,¶,‖,§
PMCID: PMC1820685  PMID: 17360454

Abstract

To systematically identify genes that maintain genome structure, yeast knockout mutants were examined by using three assays that followed marker inheritance in different chromosomal contexts. These screens identified 130 null mutant strains exhibiting chromosome instability (CIN) phenotypes. Differences in both phenotype severity and assay specificity were observed. The results demonstrate the advantages of using complementary assays to comprehensively identify genome maintenance determinants. Genome structure was important in determining the spectrum of gene and pathway mutations causing a chromosome instability phenotype. Protein similarity identified homologues in other species, including human genes with relevance to cancer. This extensive genome instability catalog can be combined with emerging genetic interaction data from yeast to support the identification of candidate targets for therapeutic elimination of chromosomally unstable cancer cells by selective cell killing.

Changes in genome structure underlie many human disease states, and an important example is cancer. Changes in chromosome number or structure are commonly observed in tumors (1, 2) and many cancer cells exhibit aberrant cell architecture, including abnormal centrosomes, multipolar spindles, and breakage–fusion–bridge cycles (3, 4). Furthermore, mutations in or misregulation of genes involved in DNA damage recognition and repair, the mitotic spindle checkpoint, or proper chromosome transmission (1, 5–9) is associated with cancer development (10, 11). Genomic instability can occur early during tumorigenesis (10, 12, 13) and promotes both tumor progression and heterogeneity (14). Whether genomic instability reflects cause or effect of altered cell physiology during tumorigenesis, a comprehensive identification of genes whose mutation leads to chromosome instability [referred to as CIN genes (10)] is an important, but daunting, goal yet to be achieved. Understanding the etiology of genome instability in viable cells is fundamental to understanding the development and survival of cancers and may be instrumental in the design of therapeutic approaches that take advantage of specific vulnerabilities exhibited by cancer cells. For highly conserved biological pathways such as genome maintenance, results from model organisms can greatly facilitate functional discovery in humans.

Phenotype screening based on marker stability in budding yeast has provided a powerful approach for studying genes that act to preserve genome structure (15–22), and these genes are often functionally conserved in other eukaryotes. Genetic screens by random mutagenesis have led to identification of gene sets important for various steps in the chromosome cycle, including those functioning at kinetochores, telomeres, and origins of replication, or in microtubule dynamics, sister chromatid cohesion, DNA replication, repair, condensation and cell cycle checkpoints. All these processes must be executed at high fidelity to maintain genetic integrity. However, the random mutagenesis approach rarely achieves screen saturation because mutability varies among genes because of differences in size, base composition, and the frequency of mutable sites that can lead to viable cells with a detectable phenotype. However, the use of the gene knockout collection for Saccharomyces cerevisiae supports new and powerful strategies based on direct phenotyping of null mutants. The ≈4,700 nonessential gene-deletion mutants represent >70% of yeast genes, 30% of which remain functionally unclassified (23, 24) (Saccharomyces Genome Database (SGD); www.yeastgenome.org).

In this study, we have used the gene knockout set to carry out three systematic screens to identify genes important for maintaining genome stability in yeast (i.e., nonessential yeast CIN genes). In addition to extending the catalog of genes known to affect genome structure, several themes emerged. Because all mutants characterized are null, phenotype strength reflects the magnitude of the role played by each gene in genome stability. Thus, direct comparisons are meaningful, between different mutants in a given assay system or between different assay systems for a given mutant. We observed that some mutants exhibit phenotypes that are screen-specific. This result confirms the idea that structural context in the genome determines what pathways predominate in protecting against genomic change. Also, protein similarity searches were used to identify candidate CIN homologues in other species, including human genes with relevance to cancer. We specifically discuss a strategy that uses both the yeast CIN gene catalog and emerging yeast genetic interaction data to identify “common nodes” in synthetic lethal interaction networks based on yeast CIN genes whose human counterparts are mutated in cancers. Human homologs of these common nodes may be useful as drug targets with broad spectrum applicability for selective elimination of genomically unstable cancer cells.

Results

Genome-Wide Marker Loss Screens Identify 130 Yeast Knockout Strains.

The three screens chosen represent complementary marker loss assays. In the first screen [chromosome transmission fidelity (CTF); Fig. 1a), inheritance of an artificial chromosome fragment (CF) was monitored by using a colony color marker. Synthetic genetic array (SGA) methodology (25) was modified to construct deletion strains carrying a CF. A colony color-sectoring assay indicating chromosome instability was then performed (18, 26). Linear artificial CFs serve as sensitive indicators because their presence or absence does not affect viability, and they resemble natural chromosomes in their structure and stability (with 1.7–7 loss events per 104 divisions) (27–31). In this assay system, whole chromosome loss is the predominant mechanism observed.

Fig. 1.

Fig. 1.

Open in a new tab

Three marker loss screens. (a) Haploid yeast knockout mutants (ykoΔ) containing a chromosome fragment (CF, blue line whose centromere is depicted as a circle) and ade2-101 were generated. Red colony color is caused by accumulation of pigment because of a block in adenine production caused by the ade2-101 (ochre) mutation. This block is relieved in the presence of the SUP11 gene (blue rectangle) located on the telocentric arm of the CF, encoding an ochre-suppressing tRNATyr. Cells that contain the CF are therefore unpigmented, whereas cells that lose the CF develop red color (27–29). Colonies exhibiting unstable inheritance of this CF develop red sectors. (b) Homozygous diploid yeast knockout mutants were tested for bimater phenotype. For example, loss of the MATa allele (depicted in gray, Left) causes the development of an α-type mating cell, which is detected by its ability to mate with a MATa tester strain containing complementing auxotrophy to support selection of mated diploids. Mutant strains exhibiting unstable inheritance of the MAT locus will lose either allele in individual cell and exhibit a bimater phenotype in a population. The mutants in the squares show elevated formation of mated cells when exposed to either MATα or MATa testers (18). (c) MATα haploid yeast knockout mutants were tested for elevated frequency of ALF cells. Loss of the MATα locus in haploids results in dedifferentiation to a-mating type. The presence of these cells is detected by selection for mated products after exposure to a MATα tester strain.

In the second and third screens, an endogenous locus (the mating type locus MAT on chromosome III) was exploited as a marker. A bimater screen (designated BiM; Fig. 1b) followed inheritance of the MATa and MATα loci in homozygous diploid deletion mutants. Diploid cells heterozygous at MAT do not mate due to codominant suppression of haploid-specific cell differentiation pathways. Loss of either MATa or MATα allele results in mating competence, where mating type is determined by the remaining allele. Reciprocal mating tests with MATa or MATα mating testers were performed on the homozygous diploid deletion set to identify those which form mated products at high rates. The endogenous rate of loss of either MAT allele in wild-type cells is two to four events in 105 divisions (18, 32), where the predominant mechanism is mitotic recombination between homologs. Loss of heterozygosity in this assay can also be due to chromosome loss, chromosomal rearrangement (deletions or translocations with loss), or gene conversion (allele replacement).

In the third screen [designated a-like faker (ALF); Fig. 1c], the MATα locus inheritance was similarly followed by a mating test, but performed by using the MATα haploid deletion set. The MATα locus encodes transcription factors that suppress a-specific and promote α-specific gene expression (33). Loss of the MATα locus leads to the default mating type in yeast, which is the a-type differentiation state. Thus, MATα cells that lose the MAT locus will mate as a-type cells and are called “a-like fakers” (ALFs) (33). The frequency of ALF cells in a population is detected by prototrophic selection of mated products. In wild-type yeast, ALF mitotic segregants are generated at a rate of ≈10−6 (34, 35); C.D.W. and F.A.S., unpublished results). Mechanisms leading to ALF cells include whole chromosome loss, MAT allele disruption by chromosomal rearrangement, and gene conversion from the silent mating type locus HMRa. The relative frequencies of these events in wild-type cells are reported below.

The mutants identified in the three assays were subjected to additional validations. First, positive mutations from each primary screen were retested in knockout strains for all three assays to establish phenotype reproducibility. Three hundred ten knockout mutations were identified after secondary screening (84 CTF, 130 BiM, and 247 ALF strains). Next, the effect of cross-well contamination was evaluated by determining the identity of the knockout mutations present in each of the 310 well locations in each of the three deletion arrays [details are in supporting information (SI) Tables 3 and 4]. This validation was accomplished by sequencing the oligonucleotide “tag” unique to each deletion allele (24). The presence of >1 tag sequence or an incorrect tag sequence indicated well contamination, and the phenotypes of these locations were discarded. (The 310 well positions exhibited 22%, 9%, and 14% error in the MATa, MATα, and homozygous diploid sets, respectively). After adjustment, 293 knockout strains were verified as exhibiting CIN in at least one of the three assays. Of these 293 knockout strains, 210 (72%) were uncontaminated in all three sets. To investigate the overall error rate in each deletion set used, we sequenced strains from 60 randomly chosen well addresses and found 12%, 3%, and 3% contaminated wells in MATa, MATα, and homozygous diploid sets, respectively. The higher well contamination rate among yeast CIN mutants may reflect a high representation of slow growing yeast strains readily replaced by faster growing contaminants.

An additional source of error in deletion collection phenotyping is the occasional presence of undesired “secondary” mutations that cause the phenotype being screened; that is, mutations that are not at the site of the knockout allele. Giaever et al. (24) estimate the presence of lethal or slow-growth phenotypes caused by mutations in genes that do not segregate with a knockout allele to occur at a frequency of 6.5%. To estimate the prevalence of secondary CIN mutations, subsets of knockout mutants were regenerated by independent transformation and phenotyped. Knockout mutations identified with phenotypes in at least two screens were reconfirmed as CIN mutants in new transformants at a high rate (13/13 CTF, 9/10 BiM, 9/11 ALF). On the other hand, mutants identified with phenotypes in a single assay were reconfirmed in new transformants at a lower rate, ≈43% for the haploid collections (2/6 of mutants exhibiting CTF only, 4/8 of mutants with ALF phenotype only), and ≈75% for the diploid collection (3/4 of mutants exhibiting BiM phenotype only). These data indicate a significant frequency of secondary mutation effects in the assay-specific subsets of CIN mutants identified in the primary screens, and emphasize the validation inherent in performing screens in multiple collections. The higher reconfirmation rate of BiM from the homozygous deletion mutants is consistent with the presence of secondary mutations, which would often be covered by the wild-type allele during the construction of diploids when independent haploid segregants were mated.

In total, 130 mutants are of high confidence (the 115 deletion strains identified in more than one assay, together with 15 mutants reconfirmed independently). These 130 are listed in Fig. 2 and SI Table 3, which reflect current status of the data including all confirmations performed to date. The remaining 163 mutants identified in only one screen are listed in SI Table 4. These 163 are regarded with lower confidence, containing an estimated ≈43% and ≈75% of true positives for haploid and diploid mutant screens, respectively. SI Tables 3 and 4 report measures of phenotype severity, as well as annotations for well contamination in any of the 3 deletion sets. It is important to note that the error rates described relate to the specific copies of the MATa, MATα, and MATa/MATα deletion collections that were obtained by our laboratories from commercial distribution sources. In addition to our manipulations, these collections are related to the original consortium collections by an unknown sequence of replication and outgrowth steps.

Fig. 2.

Fig. 2.

Open in a new tab

One hundred thirty nonessential yeast CIN genes. The small diagram (Upper Left) depicts the distribution of the 293 mutants identified initially across the three screens. The numbers in parentheses denote single-assay knockouts confirmed in independent transformants, which are included in the detailed diagram to the right. The detailed diagram (Right) summarizes the 130 high confidence genes described in SI Table 3. (For the other 163 genes, see SI Table 4). All gene names are connected to one or more of the three major nodes, indicating the screen phenotypes for which each knockout was positive (CTF, BiM, or ALF). Gene names in black typeface are those validated in all three deletion arrays: i.e., tag sequencing indicated the presence of the correct mutation in uncontaminated form. For these mutants, present and absent phenotypes are meaningful. Gene names in blue italic typeface failed tag validation in at least one of the deletion collections, and therefore phenotype information is missing from at least one screen. These partially characterized genes are placed to indicate observed tag-sequence validated phenotypes (SI Table 3 contains details). The node colors indicate biological process. Genes associated with more than one biological process are represented by the one highest in the color key for simplicity.

Functional Distribution of Yeast CIN Genes.

Gene Ontology (GO) annotations among these 293 CIN genes in comparison to the entire yeast genome (36) indicated enrichment in numerous expected cellular components: nucleus, chromosome, kinetochore, microtubule, cytoskeleton, spindle, nuclear pore, spindle pole body, replication fork, and chromatin (SI Table 5). The CIN gene list is enriched in the GO biological processes of cell cycle, cell proliferation, response to DNA damage response, and nuclear division (SI Table 6). These GO annotations reflect current knowledge of studied genes, indicating that the screens identified genes known to be functioning in genome maintenance. Interestingly, genes not previously known to contribute to stability were also identified. For example, seven yeast mutants in the adenine biosynthetic pathway (ade1, ade2, ade4, ade5/7, ade6, ade8, and ade17) gave rise to elevated ALFs at frequencies ranging from 2- to 31-fold above wild-type (SI Tables 3 and 4). Three of these mutants (ade1, ade6, and ade17 shown in Fig. 2) were validated in fresh transformants. This results indicates that cellular adenine pathway intermediates, or derivative metabolites, are important for genome stability, and that compensatory mechanisms used by cells when de novo synthesis is blocked are not fully sufficient. In addition, several CIN genes identified in the screens were recently characterized to play a role in genomic stability. Examples are Dia2 [an F-box protein in the SCF E3 ubiquitin ligase complex (37, 38)], Nce4 [which associates with the Sgs1-Top3 helicase-topoimerase complex (39, 40)], and Mms22 [which functions with Mms1 in a DNA damage repair pathway (41)]. Integrating the CIN gene catalog with other phenotypic, genetic, and physical interaction data available in yeast proves to be a fruitful avenue to further our understanding in mechanisms for genomic stability.

Chromosome Loss Is the Major Mechanism of MATα Loss in a-Like Fakers.

The CTF and BiM phenotypes have been widely used to study genome instability. However, the ALF phenotype has been only rarely used (32, 42, 43) and has not been well characterized. The electrophoretic karyotype of mated colonies obtained after selection can be analyzed to infer the mechanism of MATα locus loss. The chromosome III in the mating tester was visually differentiated from that in knockout strains by pulsed-field gel electrophoresis (Fig. 3A). Hybridization with a probe that bound 3 distant sites on chromosome III (the MAT locus, and silent cassettes located distally on each arm) allowed detection of the two parental chromosome III bands as well as aberrant chromosome III derivatives. Aberrant chromosomes were observed in a variety of sizes. These chromosomes included an expected 200 kb product likely to represent homologous recombination between MATα and silent locus HMRa. This event is known to generate an active MATa locus concomitant with a large deletion on the right arm of chromosome III (reviewed in ref. 35).

Fig. 3.

Fig. 3.

Open in a new tab

A-like fakers result from whole chromosome loss, gross chromosomal rearrangement, and gene conversion. (A) Electrophoretic karyotypes of ALF mated products were examined for chromosome III status as described (42), and examples are shown. Individual colonies selected after mating were characterized by using pulsed-field gel electrophoresis (top, ethidium bromide stained gel) and in-gel hybridization with a radiolabeled probe (bottom, autoradiogram) that hybridizes chromosomal bands containing the mating type locus and/or silent mating type loci located distally on each arm. Chromosomes III from the mating tester and deletion mutant were of distinct size (top and bottom chr III bands, respectively). In some strains, a less intense signal for rearrangement chromosomes reflects poor mitotic transmission, or hybridization only to HMRa which has imperfect homology to the radiolabeled probe. (B) Discordant CTF sectoring phenotypes are observed in knockout mutations with similar ALF frequencies.

Electrophoretic karyotypes of mated products from wild-type cells indicated that 68% of events were due to whole chromosome loss, 20% to chromosomal rearrangement, and 12% to gene conversion (Table 1). We analyzed independent mated colonies for 13 high-frequency ALF mutants. In 11, loss of whole chromosome III was the predominant event, similar to wild-type. Statistically significant exceptions were observed for rad27Δ and sov1Δ, which showed predominant chromosome rearrangement or an intact chromosome III, respectively. RAD27 encodes an endonuclease that promotes Okazaki fragment maturation during DNA replication. The ALF-associated rearrangements are consistent with previous observation that RAD27 protects against gross chromosomal rearrangements (44). SOV1 is a reserved name (SGD) for a nuclear gene important for respiration whose encoded protein may localize to mitochondria from high-throughput studies (45). Its gene conversion phenotype was unusual. To further define the events giving rise to ALFs, PCR was used to detect the presence of MATa and MATα loci in the mated products (46). Interestingly, all mated products from the sov1Δ mutant contained both MATa and MATα loci, indicating introduction of the MATa allele into MAT by gene conversion. This was not the general pattern observed in wild-type or in other mutants, where only 3% (1/39) or 6% (24/386) of isolates tested were of this type, respectively.

Table 1.

Summary of electrophoretic karyotypes from 13 ALF mutants

Strain ALF fold-change Fraction ykoΔ chr III loss Fraction ykoΔ chr III GCR Fraction ykoΔ chr III retained
Wild-type 1 0.68 0.20 0.12
rad27Δ 63 0.30 0.60 0.10
dia2Δ 60 0.70 0.30 0.00
ybr113wΔ 56 1.00 0.00 0.00
nce4Δ 56 0.70 0.20 0.10
xrs2Δ 48 0.86 0.14 0.00
esc2Δ 43 0.50 0.30 0.20
top3Δ 42 0.60 0.40 0.00
kar3Δ 40 1.00 0.00 0.00
rad50Δ 36 0.64 0.29 0.07
sic1Δ 31 0.70 0.30 0.00
ade1Δ 31 0.70 0.20 0.10
oma1Δ 30 0.80 0.10 0.10
sov1Δ 8 0.00 0.00 1.00
Open in a new tab

ALF frequency is shown as fold over wild-type. Event percentages (chr III loss, gross chromosome rearrangement (GCR), or retention of normal structure) are calculated from independent wild-type or mutant mated products (n = 40 and n ≥ 10, respectively). The outcome distributions for sov1Δ and rad27Δ are significantly different from wild type (χ2, P < 0.01).

Interestingly, some high-frequency ALFs that showed whole chromosome III loss failed to exhibit a sectoring phenotype in the CTF screen. Of 13 frequent ALF mutants analyzed in Fig. 3, only 5 were identified for CTF phenotypes by the high throughput screen: 3 with strong (kar3Δ, sic1Δ, and dia2Δ) and 2 with weak (rad27Δ, nce4Δ) phenotypes. To confirm the presence of assay difference, 5 frequent ALF mutants were directly retested for the CTF phenotype in fresh transformants. Two of these mutants (kar3Δ, sic1Δ) exhibited a strong CTF phenotype as expected, and 3 were confirmed to show mild or absent sectoring (esc2Δ, rad50Δ, xrs2Δ; Fig. 3B). Thus, frequent ALF production does not strictly correlate with frequent CF loss. This result may indicate that different factors influence stable inheritance of endogenous chromosome III and the CF. One explanation is that the telocentric structure of the CF may enhance instability in some mutants. Another is that the presence of a partial homologous chromosome may suppress instability. Further work will be required to determine the underlying biological mechanisms that explain these uncorrelated phenotypes.

Many Yeast CIN Genes Are Conserved.

Current understanding of mechanisms that contribute to genome stability has been largely fueled by work from model systems. This approach has been informative for human biology because of remarkable functional conservation within the chromosome cycle. To evaluate conservation of yeast CIN genes, BLASTp searches using yeast amino acid sequences against proteomes from Schizosaccharomyces pombe, Caenorhabditis elegans, Drosophila melanogaster, Mus musculus, and Homo sapiens were performed. Among the 293 yeast CIN genes, 103 (35%) had homologs with e-values ≤10−10 in all five organisms searched. SI Table 7 contains alignment results and links to functional summaries. Previously published work showed that ≈40% of yeast proteins are conserved in eukaryotic evolution (47), and 30% of known genes involved in human disease have yeast homologs (48). In agreement, 42% of yeast CIN genes have homologs in human with e-values ≤10−10. Human homologs of yeast CIN genes represent candidates that may cause a CIN phenotype when mutated. In the high confidence CIN gene list, there were 10 examples of human homologs (“top hit” BLASTp e-values <10−10) that have been previously shown to exhibit somatic mutation in cancer (Table 2 and SI Table 8).

Table 2.

Human proteins homologous to yeast CIN genes are mutated in cancer

Yeast gene Top human hit e-value Disease description MIM no. (gene) Source
ADE17 ATIC 0 Anaplastic large cell lymphoma CC
RAD54 RAD54L 1E-164 Non-Hodgkin lymphoma; breast cancer; colon adenocarcinoma 603615 OMIM
RAD51 RAD51 1E-122 Susceptibility to breast cancer 179617 OMIM
RDH54 RAD54B 1E-121 Non-Hodgkin lymphoma; colon adenocarcinoma 604289 OMIM
SGS1 BLM 1E-115 Bloom syndrome 604610 OMIM, CC
MRE11 MRE11A 1E-108 Ataxia-telangiectasia-like disorder 600814 OMIM
DUN1 CHK2 6E-55 Li-Fraumeni syndrome; osteosarcoma; susceptibility to breast cancer; prostate cancer; susceptibility to colorectal cancer 604373 OMIM
BUB1 BUB1 1E-41 Colorectal cancer with chromosomal instability 602452 OMIM
MAD1 MAD1 5E-12 Lymphoma; prostate cancer 602686 OMIM
CDC73 HRPT2 9E-12 Hyperparathyroidism-jaw tumor syndrome; hyperparathyroidism; parathyroid adenoma with cystic changes 607393 OMIM
Open in a new tab

The protein sequences corresponding to 130 high-confidence yeast CIN genes were used as queries in a BLASTP search against the human RefSeq protein database. Online Mendelian Inheritance in Man (OMIM; www.ncbi.nlm.nih.gov/omim) and cancer census (CC; ref. 58) databases were used to identify cancer associated mutations in “top hit” human genes.

Discussion

This work identified an extensive catalog of genome instability mutants based on phenotype tests of haploid and diploid yeast knockout collections for CTF, elevated ALF frequency, and BiM behavior. Nonessential yeast genes were characterized because of their accessibility for phenotyping. Because many essential genes are known to contribute to genome stability from traditional approaches, a similar systematic screening effort for these genes would be of great interest but will first require the development of a comprehensive hypomorphic mutation resource.

An extensive catalog is useful for understanding mechanisms that maintain or alter genome structure, for the identification of new pathways important for genome maintenance, and for the organization of functional networks. Systematic screening of arrayed nonessential mutants avoids the sampling problem in traditional mutagenesis methods, and supports the direct comparison of phenotypes observed because all alleles are null. As examples, rad27Δ, dia2Δ, nce4Δ, and xrs2Δ exhibited the strongest ALF phenotypes (>56-fold above wild type) among the high confidence yeast CIN genes, whereas well-studied damage response genes such as mec3Δ, mrc1Δ, ddc1Δ, and rad9Δ showed milder phenotypes (≈11-fold). Apparently, under the growth conditions used in the screen, damage caused by absence of Rad27, Dia2, Nce4, or Xrs2 proteins exceeds that resulting from checkpoint loss. In addition to phenotype comparisons within a given assay, results from different assays can be compared. For example, xrs2Δ exhibited one of the highest ALF frequencies but a mild or absent CTF phenotype, indicating that the damage associated with xrs2Δ is more relevant to the maintenance of a haploid chromosome III than to the artificial CF. In general terms, different chromosome marker stability assays (CTF, ALF, and BiM) defined both distinct and overlapping gene sets. The screen specificities likely reflect functional distinctions revealed by the assay systems and deserve further study as indicators of in vivo roles played by genes that safeguard genome structure.

The errors observed in the deletion arrays underscore the importance of mutant validation. It is widely known that mutant arrays accumulate change because of manipulation and selective pressure (i.e., cross-well contamination, aneuploidy, second site mutation, etc.), but parameters indicative of array quality are under-reported in published studies. This issue is increasingly important as phenotypic data derived from distantly related replicates of the deletion resource are compared and integrated. In this study, tag sequence analysis of CIN mutant strains suggests false negative observations from well contamination were between 9 and 22% in different screens (details in SI Table 3). An empirical measure agrees: the CTF screen of the knockout collection identified 12 of 15 nonessential ctf mutants found previously in a traditional mutagenesis (18). We can also estimate the false positive phenotype detection frequency because of secondary mutations in the deletion collection strains. For the haploid collections, the overall false positive rates were relatively high (≈18% and 31% for the MATa and MATα collections, respectively). For the diploid collection, the false positive rate was lower (≈5%). These frequencies are consistent with observation of unlinked recessive lethal mutations segregating independently of the deletion mutations created during construction of the mutant resource (24). In this study, false-positive observations were rare among genes identified in more than one marker loss assay (and therefore by mutants in more than one deletion resource), emphasizing the validation that is inherent in screening more than one collection for a mutant phenotype. We therefore partitioned our results into 130 high confidence genes (115 genes identified by more than one screen, plus 15 genes confirmed in new transformants), and 163 lower confidence genes identified by single screens only. Our data indicate that the lower confidence gene set contains many true positive observations, which we roughly estimate as ≈33% and 75% of the lower confidence haploid and diploid knockouts, respectively.

A full catalog of yeast CIN genes will provide a rich resource for studies of genomic instability in many organisms, including human. Inclusion of additional screens of nonessential yeast mutants (e.g., refs. 21 and 22) and systematic incorporation of essential mutants will enhance the utility of the yeast model system. Yeast CIN genes define cross-species candidate genes in humans that could be mutated or misregulated during tumorigenesis. A recent survey of CIN colorectal tumors (49) provides a stunning proof of principle. One hundred human candidate genes (chosen for similarity to yeast and fly CIN genes) were screened for mutations in a panel of colorectal tumor samples. This study identified five human CIN genes mutated in cancer (hMRE11, hZW10, hZwilch, hRod, and hDing) adding to two previously known (hCDC4 and hBub1; reviewed in ref. 50). These seven CIN genes account for <20% of the CIN mutational spectrum in colon cancer, and many other candidate CIN genes remain untested. Indeed, 103 human homologs of yeast CIN genes (based in part on this work) were tested for somatic mutations in a panel of colorectal cancer samples, identifying eight human CIN genes mutated in colon cancer (K.W.Y.Y., T. Barber, M. Reis, K. McManus, F.A.S., B. Vogelstein, V. Velculescu, P.H., and C. Lengauer, unpublished results). These studies demonstrate that systematic analysis of the mutational spectrum leading to a CIN phenotype in a model organism such as yeast will help to define the mutational spectrum leading to a CIN phenotype in human cancer and may accelerate the development of effective cancer therapies.

Knowledge of the mutational spectrum of CIN genes in human cancer has several important practical implications. In particular, if a defined subset of CIN genes represents the major mutational target in a specific tumor type, the spectrum of mutations might provide a rationale for therapeutic design. The specific class of mutations causing CIN may define a genetic “Achilles heel” that would provide selective killing of tumor cells relative to adjacent normal cells. In this strategy, genetic interactions resulting in cell lethality may be useful for the design of therapeutic approaches where the altered genotype of a cancer cell is used to leverage its specific vulnerability. One kind of genetic interaction with properties useful for this strategy is synthetic lethality, observed when two mutations individually capable of supporting viability cause cell death when present together. Synthetic lethal mutant pairs identify genes that function in parallel or related pathways that cannot be simultaneously lost. Following this logic, cancer cells with a specific CIN mutation can be killed through loss of function of a synthetic lethal partner, while sparing normal cells (51, 52). Large-scale, systematic synthetic lethality analysis in yeast provides a means for identifying such second-site loss of function mutations (53–55). The budding yeast studies suggest candidate human proteins whose inhibition (e.g., by a drug) may specifically kill tumor cells relative to normal cells. In this regard, gene deletions that exhibit synthetic lethality with multiple different CIN gene mutants are particularly attractive, as they might define broad-spectrum therapeutic targets.

To address this concept, we analyzed all known synthetic lethal interactions available for yeast CIN genes from Table 2. Eight of the 10 genes had published synthetic lethal data. The corresponding eight deletion mutations were connected to 250 partners by 371 synthetic lethal interactions based on BioGrid (56) (data not shown). Among the 250 partners, 61 bridged at least two yeast cancer homologs (Fig. 4). Notably, three mutants exhibited synthetic lethality, with at least six cancer gene homologs. Interestingly, these three yeast genes share a role in sister chromatid cohesion (42, 57). The “hub” position of these three mutants in the interaction network represents a common set of genetic vulnerabilities that may suggest broad spectrum targets.

Fig. 4.

Fig. 4.

Open in a new tab

Common synthetic lethal interactions among yeast CIN genes that have human homologs mutated in cancer. Eight yeast CIN genes with top hit human homologs mutated in cancer (e-value <10−10) are also found in the public interaction database BioGrid (56). These 8 yeast CIN genes are placed peripherally and are shown in black. There were 61 interactors with at least two synthetic lethal connections to the yeast cancer homologs. The arrows point from a query to a target gene hit in the synthetic lethal screens. The interactors include high-confidence CIN genes (blue), low-confidence CIN genes (purple), and genes not detected by the CIN screens (gray). The three genes that have six common synthetic lethal interactions with the cancer gene homologs are indicated by their blue connections. Node color indicates biological process as in Fig. 2.

The extant synthetic lethal data set in budding yeast is continuously expanding (53), and more “common nodes” may be identified. Combining these data with an increased understanding of the mutation spectrum in cancers could provide insights central to the design of therapeutic approaches in which human cancer cells are efficiently targeted for death by clinical intervention. Integration of knowledge among emerging high throughput data sets in model organisms will stimulate new research directions and solutions to challenges in combating human disease.

Materials and Methods

Yeast Strains.

Complete genotypes are given in SI Table 9.

CTF Screen.

SGA selection methods (25) were used to introduce the ade2-101 mutation and a CF into MATa deletion mutants obtained from Research Genetics (www.resgen.com). Donor strains (MATα ade2-101::NatMX ura3 his3 can1Δ mfa1Δ::MFA1pr-HIS3) containing SUP11-marked CFs from chromosome III or VII were mated with yeast deletion mutants (MATa ura3 his3 ykoΔ::G418R) to obtain diploids. After sporulation, NatR G418R Ura+ His+ CanR haploids were tested for stability of the CF as in ref. 18. For additional detail, see SI Fig. 5.

Bimater Screen.

A homozygous diploid deletion set obtained from Open Biosystems (www.openbiosystems.com) in 96-array format was used. Histidine prototrophs were selected by replica plating mutants exposed to MATa his1 and MATα his1 tester strains on solid medium lacking histidine. Mutants with elevated mating frequency detected with both testers were scored as BiM. Details are provided in SI Fig. 6.

a-Like Faker Screen.

The presence of a-type mating cells in strains from the MATα haploid knockout collection was detected as described (42). A deletion collection obtained from Research Genetics (MATα his3) was manually arrayed in 1-cm2 patches and mated to a MATa his1 tester lawn by replica plating. His+ prototrophs were selected on synthetic complete medium lacking histidine, uracil, lysine, adenine, tryptophan, and leucine. Details are provided in SI Fig. 7.

Electrophoretic Karyotypes.

Sample preparation, pulsed field gel analysis, and in-gel hybridizations were performed as described (42).

Supplementary Material

Supporting Information
pnas_0610642104_index.html (909B, html)

Acknowledgments

We thank J. Blumberg for technical development of the ALF screen; D. Yuan, D. Church, A. Pandey, J. Amberger, Q. Shi, J. Fox, and Y. Huang for discussions; and J. Boeke, C. V. Dang, D. Koshland, C. Lengauer, J. Mendell, K. D. Smith, M. Vuica, and X. C. Zhao for comments on the manuscript. This work was supported by grants from the National Institute of General Medical Sciences (to F.A.S.) and by the National Cancer Institute and the Canadian Institutes of Health Research (to P.H.). K.Y. was supported by a fellowship from the Natural Sciences and Engineering Research Council (Canada).

Abbreviations

CIN

chromosome instability

CTF

chromosome transmission fidelity

BiM

bimater

ALF

a-like faker

CF

chromosome fragment.

Footnotes

The authors declare no conflict of interest.

This article contains supporting information online at www.pnas.org/cgi/content/full/0610642104/DC1.

References

  • 1.Cahill DP, Lengauer C, Yu J, Riggins GJ, Willson JK, Markowitz SD, Kinzler KW, Vogelstein B. Nature. 1998;392:300–303. doi: 10.1038/32688. [DOI] [PubMed] [Google Scholar]
  • 2.Rajagopalan H, Nowak MA, Vogelstein B, Lengauer C. Nat Rev Cancer. 2003;3:695–701. doi: 10.1038/nrc1165. [DOI] [PubMed] [Google Scholar]
  • 3.Saunders WS, Shuster M, Huang X, Gharaibeh B, Enyenihi AH, Petersen I, Gollin SM. Proc Natl Acad Sci USA. 2000;97:303–308. doi: 10.1073/pnas.97.1.303. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Gisselsson D. Adv Cancer Res. 2003;87:1–29. doi: 10.1016/s0065-230x(03)87164-6. [DOI] [PubMed] [Google Scholar]
  • 5.Jin DY, Spencer F, Jeang KT. Cell. 1998;93:81–91. doi: 10.1016/s0092-8674(00)81148-4. [DOI] [PubMed] [Google Scholar]
  • 6.Michel LS, Liberal V, Chatterjee A, Kirchwegger R, Pasche B, Gerald W, Dobles M, Sorger PK, Murty VV, Benezra R. Nature. 2001;409:355–359. doi: 10.1038/35053094. [DOI] [PubMed] [Google Scholar]
  • 7.Zou H, McGarry TJ, Bernal T, Kirschner MW. Science. 1999;285:418–422. doi: 10.1126/science.285.5426.418. [DOI] [PubMed] [Google Scholar]
  • 8.Rajagopalan H, Jallepalli PV, Rago C, Velculescu VE, Kinzler KW, Vogelstein B, Lengauer C. Nature. 2004;428:77–81. doi: 10.1038/nature02313. [DOI] [PubMed] [Google Scholar]
  • 9.Dove W. Nat Genet. 2003;34:353–354. doi: 10.1038/ng0803-353. [DOI] [PubMed] [Google Scholar]
  • 10.Vogelstein B, Kinzler KW. Nat Med. 2004;10:789–799. doi: 10.1038/nm1087. [DOI] [PubMed] [Google Scholar]
  • 11.Hanks S, Coleman K, Reid S, Plaja A, Firth H, Fitzpatrick D, Kidd A, Mehes K, Nash R, Robin N, et al. Nat Genet. 2004;36:1159–1161. doi: 10.1038/ng1449. [DOI] [PubMed] [Google Scholar]
  • 12.Shih IM, Zhou W, Goodman SN, Lengauer C, Kinzler KW, Vogelstein B. Cancer Res. 2001;61:818–822. [PubMed] [Google Scholar]
  • 13.Chin K, de Solorzano CO, Knowles D, Jones A, Chou W, Rodriguez EG, Kuo WL, Ljung BM, Chew K, Myambo K, et al. Nat Genet. 2004;36:984–988. doi: 10.1038/ng1409. [DOI] [PubMed] [Google Scholar]
  • 14.Jallepalli PV, Lengauer C. Nat Rev Cancer. 2001;1:109–117. doi: 10.1038/35101065. [DOI] [PubMed] [Google Scholar]
  • 15.Maine GT, Sinha P, Tye BK. Genetics. 1984;106:365–385. doi: 10.1093/genetics/106.3.365. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Kouprina N, Pashina OB, Nikolaishwili NT, Tsouladze AM, Larionov VL. Yeast. 1988;4:257–269. doi: 10.1002/yea.320040404. [DOI] [PubMed] [Google Scholar]
  • 17.Hoyt MA, Stearns T, Botstein D. Mol Cell Biol. 1990;10:223–234. doi: 10.1128/mcb.10.1.223. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Spencer F, Gerring SL, Connelly C, Hieter P. Genetics. 1990;124:237–249. doi: 10.1093/genetics/124.2.237. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Meeks-Wagner D, Wood JS, Garvik B, Hartwell LH. Cell. 1986;44:53–63. doi: 10.1016/0092-8674(86)90484-8. [DOI] [PubMed] [Google Scholar]
  • 20.Ouspenski II, Elledge SJ, Brinkley BR. Nucleic Acids Res. 1999;27:3001–3008. doi: 10.1093/nar/27.15.3001. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Huang ME, Rio AG, Nicolas A, Kolodner RD. Proc Natl Acad Sci USA. 2003;100:11529–11534. doi: 10.1073/pnas.2035018100. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Smith S, Hwang JY, Banerjee S, Majeed A, Gupta A, Myung K. Proc Natl Acad Sci USA. 2004;101:9039–9044. doi: 10.1073/pnas.0403093101. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Winzeler EA, Shoemaker DD, Astromoff A, Liang H, Anderson K, Andre B, Bangham R, Benito R, Boeke JD, Bussey H, et al. Science. 1999;285:901–906. doi: 10.1126/science.285.5429.901. [DOI] [PubMed] [Google Scholar]
  • 24.Giaever G, Chu AM, Ni L, Connelly C, Riles L, Veronneau S, Dow S, Lucau-Danila A, Anderson K, Andre B, et al. Nature. 2002;418:387–391. doi: 10.1038/nature00935. [DOI] [PubMed] [Google Scholar]
  • 25.Tong AH, Evangelista M, Parsons AB, Xu H, Bader GD, Page N, Robinson M, Raghibizadeh S, Hogue CW, Bussey H, et al. Science. 2001;294:2364–2368. doi: 10.1126/science.1065810. [DOI] [PubMed] [Google Scholar]
  • 26.Hieter P, Mann C, Snyder M, Davis RW. Cell. 1985;40:381–392. doi: 10.1016/0092-8674(85)90152-7. [DOI] [PubMed] [Google Scholar]
  • 27.Gerring SL, Spencer F, Hieter P. EMBO J. 1990;9:4347–4358. doi: 10.1002/j.1460-2075.1990.tb07884.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Hegemann JH, Shero JH, Cottarel G, Philippsen P, Hieter P. Mol Cell Biol. 1988;8:2523–2535. doi: 10.1128/mcb.8.6.2523. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Warren CD, Brady DM, Johnston RC, Hanna JS, Hardwick KG, Spencer FA. Mol Biol Cell. 2002;13:3029–3041. doi: 10.1091/mbc.E02-04-0203. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Koshland D, Hieter P. Methods Enzymol. 1987;155:351–372. doi: 10.1016/0076-6879(87)55024-8. [DOI] [PubMed] [Google Scholar]
  • 31.Shero JH, Koval M, Spencer F, Palmer RE, Hieter P, Koshland D. Methods Enzymol. 1991;194:749–773. doi: 10.1016/0076-6879(91)94057-j. [DOI] [PubMed] [Google Scholar]
  • 32.Liras P, McCusker J, Mascioli S, Haber J. Genetics. 1978;88:651–671. [PMC free article] [PubMed] [Google Scholar]
  • 33.Strathern J, Hicks J, Herskowitz I. J Mol Biol. 1981;147:357–372. doi: 10.1016/0022-2836(81)90488-5. [DOI] [PubMed] [Google Scholar]
  • 34.Herskowitz I. Microbiol Rev. 1988;52:536–553. doi: 10.1128/mr.52.4.536-553.1988. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Herskowitz I. Genetics. 1988;120:857–861. doi: 10.1093/genetics/120.4.857. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Harris MA, Clark J, Ireland A, Lomax J, Ashburner M, Foulger R, Eilbeck K, Lewis S, Marshall B, Mungall C, et al. Nucleic Acids Res. 2004;32:D258–D261. doi: 10.1093/nar/gkh036. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Koepp DM, Kile AC, Swaminathan S, Rodriguez-Rivera V. Mol Biol Cell. 2006;17:1540–1548. doi: 10.1091/mbc.E05-09-0884. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Blake D, Luke B, Kanellis P, Jorgensen P, Goh T, Penfold S, Breitkreutz BJ, Durocher D, Peter M, Tyers M. Genetics. 2006;174:1709–1727. doi: 10.1534/genetics.106.057836. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Chang M, Bellaoui M, Zhang C, Desai R, Morozov P, Delgado-Cruzata L, Rothstein R, Freyer GA, Boone C, Brown GW. EMBO J. 2005;24:2024–2033. doi: 10.1038/sj.emboj.7600684. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Mullen JR, Nallaseth FS, Lan YQ, Slagle CE, Brill SJ. Mol Cell Biol. 2005;25:4476–4487. doi: 10.1128/MCB.25.11.4476-4487.2005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41.Baldwin EL, Berger AC, Corbett AH, Osheroff N. Nucleic Acids Res. 2005;33:1021–1030. doi: 10.1093/nar/gki246. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42.Warren CD, Eckley DM, Lee MS, Hanna JS, Hughes A, Peyser B, Jie C, Irizarry R, Spencer FA. Mol Biol Cell. 2004;15:1724–1735. doi: 10.1091/mbc.E03-09-0637. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Lemoine FJ, Degtyareva NP, Lobachev K, Petes TD. Cell. 2005;120:587–598. doi: 10.1016/j.cell.2004.12.039. [DOI] [PubMed] [Google Scholar]
  • 44.Chen C, Kolodner RD. Nat Genet. 1999;23:81–85. doi: 10.1038/12687. [DOI] [PubMed] [Google Scholar]
  • 45.Sickmann A, Reinders J, Wagner Y, Joppich C, Zahedi R, Meyer HE, Schonfisch B, Perschil I, Chacinska A, Guiard B, et al. Proc Natl Acad Sci USA. 2003;100:13207–13212. doi: 10.1073/pnas.2135385100. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46.Huxley C, Green ED, Dunham I. Trends Genet. 1990;6:236. doi: 10.1016/0168-9525(90)90190-h. [DOI] [PubMed] [Google Scholar]
  • 47.Rubin GM, Yandell MD, Wortman JR, Gabor Miklos GL, Nelson CR, Hariharan IK, Fortini ME, Li PW, Apweiler R, Fleischmann W, et al. Science. 2000;287:2204–2215. doi: 10.1126/science.287.5461.2204. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 48.Bassett DE, Jr, Boguski MS, Spencer F, Reeves R, Kim S, Weaver T, Hieter P. Nat Genet. 1997;15:339–344. doi: 10.1038/ng0497-339. [DOI] [PubMed] [Google Scholar]
  • 49.Wang Z, Cummins JM, Shen D, Cahill DP, Jallepalli PV, Wang TL, Parsons DW, Traverso G, Awad M, Silliman N, et al. Cancer Res. 2004;64:2998–3001. doi: 10.1158/0008-5472.can-04-0587. [DOI] [PubMed] [Google Scholar]
  • 50.Yuen KW, Montpetit B, Hieter P. Curr Opin Cell Biol. 2005;17:576–582. doi: 10.1016/j.ceb.2005.09.012. [DOI] [PubMed] [Google Scholar]
  • 51.Hartwell LH, Szankasi P, Roberts CJ, Murray AW, Friend SH. Science. 1997;278:1064–1068. doi: 10.1126/science.278.5340.1064. [DOI] [PubMed] [Google Scholar]
  • 52.Kaelin WG., Jr Nat Rev Cancer. 2005;5:689–698. doi: 10.1038/nrc1691. [DOI] [PubMed] [Google Scholar]
  • 53.Tong AH, Lesage G, Bader GD, Ding H, Xu H, Xin X, Young J, Berriz GF, Brost RL, Chang M, et al. Science. 2004;303:808–813. doi: 10.1126/science.1091317. [DOI] [PubMed] [Google Scholar]
  • 54.Pan X, Ye P, Yuan DS, Wang X, Bader JS, Boeke JD. Cell. 2006;124:1069–1081. doi: 10.1016/j.cell.2005.12.036. [DOI] [PubMed] [Google Scholar]
  • 55.Ooi SL, Pan X, Peyser BD, Ye P, Meluh PB, Yuan DS, Irizarry RA, Bader JS, Spencer FA, Boeke JD. Trends Genet. 2006;22:56–63. doi: 10.1016/j.tig.2005.11.003. [DOI] [PubMed] [Google Scholar]
  • 56.Stark C, Breitkreutz BJ, Reguly T, Boucher L, Breitkreutz A, Tyers M. Nucleic Acids Res. 2006;34:D535–D539. doi: 10.1093/nar/gkj109. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 57.Mayer ML, Pot I, Chang M, Xu H, Aneliunas V, Kwok T, Newitt R, Aebersold R, Boone C, Brown GW, Hieter P. Mol Biol Cell. 2004;15:1736–1745. doi: 10.1091/mbc.E03-08-0619. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 58.Futreal PA, Coin L, Marshall M, Down T, Hubbard T, Wooster R, Rahman N, Stratton MR. Nat Rev Cancer. 2004;4:177–183. doi: 10.1038/nrc1299. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supporting Information
pnas_0610642104_index.html (909B, html)
pnas_0610642104_4.pdf (172.7KB, pdf)
pnas_0610642104_10642Tables_3-9.xls (1.4MB, xls)
pnas_0610642104_1.pdf (1.9MB, pdf)
pnas_0610642104_2.pdf (1.9MB, pdf)
pnas_0610642104_3.pdf (1.9MB, pdf)

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

RESOURCES