Chromosome-Scale Genetic Mapping Using a Set of 16 Conditionally Stable Saccharomyces cerevisiae Chromosomes

Reagents and yeast media:

G418 was purchased from MediaTech (Herndon, VA). Five (5)-fluoroorotic acid (5-FOA) was purchased from American Bioanalytical (Natick, MA). Taq DNA polymerase was purchased from Continental Laboratory Products (San Diego).

Yeast extract–peptone–dextrose (YPD) medium, synthetic complete (SC) medium, synthetic dropout (e.g., SC −trp) media and 5-FOA medium were prepared as described (Sherman et al. 1986) except that 60 μg/ml of leucine were added to synthetic plates where appropriate. Sporulation medium was prepared as in Klapholz and Esposito (1982). Lithium acetate (LiOAc) transformations were performed as described (Schiestl and Gietz 1989).

CEN-UG plasmid construction:

A PCR product with the Kluyveromyces lactis URA3 gene was cloned into the BglII and EcoRI sites in pUK21 (Vieira and Messing 1991) to generate plasmid pUK21-KlUra3 (M3739), and a PCR product with the GAL1 promoter was cloned into the BamHI and XbaI sites in plasmid pUC21 to generate plasmid pUC21-GAL (M3738). A FseI–EagI fragment from pUC21-GAL containing the GAL1 promoter was then cloned into pUK21-KlUra3 to generate pUK21-KlUGAL (M3756). A SacI–BssHII fragment from pUK21-KlUra3 was then cloned into pUC21-NotI (Voth et al. 2001) to generate pKlUGAL-NotI (M3828) containing the K. lactis URA3 gene and the pGAL promoter flanked by multiple cloning site cassettes. CEN-proximal DNA sequences (“left of CEN” and “right of CEN”) were PCR amplified from each yeast chromosome for cloning into plasmid pKlUGAL-NotI (see supplemental Table S1 for primer sequences). The PCR primers used to amplify left of CEN DNA (∼800–1200 bp DNA to the immediate left of each centromere), added terminal SacI and SfiI restriction sites. Likewise, PCR primers used to amplify right of CEN DNA (∼800–1200 bp to the right of the centromere) added BssHII and BsiWI restriction sites. The left and right DNA fragments were then digested at the primer-encoded restriction sites for cloning into pKlUGAL-NotI to generate the pCEN-UG plasmid for 15 of the yeast chromosomes (Figure 1A). These plasmids each contain a specific centromere region interrupted by a K. lactis URA3 gene and the GAL1 promoter (Figure 1A). Integrating DNA fragments were liberated from the plasmids by NotI digestion before transformation into strain W303-1A to replace the native centromere. The CEN-proximal insertion for chromosome 14 was not isolated after two attempts using the plasmid-based clone. We therefore moved the GAL1 promoter insertion to the other side of the chromosome 14 centromere using the PCR fusion method described below.

CEN-UG PCR fragment construction:

A conditional centromere for chromosome 14 was constructed by PCR amplification of DNA to the left and right of the chromosome 14 centromere followed by PCR-based fusion to overlapping fragments of the K. lactis URA3-GAL1 promoter cassette (Figure 1, B–D). Amplification and PCR fusions were performed as described (Reid et al. 2002). In the first round of PCR, a 418-bp left DNA fragment that included CEN14 was amplified using primer CHR14newUPR and adaptamer CHR14newUPF (see supplemental Table S1 for primer sequences). A left fragment of the K. lactis URA3-GAL1 cassette from pCEN-UG was amplified using primer Kl-GALintF and adaptamer Kl-GALR. A 567-bp fragment of chromosome 14 DNA adjacent to and right of the centromere was amplified using primer CHR14newDNF and adaptamer CHR14newDNR. A right fragment of the K. lactis URA3-GAL1 cassette was amplified using primer Kl-GALintR and adaptamer Kl-GALF. PCR products from the above reactions were fused in a second round of PCR using the complementary adaptamer tags to generate the illustrated products (Figure 1D). Transformation of the two PCR products into yeast and subsequent recombination generated a strain with a conditional centromere 14 (Figure 1D). All strains containing conditional centromeres were verified by PCR amplification using primers flanking the region of homology paired with primers internal to the K. lactis URA3 coding region. Thus correct insertions could be monitored by amplification of a correct-sized DNA molecule (data not shown).

Induction of chromosome loss:

Diploid strains containing a conditional centromere were made by crossing a CEN-conditional strain to a test strain in patches, incubating on YPD medium for 24 hr at 30°, then replica plating to the appropriate dropout medium to select for diploids. These diploids were then replica plated to rich medium containing galactose as a carbon source (YPgal) and grown for 24 hr at 30° to induce centromere destabilization. Finally the CEN-destabilized strains were streaked onto medium containing 5-FOA to select for cells that have lost the K. lactis URA3 gene and become 2n − 1.

Camptothecin treatment:

Camptothecin (CPT) (Sigma, St Louis) was dissolved in DMSO and added to synthetic media at a concentration of 5 μg/ml with 0.25% DMSO final concentration. Drug-free control plates also contained 0.25% DMSO. Strains tested for CPT sensitivity in the verification experiments contained an additional wild-type TOP1 allele on a CEN vector.

TOP1 expression plasmid:

The copper-inducible promoter from the CUP1 gene was obtained from plasmid pPW66R (Dohmen et al. 1994) after insertion of an MluI linker to replace the DHFR-CDC28 fusion. The pCUP1 promoter was excised on a 440-bp BamHI–MluI fragment and cloned into those same sites in plasmid YCpScTOP1 (Kauh and Bjornsti 1995). The resulting pCUP1-TOP1 fragment was then subcloned into plasmid pRS415 (Sikorski and Hieter 1989) to produce a LEU2-marked CEN plasmid expressing Top1.

Strain construction:

Sixteen haploid yeast strains with conditional centromeres were made by integrating a GAL1 promoter into each chromosome directly adjacent to consensus centromere sequences (Figure 1 and Table 1). These conditional centromeres were constructed by cloning the GAL1 promoter along with K. lactis URA3 between segments of CEN-proximal DNA followed by integration into the genome, or by a PCR-based construction and integration (as in Figure 1). Proper integration of the GAL1 promoter and URA3 into each chromosome was assessed by PCR (see materials and methods) and the correct PCR fragment sizes were obtained for each modified CEN sequence (data not shown). In addition, each haploid strain grew normally on glucose medium but showed reduced viability on galactose medium, indicating that GAL1 promoter induction adjacent to a centromere is not tolerated in a haploid cell (data not shown).

Analysis of conditional chromosome loss:

Each strain bearing a conditional chromosome was tested for its ability to lose genetic markers on that chromosome upon galactose-induced promoter activation. These experiments were performed in heterozygous diploids and chromosome loss was ascertained by LOH. In most of these experiments, the CEN-conditional strain was crossed to a test strain with one or more recessive auxotrophic markers covered by the corresponding wild-type alleles on the conditional chromosome. Chromosome loss and subsequent LOH was monitored by loss of the wild-type allele and acquisition of the auxotrophic phenotype. For example, Figure 2A shows a diagram of the relevant genetic markers on chromosome 14 after a cross of the CEN14-conditional strain (W3617-1B) to a tester strain (K393-27C). This diploid is heterozygous for the met4 and pet8 mutations on chromosome 14 and grows on SC −methionine as well as YPG (glycerol) media (Figure 2B). After induction of chromosome 14 loss (LOH₁₄) by growth on synthetic galactose medium followed by selection on 5-FOA medium, the LOH₁₄ strain fails to grow on SC −uracil, SC −methionine, and YPG, indicating the simultaneous loss of three different genetic markers: the CEN-linked URA3 marker, the wild-type MET4, and PET8 alleles. In contrast, this strain remained prototrophic for histidine and lysine showing no LOH for heterozygous markers on chromosomes 6 and 9.

Each of the 16 CEN-conditional strains was similarly tested for LOH. Table 2 lists the strains used, the markers tested for each CEN-conditional chromosome, and the LOH phenotype observed upon chromosome loss. Multiple independent diploids were produced for each experiment and 20 or more single colonies were examined for LOH. For chromosomes 3, 4, 7, and 14, multiple markers were tested for LOH events. For these and all other chromosomes, the appropriate marker loss was observed for each colony tested. At the same time, many heterozygous markers on chromosomes with unaltered centromeres were analyzed during these experiments and no LOH events were observed for those markers.

There are multiple reports of haplo-insufficiency for specific yeast genes in a diploid background (Stevens and Davis 1998; Chial et al. 1999; Deutschbauer et al. 2005) and there are also deleterious physiological effects of aneuploidy for 13 yeast chromosomes (Torres et al. 2007). We were therefore surprised to find that whole chromosome LOH events were easily obtained for all sixteen chromosomes. To further investigate these LOH events, we performed meiotic analysis for seven of the crosses listed in Table 2, plus three additional crosses using W303-derived strains. The CEN-conditional strains used for meiotic analysis are listed in Table 3 along with the tester strains that contained recessive auxotrophic marker(s). All 10 diploids were induced for chromosome loss, sporulated, and dissected. The LOH₁ and LOH₆ strains each showed 2:2 spore viability. All of the viable spores were uracil auxotrophs and the respective diploids also exhibited LOH for markers on chromosomes 1 or 6 (ade1 and his2), demonstrating that, in both cases, the chromosome containing the conditional centromere was lost. In contrast, LOH events for the other seven chromosomes tested in Table 3 showed a preponderance of four viable spores. In these cases, all spore products were uracil auxotrophs and showed 4:0 segregation for the marker listed in the LOH column (Table 3). The MAT locus showed 2:2 segregation as did any additional unlinked markers that were tested (fourth column in Table 3). Thus, it appears that for most whole chromosome LOH events, chromosome loss is followed by endoduplication of an entire chromosome.

We suspect that there is strong selective pressure for such an endoduplication event, especially for the large chromosomes, due to haplo-insufficiency (Stevens and Davis 1998; Chial et al. 1999; Deutschbauer et al. 2005). In support of this notion, we noted that during selection for LOH strains on galactose and 5-FOA media, colonies were heterogeneous in size. After additional passage, these various-sized colonies grow more uniformly, perhaps due to endoduplication. However, it is noteworthy that even after endoduplication, these diploids maintain the LOH, which permits the mapping methods presented below.

Mapping a recessive marker by chromosome LOH:

The experiments described above demonstrate that conditional chromosomes efficiently induce LOH after induction. We used conditional LOH induction to develop a method to rapidly map a recessive genetic marker to a specific chromosome. For the first test of this method, we used an unknown camptothecin-sensitive gene (cptS) in the his5Δ strain from the yeast gene disruption library (Winzeler et al. 1999) that we found during a screen for camptothecin sensitivity (R. J. D. Reid and R. Rothstein, unpublished observations). Genetic analysis showed that the cptS gene is unlinked to his5Δ. To map the unknown mutation, a cptS HIS5 segregant was isolated and crossed to each of the 16 CEN-conditional strains. Diploids were selected and induced for chromosome loss by growth on synthetic galactose medium followed by selection on 5-FOA medium. The LOH strains were then grown, serially diluted, and spotted on synthetic medium with and without camptothecin. As shown in Figure 3, the cptS LOH₂ strain shows a >10,000-fold growth reduction on camptothecin compared to the no-drug control. A control cross of a wild-type strain to all of the CEN-conditional strains using the same conditions for induction of chromosome loss resulted in no increased camptothecin sensitivity in any particular LOH strain. Although the chromosome 2 heterozygous strain initially grows slowly on galactose, it is the only one that becomes CPT sensitive when the conditional centromere-bearing chromosome is lost (Figure 3). Thus, the unknown camptothecin-sensitive phenotype in the MATα his5Δ∷KanMX deletion strain is due to a recessive gene on chromosome 2.

Next, we tested candidate genes on chromosome 2 for complementation of the cptS drug sensitivity. To choose the candidates, we took advantage of the fact that the cptS gene showed very high CPT sensitivity, but was not sensitive to gamma irradiation at 200 Gy (data not shown). A gene on chromosome 2, MMS4, has a similar phenotype when it is mutated (Xiao et al. 1998; Bastin-Shanower et al. 2003). Two other mutants that map on chromosome 2 (the weakly CPT-sensitive atg14Δ∷KanMX mutant and a drug-resistant mutant, ybr204cΔ∷KanMX) were also included in the complementation test (Figure 4). Diploids from the cross of the cptS strain to atg14 and ybr204c are resistant to camptothecin and thus showed complementation. However, diploids from the cross of cptS to mms4Δ fail to complement as they show camptothecin sensitivity comparable to both the mms4Δ and the cptS haploid strains indicating that the unknown cptS gene on chromosome 2 is allelic to MMS4.

Not every case of chromosome-scale mapping will provide potential candidate genes as in the example above. As an alternative, the unknown mutant can be crossed with all of the members of the haploid gene deletion set for a particular chromosome and then be tested for complementation. This is feasible with a simple phenotypic assay such as drug sensitivity. However, even after systematic crosses, the unknown gene may not exist in the haploid deletion library (e.g., essential genes), and this approach may fail to identify the locus. Therefore, another approach to localize the unknown gene is via traditional two-point crosses. For example, chromosome 2 is 813 kb, with a physical-to-genetic distance ratio of 0.3 kb/cM (Cherry et al. 1997). Tetrad analysis can provide a reasonable measure of genetic distance up to 50 cM. Thus approximately eight equally spaced genetic markers on chromosome 2 could be used to map the genetic location of the unknown gene. Since the deletion library is marked with KanMX, appropriately spaced deletion strains can be used directly for this mapping.

Finally, the genetic location can be ascertained systematically with multiple strains from the deletion library using the synthetic genetic array method (Jorgensen et al. 2002). Using this approach, the mutant phenotype is screened after crosses with all of the markers on the chromosome of interest. The use of “magic” markers ensures that after sporulation, only haploid progeny grow, permitting selection for the KanMX marker and subsequent screening for the mutant phenotype. In this case, the gene is localized when it fails to appear in the KanMX segregants that surround its map location.

Mapping a dominant marker by chromosome LOH:

The CEN-conditional strains can also be used to map dominant alleles after LOH induction. In this case, the LOH event must be analyzed after meiosis. For example, a dominant allele on chromosome 2 that is crossed to the CEN2-conditional strain becomes homozygous after LOH₂. Since chromosome 2, like most chromosomes, undergoes endoduplication, sporulation and dissection of only the LOH₂ strain produces 4:0 spores for the dominant phenotype. In contrast, among the remaining CEN-conditional crosses (note, LOH₃ does not sporulate), the dominant allele on chromosome 2 remains heterozygous after LOH for each chromosome and equal numbers of mutant and wild-type meiotic products are produced upon sporulation (2:2). The difference between 2:2 and 4:0 segregation can be easily determined by tetrad analysis. In LOH events where endoduplication does not occur, such as LOH₁, the same analysis is made except that tetrads of LOH₁ strains are 2:2 for viability. Therefore among the live spore products, an unlinked gene segregates 1:1, while a chromosome 1-linked gene segregates 2:0.

To facilitate the mapping of a dominant gene, a dissection-free protocol can be performed by random spore analysis. In the following example, we show how we confirmed the chromosome map position of a gene disruption from the yeast deletion library by scoring the KanMX dominant selectable marker. The yor364w∷KanMX4 strain from the MATα library, which contains a wild-type CAN1 gene, was crossed to each of the CEN-conditional chromosome strains, which contain the can1-100 mutation. The CAN1 gene is useful in this assay since CAN1/can1-100 diploids and CAN1 haploids are sensitive to canavanine and thus the only canavanine-resistant colonies that grow after sporulation are can1-100 haploids. Diploids were isolated and induced for chromosome loss. The LOH strains were sporulated and the asci were digested with glusulase to disrupt the tetrads and produce single spores (Sherman et al. 1986). Next, the spores were diluted and plated so that ≤300 colonies formed on synthetic medium containing canavanine. The haploid colonies were replica plated to G418 medium and the number of G418-resistant colonies was counted and compared to the total number of canavanine-resistant colonies (Table 4). For this analysis, chromosomes 3 and 5 cannot be assayed since LOH₃ events homozygose the MAT locus and do not sporulate, while the can1-100 allele is lost with the conditional chromosome 5. For 13 of the 14 remaining crosses, ∼50% of the colonies in the random spore analysis are G418 resistant indicating random segregation of the KanMX-marked chromosome. In contrast, for the LOH₁₅ strain, all of the canavanine-resistant haploids are also G418 resistant, indicating that the KanMX marker has become homozygous due to LOH. Thus the yor364Δ∷KanMX allele is mapped to chromosome 15 confirming its chromosome position.

There are some limitations to mapping procedures using the CEN-conditional strains described here. For example, any analyses that require sporulation, such as mapping a dominant gene, fail for chromosome 3. Therefore, we assign a gene to chromosome 3 by process of elimination. However, this limitation can be overcome by providing a plasmid-borne (or nonchromosome 3) copy of the same mating-type locus in the CEN3-conditional strains. Thus, when the native chromosome 3 is lost, the additional MAT locus provides mating-type function for sporulation. Another problem occurs during the random spore procedure for meiotic LOH analysis. The can1-100 allele is used to select haploid spore clones. Since LOH₅ results in loss of that recessive drug-resistant mutation, we are unable to select haploid strains during plating for chromosome 5. This limitation can be overcome by introducing lyp1 or cyh2 (rpl28) mutations on chromosomes 14 or 7 into the chromosome 5 strain, since these alleles cause recessive drug resistance to thiol-lysine (Grenson 1966) and cycloheximide (Kaufer et al. 1983), respectively.

Comparison to other mapping methods:

Several methods have previously been developed to map a gene to a specific chromosome. In some of these methods, chromosome loss is randomly induced in certain genetic backgrounds, for example by sporulation of a triploid strain as in Wickner (1979). Behavior of the unknown gene is then compared to chromosome-specific markers to narrow the map position to a single chromosome. Another mapping method employs a spo11-1 mutation to suppress recombination during sporulation, thus limiting the segregation of an unknown marker vs. chromosome-specific markers in predictable ways (Klapholz and Esposito 1982). In these random chromosome loss and rec⁻ methods, a mutation must first be crossed into appropriate strain backgrounds before mapping occurs. In contrast, no strain construction is required for our LOH mapping procedure. The unknown mutation need only be in a ura3 strain.

A mapping method that induces specific chromosome loss to allow analysis by LOH was also developed on the basis of 2μ-integration into all 16 yeast chromosomes in cir⁰ strains (Wakem and Sherman 1990). However in this method of chromosome loss, LOH events can routinely occur for only one chromosome arm depending on the site of 2μ-integration (R. Rothstein and G. Chrebet, unpublished observations). Thus, many more than 16 strains are required for crosses to an unknown to cover the entire genome. On the other hand, the method of chromosome loss induction described here produces LOH for all markers on a given chromosome and thus greatly simplifies the LOH analysis compared to the 2μ-integrations.

Rapid validation of library screens:

The methods described here rely on lack of complementation after an LOH event to identify the chromosome containing a given recessive marker. This principle can be used to validate a screen and verify that the gene responsible for a phenotype is likely linked to the corresponding gene disruption. As an example we used the centromere-conditional strains described in this report to verify all of our “hits” in a recently completed screen of the yeast gene disruption library. In that screen, we expressed a TOP1 allele (top1-T₇₂₂A) that acts as a CPT mimetic and damages DNA (Megonigal et al. 1997). The full results of this screen will be described elsewhere (R. J. D. Reid and R. Rothstein, unpublished results). In brief, the mutant TOP1 allele was expressed from the copper-inducible CUP1 promoter on a CEN plasmid that had been transferred into each strain in the deletion library. Heterologous expression of the mutant allele induces DNA damage and most of the deletion mutants identified as sensitive to expression of the top1-T₇₂₂A are also CPT sensitive. Approximately 140 strains showed increased sensitivity to expression of the top1-T₇₂₂A mutant compared to wild type. To assure that DNA damage sensitivity was due to the specific KanMX-marked mutations, we performed a secondary screen by crossing each of the 140 strains to the conditional centromere strain corresponding to the deletion allele's chromosome. These hybrid diploids also contain an additional wild-type copy of TOP1 on a plasmid, which effectively increases the sensitivity of yeast strains to CPT (Reid et al. 1998). All 140 strains were retested for CPT sensitivity after LOH induction and an example of the retest results from four different chromosome-8 mutants isolated in our screen are shown in Figure 5. These four mutant strains were crossed to the CEN8-conditional strain and induced for LOH₈. Any spurious recessive factors residing on another chromosome that might result in DNA damage sensitivity would be complemented in the LOH₈ strain. None of the heterozygous diploids, including a wild-type control, exhibits CPT sensitivity before LOH. We found that 3 of the 4 mutant strains are drug sensitive after LOH₈ induction, indicating that for those 3 strains the gene disruption on chromosome 8 was the likely cause of drug sensitivity. On the other hand, the dia4 mutant showed no CPT sensitivity in LOH₈ diploids and thus we could eliminate further study of this strain. Approximately 70% of the mutants isolated in the top1-T₇₂₂A mutant screen validated after rescreening by LOH.

Spurious secondary mutations unlinked to the KanMX insertions have been identified in the deletion library (Lehner et al. 2007) and we showed one such example above (cptS/mms4). In principle, such a mutation can occur on the same chromosome as a library gene disruption and would not be identified as a false positive using the method described here. A hypothetical worst case occurs when verifying chromosome-4 deletion strains. But even though chromosome 4 contains ∼13% of the total coding DNA in yeast, our LOH rescreen would remove 87% of the cases of secondary mutants assuming that these would be distributed randomly on all chromosomes. We assume that the contribution of spurious secondary mutations in these screens is relatively small, so removing ∼90% of such events is quite good.

Summary:

We have produced a set of 16 centromere-conditional chromosomes that can be used to induce LOH. The systematic LOH induction described here provides a simple and effective means of mapping both recessive and dominant alleles to specific chromosomes. In addition, hybrid diploid formation and LOH can be used to complement spurious genetic variation in strains isolated in library screens. Our simple rescreening procedure effectively eradicates such variation and confirms the results from any library screen.