Spontaneous Loss of Heterozygosity in Diploid Saccharomyces cerevisiae Cells

Corresponding author: Keiko Umezu, Department of Molecular Biology, Graduate School of Biological Sciences, Nara Institute of Science and Technology, Takayama 8916-5, Ikoma, Nara 630-0101, Japan., umezu{at}bs.aist-nara.ac.jp (E-mail)

Communicating editor: M. LICHTEN

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

To obtain a broad perspective of the events leading to spontaneous loss of heterozygosity (LOH), we have characterized the genetic alterations that functionally inactivated the URA3 marker hemizygously or heterozygously situated either on chromosome III or chromosome V in diploid Saccharomyces cerevisiae cells. Analysis of chromosome structure in a large number of LOH clones by pulsed-field gel electrophoresis and PCR showed that chromosome loss, allelic recombination, and chromosome aberration were the major classes of genetic alterations leading to LOH. The frequencies of chromosome loss and chromosome aberration were significantly affected when the marker was located in different chromosomes, suggesting that chromosome-specific elements may affect the processes that led to these alterations. Aberrant-sized chromosomes were detected readily in 8% of LOH events when the URA3 marker was placed in chromosome III. Molecular mechanisms underlying the chromosome aberrations were further investigated by studying the fate of two other genetic markers on chromosome III. Chromosome aberration caused by intrachromosomal rearrangements was predominantly due to a deletion between the MAT and HMR loci that occurred at a frequency of 3.1 x 10^-6. Another type of chromosome aberration, which occurred at a frequency slightly higher than that of the intrachromosomal deletion, appeared to be caused by interchromosomal rearrangement, including unequal crossing over between homologous chromatids and translocation with another chromosome.

GENETIC alterations leading to phenotypic changes in diploid cells are more complex than those occurring in haploid cells. Except in the case of dominant mutations, cells carrying functionally normal alleles on both homologous chromosomes usually require two genetic events, wherein both alleles are altered, to obtain a phenotypic change. For example, the first event might be a point mutation resulting in a heterozygous state with a recessive mutant allele and an unchanged normal one, while the second event could be any genetic alteration that functionally inactivates the remaining normal allele. The latter event leads to phenotypic expression of the recessive allele and is referred to as loss of heterozygosity (LOH). As proposed by the two-hit hypothesis (KNUDSON 1993 ) and as shown by extensive studies with familial cancers (LASKO et al. 1991 ), in most malignant cells, tumor-suppressor genes have been functionally inactivated by such mutational processes. Thus, determining the mechanisms leading to LOH is crucial in helping to understand the generation and progression of cancer as well as other diseases caused by somatic mutations.

The process of LOH can be caused not only by intragenic mutations but also by various genetic alterations unique to diploid somatic cells, namely, chromosome loss with or without reduplication, gross rearrangement of chromosomes, and mitotic recombination between alleles (TISCHFIELD 1997 ; LENGAUER et al. 1998 ). The LOH of tumor-suppressor genes in malignant cells is predominantly associated with such chromosomal alterations and allelic recombination. Different types of multiple chromosome rearrangements, such as gene amplifications, large deletions, and translocations, are often found in various kinds of cancer cells and some of the loci affected by the rearrangement correspond to tumor-suppressor genes. Despite their impact on carcinogenesis, however, molecular mechanisms underlying the genomic rearrangements are not well understood. This is due mainly to difficulties in identification and examination of the spontaneous rearrangements that occur in normal cells during mitotic division. Although aberrant chromosomes have been found in studies of spontaneous LOH in human T cells and mouse primary fibroblast cells, it has been difficult to determine the endpoints of the chromosome rearrangements because of the large size of the mammalian genomes (GUPTA et al. 1997 ; TISCHFIELD 1997 ). Thus, the structural constituents involved in the chromosome aberrations are not well defined. Another reason for the difficulties associated with the study of mammalian systems is that the state of local and general DNA transactions, such as replication, repair, recombination, and transcription, vary widely between cell types. LOH is the net result of such DNA transactions and the underlying mechanisms may vary significantly among different types of mammalian cells. Because of these issues, it is advantageous to study mechanisms leading to LOH in the yeast Saccharomyces cerevisiae, since its genome is small and its life cycle is simple. In addition, it may be relatively easy to examine the LOH mechanisms in yeast with the availability of elaborate genetic techniques for its study and the significant amount of information about the physical and functional structure of the whole genome of the organism. Furthermore, mechanisms involved in the DNA transactions have been well documented in yeast cells as a model organism of eukaryotic systems.

LOH in yeast has been studied by examining mitotic segregation of a given genetic marker, heterozygously situated in diploid strains or in hyperhaploid strains disomic for a certain chromosome (CAMPBELL et al. 1975 ; MALONE et al. 1980 ; ESPOSITO et al. 1982 ; HARTWELL and SMITH 1985 ; SUROSKY and TYE 1985 ; MEEKS-WAGNER and HARTWELL 1986 ; KUPIEC and PETES 1988 ; BRUSCHI et al. 1995 ). Although there were some differences between experiments with different genetic markers, these studies have shown that, in general, LOH in yeast cells is due mainly to mitotic recombination between alleles. In contrast with what was seen for mammalian cells, chromosome loss was the major chromosome aberration observed in the yeast LOH clones, and aberrant chromosomes were rarely found associated with LOH. Some of these differences between yeast and mammalian cells may be due to intrinsic differences between the organisms. However, it is also probable that various DNA elements spread over chromosomes affect the way LOH is generated differently. In other words, it is hard to rule out the possibility that the loci previously utilized for the yeast and mammalian studies may undergo LOH in a biased manner due to their surrounding sequences. In addition, yeast LOH that resulted from chromosome rearrangement might have been overlooked in previous studies, because LOH in yeast was analyzed primarily by genetic techniques rather than by direct examination of chromosome structures.

Studies of genetic rearrangements in yeast cells have been carried out, in the main, using haploid cells or cells in meiosis, since genetic techniques for detection of rearrangements are more applicable to haploid cells. It has been shown that spontaneous genetic rearrange ments in the yeast genome occur at a detectable level between repeated sequences, such as retrotransposable elements Ty, the ribosomal DNA array, multiple repeats of CUP1, and artificially inserted duplications (LIEBMAN et al. 1979 ; ROTHSTEIN 1979 ; ROEDER and FINK 1982 ; WINSTON et al. 1984 ; DOWNS et al. 1985 ; CHRISTMAN et al. 1988 ; KUPIEC and PETES 1988 ; BOEKE 1991 ; KEIL and MCWILLIAMS 1993 ; KLEIN 1995 ). Recently, Chen and Kolodner (CHEN et al. 1998 ; CHEN and KOLODNER 1999 ) showed that microhomology-mediated and non-homology-mediated gross rearrangements were significantly increased in strains defective in RFA1, RAD27, MRE11, XRS2, or RAD50 genes. These observations suggest that several different pathways of DNA recombination are involved in the spontaneous rearrangements of the genome. Some require relatively long homologous DNA sequences at the endpoints, and others do not require such homology. Furthermore, it seems likely that recombinogenic damage arises frequently in the genome of normal dividing cells and promotes genetic rearrange ments as the consequences of errors in the recombination processes. Nevertheless, only a limited portion of genetic rearrangements can be detected when haploid cells are used, as rearrangements leading to the loss of an essential gene or DNA segment go unrecognized.

Our ultimate goal is to determine the breakpoints of aberrant chromosomes and identify chromosomal elements and particular nucleotide sequences that participate in the chromosome aberration, so as to elucidate the molecular mechanisms involved in genetic rearrangement. To reveal those structural features of chromosomes involved in genetic rearrangement, especially in the context of various DNA transactions during mitotic growth, it is important to collect a variety of aberrant chromosomes occurring within a natural diploid genome without any bias toward particular types of genetic rearrangement. To achieve this, we analyzed genetic alterations that functionally inactivated a URA3 gene hemizygously or heterozygously situated at particular loci in diploid yeast cells, with the assumption that genetic rearrangement would be less detrimental in diploid cells compared to haploid cells. By directly analyzing chromosomes in LOH⁺ clones by pulsed-field gel electrophoresis (PFGE) and polymerase chain reaction (PCR), we could identify several kinds of genetic alterations participating in LOH. These alterations were chromosome loss, allelic recombination, and, in particular, gross rearrangement of chromosomes. The latter had previously never been systematically examined as an event causing LOH in yeast. The chromosome rearrangements were further investigated by coupling PFGE analysis with a conventional genetic analysis. This approach allowed us to classify the chromosome rearrangements into two classes: one caused by intrachromosomal deletion and the other by interchromosomal rearrangement. In addition, the relative frequencies of the distinct LOH events were estimated and compared between the same heterozygous marker located at either of two different chromosomes.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Media:
Media for yeast strains including complex glucose (YPD), sporulation, synthetic complete (SC), and various drop-out media were made as described (ROSE et al. 1990 ). The pH of SC and drop-out plates was adjusted to 7 before autoclaving. In YPAD medium, adenine sulfate was added to YPD to 0.004%. Uracil was added to YPD and YPAD to 20 µg/ml when indicated. 5-Fluoro-orotic acid (5-FOA) plates were prepared as described (ROSE et al. 1990 ). Escherichia coli cells were grown in Luria broth medium (MILLER 1972 ) supplemented with 100 µg/ml ampicillin or 40 µg/ml each of 5-bromo-4-chloro-3-indolyl-ß-D-galactoside and isopropylthio-ß-D-galactoside when required (SAMBROOK et al. 1989 ). SOC medium (SAMBROOK et al. 1989 ) was used for incubation of the cells after electrotransformation.

Genetic and nucleic acid techniques:
Standard methods for yeast genetics were followed (ROSE et al. 1990 ; AUSUBEL et al. 1994 ). Yeast transformations were performed using lithium acetate (GIETZ et al. 1992 ). E. coli transformation was carried out by electroporation using Gene Pulser (Bio-Rad, Hercules, CA) as recommended by the supplier. Restriction enzymes, T4 ligase, alkaline phosphatase, and Klenow fragment were purchased from Toyobo or Takara and used according to manufacturer's specifications. Yeast genomic DNA was isolated by a glass bead lysis method (AUSUBEL et al. 1994 ). Purification of DNA from agarose gel or DNA plugs was performed using QIAquick gel extraction kit (QIAGEN, Hilden, Germany). Plasmid DNA was isolated from E. coli by an alkali-lysis method (SAMBROOK et al. 1989 ) except for that used as PCR template, which was prepared with the QIAGEN Plasmid Midi kit. PCR products were purified with QIAquick PCR purification kit (QIAGEN) when used as probes for Southern hybridization. General DNA manipulations were performed as described (SAMBROOK et al. 1989 ). To quantify DNA in ethidium bromide-stained gels, the gels were photographed using a CCD video camera (Atto, Tokyo) with DensitoGRAPH software (Atto) and the intensities of DNA bands were quantified against standard DNA with Image 1.47 software (Wayne Rasband, National Institutes of Health).

S. cerevisiae and E. coli strains:
All yeast strains used were derivatives of the S288c parental strains, FY838 and FY23, kindly provided by Dr. Fred Winston (Harvard Medical School) and are listed in Table 1. Strains YKU1 and YKU21 were isolated as spore clones from the diploid constructed by the cross between FY838 and FY23. Strain YMH1 is similar to YKU1 but has the URA3 insert at about 204,900 nucleotides on chromosome III [designated here as III-205 locus: nucleotide coordinates are given in the Saccharomyces Genome Database (SGD; http://genome-www.stanford.edu/Saccharomyces)] and was constructed by transforming YKU1 with the XbaI-XmaI URA3 fragment of pMH116 (see below). Strain YMH5, in which ura3-52 was replaced with wild-type URA3, was obtained in a similar way. Strain YMJ1 was constructed in a similar way to YMH1 except FY838 was used for the transformation instead of YKU1. Strain YMJ2 was obtained by transforming YMJ1 with the EcoRI fragment of pMJ118 containing ura3-91 (see below) and selecting for 5-FOA resistant (5-FOA^r) transformants. For strain YMJ4, the NdeI-StuI fragment of pMJ118 was used to introduce the ura3-91 at the URA3 locus of YKU21. YKU23 is the same as FY838 except for the presence of ade2::hisG, which was made by the method using the hisG-URA3-hisG construct as described (ALANI et al. 1987 ). The ADE2 disruption fragment was constructed by PCR and was composed of three parts, the first 60 bp of the ADE2 open reading frame (ORF), the hisG-URA3-hisG region of plasmid pNKY51 (ALANI et al. 1987 ), followed by the last 60 bp of the ADE2 ORF. The primers used were dAFh505 and dARh2214 (Table 2). YKU34 is the same as YMH1 except for the presence of ade2::hisG and two markers on chromosome III, LEU2 and the ADE2 insert at about 314,000 (designated as III-314 locus) in addition to III-205::URA3. YKU34 was constructed by the following steps: YKU1 was converted to ade2::hisG as for YKU23. The URA3 insertion was as described for YMH1. leu21 was transplaced to LEU2 with the AccI-HpaI fragment of pRS415 (SIKORSKI and HIETER 1989 ; CHRISTIANSON et al. 1992 ) that contained LEU2. The ADE2 marker was inserted into III-314 with the EcoRI-SphI fragment of plasmid pKU7 (see below). Correct targeting of these manipulations was verified by PCR analysis of these regions at the target site and at both targeting junctions. The integration of URA3 and ADE2 in the targeted positions on chromosome III in YKU34 was also verified by meiotic mapping of the two markers with MATa. The distances between MATa and URA3, MATa and ADE2, and URA3 and ADE2 were calculated as 8.0, 39, and 32 cM, respectively, which are consistent with the values of the markers mapped adjacent to them on the same chromosome in the SGD database. E. coli strains DH5 and DH5 were used for all plasmid manipulations.

Plasmids:
Plasmid pMH116, which carried the DNA cassette for the URA3 insertion into III-205, was constructed in four steps as follows. (1) The flanking regions of the target site III-205 were amplified separately by PCR and unique restriction sites were incorporated at both ends of the amplified fragments. From 5' to 3', the primers used consisted of three portions: five random sequence nucleotides followed first by six restriction site nucleotides and then by the priming sequences to amplify the relevant yeast genome region (Table 2). Primers d3W205-X and d3C205-B, used to amplify the centromere-side region of III-205, added an XbaI site at the upstream end and a BssHII site at the downstream end of the fragment, respectively. Primers d3W205-B and d3C205-M amplified the telomere-side region of III-205 and placed a BssHII site at the upstream end and an XmaI site at the downstream end of the fragment, respectively. The products were digested with restriction enzymes. (2) The XbaI-BssHII and BssHII-XmaI fragments were ligated into the XbaI-XmaI sites of plasmid pUC19 to construct plasmid pMH100. (3) The region including the URA3 of pRS416 (SIKORSKI and HIETER 1989 ; CHRISTIANSON et al. 1992 ), position 187 to 1298 (GenBank accession no. U03450), was amplified by PCR with primers containing a BssHII site at their 5' ends. Primers dp416W-B and dp416C-B (Table 2) were used for this amplification. The products were digested with BssHII. (4) The BssHII URA3 fragment was then inserted into the BssHII site of pMH100. The resulting plasmid, which carries the URA3 marker in the direction opposite to the lacZ, was designated as pMH116. It should be noted that the XbaI-XmaI fragment of pMH116 contains 41 and 39 bp of unrelated sequences upstream and downstream of the URA3 marker between the target site, respectively. Plasmid pMJ118 was constructed by digesting pMH116 at the NcoI site inside the URA3 gene, filling the site using Klenow fragment and recircularizing the DNA. A new EcoT22I site was created at the destroyed NcoI site of the URA3 sequence in this plasmid, and the resulting ura3 mutation was designated as ura3-91. Plasmid pKU7, which carried the DNA cassette for the ADE2 insertion into III-314, was constructed as follows. (1) The ADE2 coding region including possible transcription activating sequences (GEDVILAITE and SASNAUSKAS 1994 ) was amplified by PCR, which added an ApaI site at the upstream end and a SacI site at the downstream end. Primers used were d15C566-A and d15W564-S (Table 2). The products were digested with ApaI and SacI. (2) The ApaI-SacI fragment was cloned into the ApaI-SacI sites of pRS415 to construct the plasmid pKU4. It was confirmed that pKU4 could complement an Ade^- phenotype of ade2::hisG strains. (3) The flanking regions of the target site III-314 were amplified separately by PCR, which incorporated unique restriction sites at each end of the amplified fragments. Primers used to amplify the centromere-side region were d3W313-E and d3C313-BK (Table 2), which added an EcoRI site at the upstream end and a KpnI and a BglII site at the downstream end of the fragment, respectively. The products were digested with EcoRI and BglII. Primers used to amplify the telomere-side region were d3W315-BS and d3C315-S (Table 2), which added a BglII and a SacI site at the upstream end and an SphI site at the downstream end of the fragment. The products were digested with BglII and SphI. (4) These EcoRI-BglII and BglII-SphI fragments were cloned into the EcoRI-SphI sites of plasmid pUC19 to construct plasmid pKU5. (5) The KpnI-SacI fragment of pKU4 containing ADE2 was inserted into the KpnI-SacI sites of pKU5, and the resulting plasmid was designated as pKU7.

PCR procedures:
All primers used in this study were supplied by Griner Japan and are listed in Table 2. PCR mixtures (25–100 µl) contained 10 mM Tris-HCl (pH 8.3), 50 mM KCl, 1.5 mM MgCl₂, 200 µM of each dNTP, 0.4 µM of each primer, 0.02 unit/µl of rTaq DNA polymerase (Takara) and 10⁴–10⁷ molecules of template DNA. Ex Taq DNA polymerase (Takara) was used instead of rTaq DNA polymerase to amplify the fragments for cloning. Standard PCR was performed with GeneAmp PCR system 9600 (PE Biosystems, Foster City, CA) using a program consisting of one cycle of 1 min at 95° followed by 25 to 30 cycles of 92° for 1 min, 58° for 2 min, and 72° for 2 min, unless otherwise indicated. For amplification of DNA fragments >2 kb, the extension step at 72° was increased by 1 min for each 1-kb increase of target region. For quantitative PCR analysis of chromosome III (Fig 2), about 10⁵ copies of yeast genomic DNA purified from PFGE plugs were used as the template in each reaction (25 µl). The primers d3W102, d3C102, d3W205, and d3C205 were added together in the same reaction. The reactions were performed with a 5-min extension step and the products were quantified after 18 or 20 cycles. For quantitative PCR analysis of chromosome V (Fig 4), the reaction was performed with 2 x 10⁵ cells in a 25-µl reaction. The primer pairs used were d5W116 + d5C117, d5W230 + d5C231, d5W543 + d5C544, and d10W587 + d10C588 and two pairs of the four were combined in the same reaction. The extension time was 3 min for all the reactions and the products were quantified at 22, 24, 26, 28, and 30 cycles. PCR analysis of the mating types was performed as described (HUXLEY et al. 1990 ) except that 0.02 unit/µl rTaq DNA polymerase and 0.4 µM of each primer were used. Primers used to analyze the loci on chromosome III were as follows: for LEU2 locus, d3C92 and d3W91; for III-205 locus, d3W205 and d3C205; for III-314 locus, d3W312 and d3C314; for the Hawthorne deletion, d3W197 and d3C294.

View larger version (57K): In this window In a new window Download PPT slide   Figure 1. Classification of 5-FOAr cells based on the copy number and the size abnormality of chromosome III as detected by PFGE and Southern blotting. (A–C) The pattern of chromosomes in representative clones of each class (for classification, see text) relative to that of parental strain RD101 (marked with P). The left half of each shows an ethidium bromide-stained gel of PFGE; the right half shows Southern blotting analysis of the same gel. Bands corresponding to the authentic chromosome III are marked by open arrowheads; aberrant chromosomes derived from chromosome III are marked by solid arrowheads. Bands corresponding to chromosomes V and VIII are marked by a shaded arrowhead.

View larger version (10K): In this window In a new window Download PPT slide   Figure 2. Analysis of the III-205 locus by quantitative PCR. The chromosome III pair of parent strain RD101 is illustrated with relative positions of primers used. Two horizontal lines depict the chromosome pair, circles indicate their centromeres, and an open triangle marks the URA3 insert at the III-205 locus. Head-to-head pairs of arrows indicate PCR primer sets. PCR product lengths resulting from each primer set are indicated below the corresponding positions. Three PCR products were amplified from the parental strain and are designated as fragments x, y, and z as indicated. Primer sets A and B are, respectively, d3W102 and d3C102, and d3W205 and d3C205 pairs shown in Table 2, all of which are added together in the same reaction.

View larger version (22K): In this window In a new window Download PPT slide   Figure 3. RFLP analysis of the III-205 locus in class B clones from the least heterozygous diploid strain, RD104. (A) Schematic depiction of the chromosome III pair of parent strain RD104 showing the relative positions of the primers used. Two horizontal lines depict the chromosome pair, circles indicate their centromeres, and open triangles mark the URA3/ura3-91 inserts at the III-205 locus. Head-to-head pairs of arrows indicate the primer set d3W205 and d3C205-R (Table 2), which amplifies the III-205 locus. E and N represent EcoT22I and NcoI sites in the amplified regions, respectively. (B) An expanded view of the region amplified by PCR, shown in A, for each allele. The URA3/ura3-91 inserts are shown by open bars with arrowheads indicating the direction of transcription. Thick lines represent the surrounding sequence at III-205. E and N represent EcoT22I and NcoI sites, respectively. The restriction fragments of the PCR products digested with either NcoI or EcoT22I are shown by thin lines with two arrowheads for each allele as indicated. Restriction fragment lengths are indicated below the corresponding lines. The six fragments (a–f) indicate the two alleles.

View larger version (13K): In this window In a new window Download PPT slide   Figure 4. Determination of copy number of chromosome V with quantitative PCR. Chromosomes V and X in parent strain RD201 are schematically depicted with relative positions of the primers used. Two horizontal lines depict a chromosome pair and circles indicate their centromeres. Head-to-head pairs of arrows indicate primer sets A–D. PCR product lengths with each primer set are indicated above the corresponding positions. Primer sets A–D consist of the d5W116 and d5C117, d5W230 and d5C231, d5W543 and d5C544, and d10W587 and d10C588 primers (Table 2), respectively.

Frequency of 5-FOA^r conversion events:
For all strains examined, freshly mated diploids were used to ensure normal karyotypes were present in the starting cells. For strains except RD301, the experiments were performed as follows. Cells from a single colony were inoculated and grown overnight in SC medium depleted of uracil so as to maintain the URA3 marker of the parent cells. A series of culture tubes with 5 ml YPD medium supplemented with 20 µg/ml uracil were each inoculated with 5 or 10³ cells of the preculture. Uracil was added to the medium to ensure that Ura^- segregants did not have a growth disadvantage due to a shortage of uracil. The frequency of 5-FOA^r cells in the inoculum was determined by plating about 10⁶ cells on 5-FOA plates to make sure that the background was low. Cultures were incubated at 30° to a concentration of 5 x 10⁷ cells/ml, after which the cells were harvested, washed twice, and resuspended in sterile distilled water. To disrupt the cell aggregate, either the cells were treated with 50 mM EDTA (pH 8.0) before washing or the cell suspension was briefly sonicated. The frequency of 5-FOA^r clones in each culture was determined by plating appropriate amounts of cells on two to six of both 5-FOA and YPD plates. Colonies were counted after incubation at 30° for 2 days. The 5-FOA^r colonies were classified according to size as either small (the diameter was <1 mm after 2 days) or normal-sized colonies, and the contribution of each was determined in the following experiments. 5-FOA^r cells used for further analyses were purified on 5-FOA plates.

For strain RD301, the experiments were modified as follows. As a preculture, SC medium depleted of uracil, leucine, and adenine was used to maintain the URA3, LEU2, and ADE2 markers of the parent cells. A total of 10³ cells from the preculture were inoculated into a series of culture tubes with 5 ml YPAD medium supplemented with 20 µg/ml uracil. Uracil and adenine were added to the medium to ensure that Ura^- and Ade^- segregants did not have a growth disadvantage due to shortage of uracil and adenine. Cultures were incubated at 30° until about 3–5 x 10⁶ cells/ml were present (about 14–15 generations). About 10⁶ cells were spread on three of each of three kinds of 5-FOA plates: the usual 5-FOA plates, ones lacking leucine (5-FOA Leu^-), and ones depleted of leucine and adenine (5-FOA Leu^- Ade^-). To measure the number of viable cells in the suspension, about 2 x 10² cells were spread on three YPD plates. Colonies were counted after incubation at 30° for 3 days. The 5-FOA^r cells used for further analyses were purified on the same kind of 5-FOA plates.

To determine if it was appropriate to estimate the frequency of 5-FOA^r Leu^- cells as the difference between the frequencies of 5-FOA^r cells and 5-FOA^r Leu⁺ cells and if it was appropriate to estimate the frequency of 5-FOA^r Leu⁺ Ade^- cells as the difference between the frequencies of 5-FOA^r Leu⁺ cells and 5-FOA^r Leu⁺ Ade⁺ cells, the following experiments were performed. The same number of cells was spread and grown on the three kinds of plates, 5-FOA, 5-FOA Leu^-, and 5-FOA Leu^- Ade^-. The primary 5-FOA plates were then replica plated onto 5-FOA Leu^- plates and the number of 5-FOA^r Leu⁺ cells on the replicas was compared to that on the primary 5-FOA Leu^- plates. The ratio of the former against the latter was 1.02 ± 0.092 in experiments using four independent cultures. In a similar way, the primary 5-FOA Leu^- plates were replica plated onto 5-FOA Leu^- Ade^- plates and the number of 5-FOA^r Leu⁺ Ade⁺ cells on the replicas was compared to that on the primary 5-FOA Leu^- Ade^- plates. The ratio of the former against the latter was 1.05 ± 0.22. On the basis of these results, we concluded that the frequency of 5-FOA^r Leu^- cells could be estimated as the difference between the frequencies of 5-FOA^r cells and 5-FOA^r Leu⁺ cells. Similarly, the frequency of 5-FOA^r Leu⁺ Ade^- cells was estimated as the difference between the frequencies of 5-FOA^r Leu⁺ cells and 5-FOA^r Leu⁺ Ade⁺ cells.

Pulsed-field gel electrophoresis:
Agarose plugs of chromosomal DNA were prepared as described (SCHWARTZ and CANTOR 1984 ; CARLE and OLSON 1987 ), except that the final concentration of cells in each solid plug was between 6 x 10⁸ and 1.2 x 10⁹ cells/ml. Electrophoresis was performed on 1% of PFGE certified agarose (Bio-Rad) in 0.5 x TBE buffer at 14° using a CHEF Mapper XA system (Bio-Rad). Chromosomes were separated at 6 V/cm of pulses angled at 120°. Switch time was linearly increased from 24.03 sec to 1 min 33.69 sec for 29 hr 57 min for standard analysis. Higher resolution around chromosome III and chromosome V was obtained by altering the switch time linearly from 24.03 to 44.69 sec for 34 hr 2 min and from 44.43 to 52.28 sec for 42 hr 46 min, respectively. After electrophoresis, the gel was stained with 0.5 µg/ml ethidium bromide for 30–60 min, destained in deionized water for 20–60 min, and photographed using a CCD video camera (Atto) with DensitoGRAPH software (Atto). Aberrant chromosome sizes were determined by comparison to normal-sized chromosomes used as standards. Copy number of chromosome III per cell was determined by comparison of the intensity of the corresponding band to band intensities of chromosomes I, VI, V, VIII, and XI in the same lane, using Image 1.47 software (Wayne Rasband, National Institutes of Health). The latter chromosomes serve as internal markers representing two copies of each chromosome.

Southern blotting:
Gels were soaked in 0.25 N HCl for 15 min to partially depurinate the DNA and then chromosomal DNA fragments were transferred to nylon membranes (Hybond-N+, Amersham, Buckinghamshire, England) by the capillary transfer method as described (SAMBROOK et al. 1989 ) or under vacuum with a vacuum blotter Model 785 (Bio-Rad) following instructions of the manufacturer. The probes hybridizing to the left arm of chromosome III were generated by PCR using the primer sets d3W54 + d3C54 and d3W102 + d3C102, respectively (Table 2). The probe to detect chromosome V on the right arm was prepared by PCR with d5W230 and d5C231 primers (Table 2). Hybridized probes were detected with enhanced chemiluminescent direct nucleic acid labeling and detection system (Amersham) or with Gene Images labeling and detection system (Amersham) according to the protocols of the supplier.

DNA sequencing:
The entire region of the PCR fragment of the MAT-HMR deletion was sequenced by the dye terminator method using BigDye terminator cycle sequencing kits (PE Applied Biosystems) with a capillary sequencer ABI PRISM310 (PE Applied Biosystems).

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Frequency of functional inactivation of the URA3 marker hemizygously inserted on chromosome III in diploid yeast cells:
To evaluate a wide variety of events involved in the process of LOH in eukaryotic cells, especially chromosome aberrations resulting from spontaneous genetic rearrangement, we have performed direct structural analyses of genetic alterations leading to LOH in a systematic way. Our experimental design was to construct a diploid yeast strain carrying a hemizygous marker at a particular site in the genome, isolate progeny showing functional inactivation of the marker, and analyze structural features of their genome using PFGE, quantitative and qualitative PCR, and DNA sequencing. We utilized the URA3 gene as a versatile reporter marker. When the URA3 marker has been inactivated or lost, progeny can be positively selected because of their resistance to 5-FOA (BOEKE et al. 1984 ). Chromosome III was considered to be suitable as a target chromosome for our analysis because it is the yeast chromosome most extensively characterized in its structural and functional aspects (CAMPBELL and NEWLON 1991 ; NEWLON et al. 1991 ; RIVIER and RINE 1992 ; ZENVIRTH et al. 1992 ; WU and LICHTEN 1994 ; BAUDAT and NICOLAS 1997 ) and it can be separated readily from other chromosomes by PFGE.

Initially, a 1183-bp segment of DNA containing the URA3 marker was inserted at a site 91 kb downstream from the centromere in the right arm of chromosome III (III-205::URA3) in a haploid strain whose authentic URA3 gene had been inactivated. This site was chosen so as to avoid disruption of putative open reading frames. The resulting strain, YMH1 (MATa ura3-52 III-205::URA3), was then mated with another haploid strain, FY838 (MAT ura3-52), to obtain the diploid parent strain, RD101. Then, 5-FOA^r clones were screened from RD101 cells exponentially grown in nonselective rich medium. The average frequency of 5-FOA^r convertants was calculated to be 2.0 x 10^-4 from analyses of 23 independent cultures, each containing 5–6 x 10⁷ cells/ml (Table 3). This frequency was 500-fold higher than the frequency of 5-FOA^r cells determined with a haploid strain carrying the same URA3 insert (YMH1). In similar experiments with a diploid strain homozygous for the inserted URA3 marker (RD103), no 5-FOA^r cells could be recovered from an average population of 5.5 ± 2.9 x 10⁷ cells in 9 independent cultures. Thus, the hemizygous URA3 marker is inactivated via a mechanism specific to diploid or disomic cells.

Distribution of RD101 5-FOA^r convertants into three classes of chromosome alterations:
Chromosome patterns from a large number of 5-FOA^r convertants from RD101 were examined by PFGE (Fig 1). To identify chromosome III and its derivatives, two probes specifically hybridizing with two different segments in the left arm of chromosome III were used in Southern blots performed with chromosomal DNA separated by PFGE. The copy number of chromosome III per cell was determined by comparison of the intensity of the ethidium bromide-stained band that corresponded to chromosome III or its derivatives with those of the neighboring chromosomes in the same lane, each of which represented two copies of chromosomal DNA. Various chromosomal changes including alterations in size or loss of chromosome III were observed. Since the resolution limit around the position of chromosome III in our PFGE analyses was 5 to 10 kb, only size aberrations more than that were detectable. This also explains why the two alleles of chromosome III in RD101 cannot be distinguished by their migration in the gel.

On the basis of copy number and size abnormality of chromosome III, the 5-FOA^r convertants could be categorized into three classes (Fig 1): class A, those retaining only one normal-sized chromosome III; class B, those having two normal-sized alleles of chromosome III; and class C, those harboring a chromosome III that was aberrant in size. The frequency of 5-FOA^r clones falling into each category was determined from six independent cultures (Table 3). For each culture, 32 to 48 5-FOA^r clones were examined for chromosome alterations. Cells monosomic for chromosome III (class A) were the most common and accounted for 60% of the total clones examined. Clones harboring an aberrant chromosome III (class C) were found in most but not all cultures and accounted for 8% of clones. The remaining 30% of total 5-FOA^r isolates showed no apparent alterations of either chromosome III or other chromosomes (class B).

Slower-growing cells present in the 5-FOA^r convertants:
During the first screening of RD101 as well as during the subsequent purification on 5-FOA plates, it was noted that some 5-FOA^r colonies were much smaller than others. These convertants formed colonies that had a diameter <1 mm after incubation at 30° for 2 days and accounted for 31% of all convertants. The contribution of such small colonies was <8% of the total number of colonies when the parental strain (RD101) was spread on SC plates or a ura3/ura3 diploid strain (RD102) was grown on 5-FOA plates. To determine whether this slower growth was related to certain genetic alterations in the cells, the distribution of small colonies in each of classes A–C was examined. In all three classes, 30–40% of 5-FOA^r clones formed small colonies, which indicated that the high incidence of slower-growing clones appeared not to be related to a certain class of genetic alteration but to be common to all classes. Among them, however, some class A and class C clones formed especially tiny colonies and their doubling time in YPD medium took up to 230 min, which was much longer than the 75 min required by the parent strain. Estimation of mutation rates from the frequency of mutants in a given population requires that the growth rate of the mutants is the same as that of parent cells (LEA and COULSON 1948 ; DRAKE 1991 ). If the growth rate is much reduced in a significant portion of 5-FOA^r clones then the rate of LOH in this population may be underestimated. For this reason, we evaluated 5-FOA^r conversion events on the basis of the frequency of the 5-FOA^r convertants in the present study rather than the rate calculated from the frequency.

Chromosome III carrying the URA3 marker is lost in class A 5-FOA^r clones:
Cells monosomic for chromosome III (class A) were further examined to confirm that 5-FOA resistance in these cells was due to loss of the chromosome. We determined which of the parental chromosome III pair, the chromosome inserted with the URA3 marker or the unaltered chromosome, was lost in the class A clones by determining the presence of MATa with PCR (HUXLEY et al. 1990 ). MATa/MAT is a heterozygous marker in the parent strain RD101 and is closely linked to the URA3 marker on chromosome III. MATa is located 5 kb from the URA3 marker on its centromere side in the same chromosome. In all 376 class A clones, the MAT but not the MATa locus could be amplified by PCR. Thus, the functional inactivation of the URA3 marker in the monosomic clones is due to loss of the chromosome III carrying the inserted URA3 and MATa markers.

Class B 5-FOA^r clones were homozygous wild type for the III-205 locus:
According to previous LOH studies with diploid or partial diploid yeast cells (CAMPBELL et al. 1975 ; MALONE et al. 1980 ; ESPOSITO et al. 1982 , ESPOSITO et al. 1994 ; HARTWELL and SMITH 1985 ; SUROSKY and TYE 1985 ; MEEKS-WAGNER and HARTWELL 1986 ; KUPIEC and PETES 1988 ; BRUSCHI et al. 1995 ), class B clones, the second most frequent class of RD101 5-FOA^r convertants, were expected to be generated by recombination between allelic chromosomes followed by segregation of the recombinant chromosomes during mitosis. However, because the PFGE technique used here could not reveal chromosomal size changes <5 kb, we were unable to rule out the involvement, in class B clones, of genetic alterations that yielded size alterations <5 kb such as deletions, insertions, duplications, and point mutations in the URA3 marker. To identify such genetic alterations, we performed PCR-based structural analyses of the local DNA configuration around the URA3 marker.

Attempts were made to amplify the marker sequence in the genomes of class B clones with primers that specifically annealed with sequences at both ends of the inserted URA3 marker. Although the expected size fragment could be generated when genomic DNA from the parent strain was used, the URA3 marker could not be amplified from DNA of any of 27 class B clones examined. Thus, the entire inserted segment that contained the URA3 marker was most likely lost in the class B clones.

The sequences surrounding the URA3 marker were then examined using primers positioned on both sides of the URA3 insertion site, III-205 (Fig 2, primer set B). To permit quantification of this amplification, primers that amplified a sequence on the left arm of chromosome III (primer set A) were added to the same reaction mixture to generate an internal control (fragment x). Reaction conditions were adjusted so that the reaction was terminated within the phase of exponential amplification (18–20 cycles with 3 ng of genomic DNA). When genomic DNA from the parent strain RD101 was used as the template, a 1853-bp DNA fragment (fragment z) that included the URA3 marker and a 680-bp DNA fragment (fragment y) that corresponded to the allelic locus without the marker were amplified. The molecular ratio of y/x was 0.43. When a control diploid strain that was homozygous wild type for the III-205 locus (RD102) was used, the y/x ratio was 0.80, about twice that of the parent strain. For all 213 class B clones isolated, fragment y but not fragment z could be amplified and the ratio of y/x was 0.80. These results clearly indicated that the inserted DNA segment that contained the URA3 marker was absent from the genome of class B convertants and that locus III-205 in these convertants was homozygous wild type. Since no clones in class B contained point mutations and structural alterations such as deletions and insertions <5 kb, their frequency in overall 5-FOA^r conversion was estimated to be lower than 3.1 x 10^-7. Thus, we concluded that there was little, if any, involvement of such alterations in the functional loss of the URA3 marker in RD101.

The loss of the URA3 marker in class B clones might be due to its replacement with the allelic locus through gene conversion around the site or crossing over in the CEN III-URA3 interval. It is also possible that the chromosome III that carried the URA3 marker was lost and the remaining chromosome III reduplicated. These possibilities were tested by examination of the MAT locus of the 213 class B clones by PCR (HUXLEY et al. 1990 ). Of these clones, 39 retained both the MATa and MAT loci, while the remaining 174 lost the MATa locus that was closely linked with the inserted URA3 marker. Thus, in the former clones, the URA3 marker was most likely lost through allelic recombination, either by gene conversion localized around III-205 locus or by crossing over in the interval between III-205 and MAT loci. In the latter clones, however, while recombination may also have been responsible for the loss of the URA3 marker through allelic crossing over in the CEN III-MAT interval, the possibility that chromosome loss with reduplication occurred cannot be excluded.

There are three distinct subtypes of class C 5-FOA^r clones carrying an aberrant chromosome:
As summarized in Table 4, 51 clones that carried an aberrant chromosome III (class C) were identified in 640 5-FOA^r convertants from RD101. The aberrant chromosomes varied widely in size from 110 kb smaller to 620 kb larger than the normal chromosome III. The aberrant chromosome in 80% of the clones was smaller than the normal chromosome. In the remaining clones it was much larger, the largest being almost three times as big as the normal chromosome. We have noticed that multiple clones carried an aberrant chromosome III that was indistinguishable by size. Among these, clones carrying an 80-kb shorter chromosome III occurred most frequently and accounted for 29% of all class C clones examined. Although these chromosomes might have different endpoints, it is probable that there are certain "hot spots" involved in generating class C clones in chromosome III.

To obtain an insight into the mechanisms generating the class C clones, we examined the DNA arrangement around locus III-205, where the URA3 marker was inserted, using the quantitative PCR analysis as described for class B clones (Fig 2). All class C clones, apart from six that could not be recovered after storage, were tested and could subsequently be further classified into three subtypes (Table 4).

The first subtype was characterized by gross deletion of the inserted URA3 marker and its flanking DNA. The majority of clones, 38 of 45, fell into this category. In these clones, fragment y but not fragment z could be amplified, and the molecular ratio of y/x was 0.43, similar to that of the parental strain. Thus, in these clones, locus III-205 became hemizygous upon losing the inserted marker. Despite deletions encompassing the URA3 marker, all six class C clones carrying larger aberrant chromosomes fell into this subtype. Such aberrant chromosomes were probably generated by the rejoining of a large segment of DNA at a site between the CEN III-URA3 interval.

The second subtype was characterized by replacement of the DNA segment that contained the inserted URA3 marker with wild-type locus III-205. Four of the remaining seven clones fell into this category. In these clones, fragment y but not fragment z could be amplified, and the molecular ratio of y/x was 0.80, similar to that of a homozygous control strain. These clones are thus probably homozygous wild type for locus III-205. This subtype might be related to class B clones and was probably generated by unequal crossing over between the CEN III-URA3 interval. Consistent with this notion was the finding that the decrease in size of the aberrant chromosomes of these clones was relatively small (40 or 30 kb; Table 4).

The remaining three clones fell into the third subtype, characterized by the presence of an aberrant chromosome III and the loss of the other chromosome III (an example is shown in lane 5 of Fig 1C). In these clones, the DNA segment containing the URA3 marker has been replaced with wild-type locus III-205, similar to what is seen in clones from the second subtype. With these clones that carried a single chromosome III, 30 or 10 kb smaller than the normal chromosome, fragment y but not fragment z could be amplified by PCR. The y/x ratio was 0.80. These clones are thus hemizygous for locus III-205 and are related to both class A clones and the second subtype of class C clones. It is likely that the aberrant chromosomes were generated, like the second subtype of class C clones, by unequal crossing over between the CEN III-URA3 interval.

Loss of heterozygosity of the URA3 marker in III-205::URA3/III-205::ura3-91 diploid cells:
In the RD101 diploid strain, the insertion of the URA3 marker results in substantial size heterogeneity between the two III-205 loci. This situation is somehow different from usual LOH in carcinogenesis, and it is possible that the large difference in loci in our experimental model might affect the pattern of LOH in diploid yeast cells. This was tested by experiments using RD104, a diploid strain with almost identical chromosome IIIs (Fig 3A). In this strain, one chromosome III is derived from YMH1 and carries URA3, while the other chromosome comes from YMJ2 and contains ura3-91, a sequence obtained by introducing a single-base frameshift mutation at the NcoI site in the wild-type URA3 sequence. The mutation completely abolishes URA3 gene function and simultaneously creates an EcoT22I site.

With this strain, the average frequency of 5-FOA^r convertants, obtained from analyses of 13 independent cultures each containing 5.9 x 10⁷ cells/ml, was 1.6 x 10^-4, similar to that obtained with RD101 (Table 3). The RD104 5-FOA^r convertants could also be classified into classes A–C according to their chromosome patterns, but the frequencies of clones falling into each class differed from those seen with RD101. Clones with a loss of a chromosome III (class A) occurred less frequently, while class B clones, which had two normal-sized chromosome IIIs, occurred slightly more frequently. As a result, class B became the most abundant class of 5-FOA^r convertants. Eight clones from 134 5-FOA^r convertants from RD104 contained an aberrant chromosome III (class C) and, thus, the frequency of class C was comparable with that for RD101. The number of class C clones obtained was, however, too small to determine the distribution of subtypes in these clones. Thus, the chromosomal changes leading to loss of URA3 function in RD104 occur in essentially the same manner as for RD101. However, the decrease in prevalence of class A clones in RD104 suggests that the large heterogeneity between the chromosome III pair in RD101 might enhance, by two- to threefold, the occurrence of the mechanism leading to loss of the chromosome.

ura3-91 could be easily identified through its unique restriction enzyme digestion pattern, allowing further examination of the nature of the LOH in the RD104 class B clones, which might have two ura3-91 chromosome IIIs. The replacement of URA3 with ura3-91 means that the PCR product of the URA3 sequence becomes totally resistant to NcoI digestion and would be completely digested into four fragments by EcoT22I (Fig 3B). Analysis of the PCR products with the restriction enzymes showed that all 69 class B clones isolated from seven independent cultures had lost the wild-type URA3 sequence and carried only the NcoI-resistant ura3-91 sequence.

Loss of heterozygosity of the autologous URA3 marker located on chromosome V:
To gain a more generalized view of chromosome instability as well as to measure its variability between individual chromosomes, LOH of the same marker on another chromosome was examined. Chromosome V was used for this analysis because it carries the autologous URA3 gene, which is 35 kb upstream from its centromere on the left arm.

RD201 is a diploid strain heterozygous for the URA3 locus on chromosome V and was constructed by mating a URA3 wild-type haploid strain, YMH5, with another haploid strain, YMJ4, which carries the ura3-91 mutation at the authentic locus (Fig 4). The frequency of total 5-FOA^r convertants in logarithmically growing cultures of RD201 (an average titer of 4.0 x 10⁷ cells/ml) was 0.34 x 10^-4, five times lower than that observed for RD104 (Table 3). The chromosome patterns of the convertants were examined in a way similar to those of the RD101 and RD104 convertants. Chromosome V could not be well separated from chromosome VIII in PFGE (Fig 1), but this was resolved by using Southern hybridization of DNA separated in PFGE with a probe specific for a DNA sequence on the right arm of chromosome V. This analysis, with a resolution limit of 10 kb around the position of chromosome V, showed that this chromosome in all of 152 5-FOA^r clones isolated from five independent cultures was not altered in size. Thus, the occurrence of class C clones in RD201 was at an undetectable level, estimated to be below 2.2 x 10^-7, and all clones fell into either class A or B.

Since the hybridization analyses did not allow precise measurement of chromosome V copy number, quantitative PCR was performed to distinguish between class A and B clones (Fig 4). The relative copy number was estimated for four loci in the yeast genome: the URA3 locus, two loci on the right arm of chromosome V, and a locus on chromosome X that served as a standard diploid locus. Of 152 5-FOA^r clones examined, 119 clones were classified as class B that carried a pair of chromosome Vs and the remaining 33 were classified as class A in which either one of the chromosome V pair was lost.

Using the PCR-based restriction fragment length polymorphism analysis, as performed with class B clones from RD104 (Fig 3), we determined which allele was retained at the URA3 locus in both class A and class B convertants from RD201. PCR products of the URA3 locus (Fig 4) were digested with either NcoI or EcoT22I. All of 33 class A and 119 class B clones were examined, and each carried only the NcoI-resistant ura3-91 allele. We concluded that the class A clones had lost the chromosome V carrying URA3, while the class B clones became homozygous for ura3-91.

Table 3 summarizes the distribution of the 5-FOA^r convertants from RD201 into classes A–C. When compared to RD104 and RD201, it was clear that the frequencies of two of the three major chromosomal alterations that led to LOH varied significantly when the genetic marker concerned was located in different chromosomes. That is, loss of chromosome V occurred 50-fold less frequently than loss of chromosome III, and size aberrations of chromosome V that resulted in LOH were at least 21-fold less frequent than that of chromosome III. On the other hand, frequencies of class B clones were at a similar level and the ratio of those in RD104 and RD201 was 2.4. This agrees well with the ratio of URA3-CEN interval size on chromosomes III and V, which is 2.6. This result suggested that class B clones resulted mainly from crossing over and that their prevalence was correlated with the distance between the given locus and its centromere.

Experimental approaches to understanding the origins of aberrant chromosomes generated in diploid yeast cells:
In the experiments described above, we identified clones with aberrant chromosomes (class C) by directly analyzing chromosome patterns of many 5-FOA^r convertants with PFGE and Southern blotting. The method proved to be effective for this purpose and an aberrant-sized chromosome III was identified in 6–8% of the 5-FOA^r convertants from RD101 or RD104 (Fig 1; Table 3). These clones could be subdivided into at least two distinct types based on the status of III-205 locus that was analyzed by PCR (Table 4). However, it seemed difficult to apply these procedures for statistical measurements of such fluctuating events unless we analyzed a large number of convertants from several independent populations. Moreover, structural analyses of the aberrant chromosomes identified in this way were available only with physical methods such as competitive PCR. To gain a more definite insight into the mechanisms generating aberrant chromosomes and to determine more precisely the frequencies of these distinct events leading to the chromosome aberrations, we improved our previous experimental system by introducing two other heterozygous markers besides URA3 into chromosome III.

A new parental strain, RD301, was constructed by mating a haploid strain YKU34 (MATa ura3-52 ade2::hisG LEU2 III-205::URA3 III-314::ADE2) with another haploid strain, YKU23 (MAT ura3-52 ade2::hisG leu21) (Fig 5). The LEU2, ADE2, and URA3 markers are located on the same chromosome in this diploid strain. Intrinsic LEU2 serves as a marker indicating the status of the left arm. Our definition of aberrant chromosome III is a chromosome showing a size different from the normal chromosome III but maintaining the left arm and the centromere of chromosome III. Thus, the aberrant chromosome III accompanied by the loss of the URA3 marker should maintain the LEU2 marker. Since 5-FOA^r convertants with simple loss of the chromosome III carrying the URA3 marker are expected to lose all three markers simultaneously (Fig 5A), this class of clones (class A), which accounts for more than half of the total LOH⁺ clones, can be selectively eliminated using a medium depleted of leucine. ADE2, inserted on the telomere side of the last ORF in the right arm of chromosome III, serves as a marker indicating the status of the region distal from the URA3 marker. By following the fate of these two markers in the 5-FOA^r convertants, it is possible to trace the way in which the original chromosome III is rearranged to produce aberrant chromosomes. If aberrant chromosomes result from intrachromosomal deletion of the DNA segment containing the URA3 marker, both LEU2 and ADE2 markers can be maintained in the resulting chromosome (Fig 5C). Thus, this type of aberrant chromosome would be found frequently in 5-FOA^r Leu⁺ Ade⁺ clones. On the other hand, when the URA3 marker is lost as a consequence of interchromosomal interactions, either by unequal crossing over between homologous chromosome IIIs or by translocation with another chromosome, the ADE2 marker would also be lost while the LEU2 marker would remain in the resulting aberrant chromosome (Fig 5B). In this case, such aberrant chromosomes would be detected often in 5-FOA^r Leu⁺ Ade^- convertants.

View larger version (41K): In this window In a new window Download PPT slide   Figure 5. Classification of genetic alterations leading to functional inactivation of the URA3 marker in strain RD301. Chromosomes III in parent strain RD301 (top) and their possible alteration in 5-FOAr convertants (bottom) are illustrated with relative positions of the three markers used for the analysis. The 5-FOAr convertants are classified with their indicated phenotypes (A–C) and with alteration patterns of chromosome III (a–c), which are distinguished as shown in Fig 1. The segments of chromosome III originally harboring the markers are shown by open bars and those of the homologous chromosome III are shown by shaded bars with their centromeres as circles. A solid bar in B-a indicates a chromosome segment translocated to the marked chromosome III from another chromosome. The URA3 insert at III-205 is indicated by an open triangle, the ADE2 insert at III-314 is shown by solid triangles, and the positions of intrinsic LEU2 loci are indicated by vertical lines, which are marked with a cross for the leu2 allele. A point mutation inactivating the URA3 insert is shown by a cross on the open triangle.

In a similar way, two distinct types of allelic recombination, both of which result in two normal-sized chromosome IIIs that are homozygous for III-205 in 5-FOA^r convertants (class B), can also be distinguished using the new parental strain. If the chromosome has lost the URA3 marker by local gene conversion not associated with crossing over, such convertants would maintain the LEU2 and ADE2 markers and show a phenotype of 5-FOA^r Leu⁺ Ade⁺ (Fig 5C). Meanwhile, crossing over between homologous chromosome IIIs in the interval from III-205 to the centromere will give rise to a hybrid chromosome III in which the telomere-side region of the right arm containing both URA3 and ADE2 markers is replaced with the corresponding wild-type sequence of another chromosome III. LOH⁺ clones with this type of allelic recombination would show a phenotype of 5-FOA^r Leu⁺ Ade^- (Fig 5B). Gene conversion that replaced the segment that contained the URA3 with the wild-type sequence and was accompanied by crossing over in the interval from URA3 to ADE2 would also be included in this kind of convertant.

We expected allelic recombination to occur more frequently than chromosome aberrations in our system. Hence, we first determined the phenotypes of the 5-FOA^r convertants as either Leu⁺ Ade⁺ or Leu⁺ Ade^- and then determined the size of their chromosomes using PFGE to distinguish clones with the chromosome aberration from those with the allelic recombination.

Frequencies of 5-FOA^r Leu⁺ Ade⁺ and 5-FOA^r Leu⁺ Ade^- convertants in the diploid cells carrying three heterozygous markers on chromosome III:
With the new parental diploid strain RD301, we first determined the frequency of 5-FOA^r convertants in the cells growing exponentially in nonselective rich medium. From experiments with 15 independent cultures, final titers of which were set around 4 x 10⁶ cells/ml, the frequency of 5-FOA^r convertants was calculated to be 1.7 x 10^-4 in arithmetical mean and 1.2 x 10^-4 in median (Table 5). As expected, these numbers were comparable to those obtained with the strain RD101 that carried the same III-205::URA3 insert (Table 3). Cells from the same cultures were spread on 5-FOA plates lacking leucine and ones depleted with leucine and adenine to measure the frequencies of 5-FOA^r Leu⁺ and 5-FOA^r Leu⁺ Ade⁺ clones, respectively (Table 5). In these experiments, we could examine the phenotypes of a large number of clones in multiple cultures, which enabled the calculation of the median frequencies. Because the frequency estimated from the arithmetical mean is overly influenced by the jackpot effect, we used the median to express the frequency of clones concerned for analyses with RD301.

5-FOA^r convertants accompanied by the Leu^- or Ade^- phenotype cannot be directly screened with the selective plates. Instead, we calculated the frequency of 5-FOA^r Leu^- clones by subtracting the frequency of 5-FOA^r Leu⁺ clones from that of 5-FOA^r cells (Table 5). The calculated frequency of 5-FOA^r Leu^-, 6.8 x 10^-5, is comparable to the frequency of clones monosomic for chromosome III (class A) determined with the strain RD101 (Table 3). Similarly, the frequency of 5-FOA^r Leu⁺ Ade^- clones was calculated by subtracting the frequency of 5-FOA^r Leu⁺ Ade⁺ clones from that of 5-FOA^r Leu⁺ cells. The calculated frequency of 5-FOA^r Leu⁺ Ade^- clones is 4.9 x 10^-5. Since the 5-FOA^r Leu⁺ Ade^- clones were expected to result from interchromosomal rearrangements, either by crossing over between homologous chromosome IIIs or by translocation with another chromosome (Fig 5B), such interchromosomal rearrangements appeared to be involved in about half of the LOH events that led to the functional inactivation of the URA3 marker.

Aberrant chromosomes resulting from intrachromosomal recombination:
As described above, 5-FOA^r Leu⁺ Ade⁺ convertants result from genetic alterations inactivating the URA3 marker without affecting the other two markers. Expected events leading to this genetic alteration are intrachromosomal rearrangements of the DNA segment including the URA3 marker, gene conversion localized around the marker, and point mutations in the marker (Fig 5C). In fact, PFGE analyses of these convertants, performed as for those of RD101 (Fig 1), revealed two types of clones showing different chromosome patterns: (1) clones having two normal-sized chromosome IIIs (class B) and (2) those carrying an aberrant-sized chromosome III along with a normal-sized one (class C). Table 6A shows the distribution of 98 5-FOA^r Leu⁺ Ade⁺ convertants into these two classes. All class B clones were further examined for their status of locus III-205 using quantitative PCR similar to the analysis performed for RD101 (Fig 2). Nine of the 11 clones appeared to be homozygous for the wild-type allele (non-URA3 insertion) of locus III-205. Thus, we concluded that 5-FOA^r Leu⁺ Ade⁺ class B clones result mainly from gene conversion around locus III-205. The remaining two clones that were isolated from the same culture were found to carry an identical point mutation (G:C to T:A base substitution) within the URA3 marker and thus probably are siblings. This result indicates that the frequency of point mutations in the URA3 marker would be at least five times lower than the frequency of gene conversion. In 87 class C clones (Table 6A), on the other hand, the aberrant chromosomes probably resulted from intrachromosomal deletions of DNA segments that included the URA3 marker.

Interestingly, all of the aberrant chromosomes identified in the class C clones migrated to a similar position in PFGE that corresponded to a size 80 kb smaller than the normal chromosome III. On the basis of this result, we expected that the aberrant chromosome might arise due to the Hawthorne deletion, a recessive lethal deletion between MAT and HMR loci located 90 kb apart in the right arm of chromosome III (HAWTHORNE 1963 ). Since the DNA segment that would be eliminated by Hawthorne deletion includes locus III-205 but not locus III-314, such a deletion occurring in the chromosome III of RD301 would produce a chromosome missing the URA3 marker only. To examine this possibility, PCR encompassing a segment between MAT and HMR was carried out with DNA from 5-FOA^r Leu⁺ Ade⁺ class C clones. PCR with the 87 clones successfully amplified the 3.0-kb fragment indicative of Hawthorne deletion in all cases. Furthermore, DNA sequence analysis of the PCR products revealed that the 1.6 kb of homologous sequence of MATa and HMR was faithfully fused within the amplified DNA, as in the case of Hawthorne deletion. From these results, we concluded that all of the aberrant chromosomes identified in the 5-FOA^r Leu⁺ Ade⁺ convertants resulted from intrachromosomal recombination between MATa and HMR with a frequency of 3.1 x 10^-6 (Table 6A). This is the first report on the frequency of Hawthorne deletion in a heterothallic diploid strain, while the frequency in homothallic strains that maintain active HO endonuclease has been previously reported as 1% (HABER et al. 1980 ).

The aberrant chromosome caused by Hawthorne deletion was similar in size to the most frequent type of aberrant chromosome isolated from the strain RD101 (Table 4). Although it is not clear whether these chromosomes resulted from intrachromosomal rearrangement, the same PCR used to detect Hawthorne deletion successfully amplified the 3.0-kb fragment with 11 of the 13 clones. The frequency of such clones in RD101 was about 3 x 10^-6, which is in good agreement with the frequency of Hawthorne deletion observed with RD301.

Aberrant chromosomes resulting from interchromosomal rearrangement:
5-FOA^r Leu⁺ Ade^- convertants formed red colonies on 5-FOA Leu^- plates and could be easily distinguished from 5-FOA^r Leu⁺ Ade⁺ clones, which formed white colonies on the same plates (ROMAN 1956 ). We isolated 140 5-FOA^r Leu⁺ Ade^- clones from seven independent cultures of RD301 by picking 20 clones from each culture. Chromosomes from all the clones were examined by PFGE analysis. We identified three types of clones with different chromosome patterns (Table 6B): (1) clones having two normal-sized chromosome IIIs (class B), (2) those carrying an aberrant- sized chromosome III along with a normal one (class C), and, unexpectedly, (3) those monosomic for chromosome III (class A), which is described in detail below. The class B clones were considered to arise from crossing over between homologous chromatids of chromosome III in the interval from locus III-205 to the centromere (Fig 5B). Their estimated frequency, 4.0 x 10^-5, was comparable to the frequency of class B convertants from the strain RD101 (Table 3).

Aberrant chromosomes identified in 5-FOA^r Leu⁺ Ade^- clones were expected to result from either unequal crossing over between homologous chromosome IIIs or translocation with another chromosome (Fig 5B). In fact, sizes of aberrant chromosomes identified in 18 class C clones varied widely. The distribution was similar to that observed with RD101 (Table 4), except that an 80 kb shorter chromosome III, the one most frequently recovered from RD101, was not obtained in 5-FOA^r Leu⁺ Ade^- clones. These class C clones were further examined for their status of locus III-205 by quantitative PCR as used for RD101 (Fig 2). Of 18 clones, 1 clone was homozygous for the wild-type allele (non-URA3 insertion) of locus III-205, which suggested the involvement of unequal crossing over between the homologous chromosome IIIs. Consistent with this idea was the finding that the aberrant chromosome had a small change in size and was 10 kb smaller than the normal chromosome III. The remaining 17 clones were found to be hemizygous for the wild-type allele of