The tRNA-Tyr Gene Family of Saccharomyces cerevisiae: Agents of Phenotypic Variation and Position Effects on Mutation Frequency

Corresponding author: John H. McCusker, 3020 Duke University Medical Center, Durham, NC 27710., mccus001{at}mc.duke.edu (E-mail)

Communicating editor: L. PILLUS

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Extensive phenotypic diversity or variation exists in clonal populations of microorganisms and is thought to play a role in adaptation to novel environments. This phenotypic variation or instability, which occurs by multiple mechanisms, may be a form of cellular differentiation and a stochastic means for modulating gene expression. This work dissects a case of phenotypic variation in a clinically derived Saccharomyces cerevisiae strain involving a cox15 ochre mutation, which acts as a reporter. The ochre mutation reverts to sense at a low frequency while tRNA-Tyr ochre suppressors (SUP-o) arise at a very high frequency to produce this phenotypic variation. The SUP-o mutations are highly pleiotropic. In addition, although all SUP-o mutations within the eight-member tRNA-Tyr gene family suppress the ochre mutation reporter, there are considerable phenotypic differences among the different SUP-o mutants. Finally, and of particular interest, there is a strong position effect on mutation frequency within the eight-member tRNA-Tyr gene family, with one locus, SUP6, mutating at a much higher than average frequency and two other loci, SUP2 and SUP8, mutating at much lower than average frequencies. Mechanisms for the position effect on mutation frequency are evaluated.

MEMBERS of clonal populations of bacteria and fungi exhibit frequent and significant instability or variation in their phenotypes (reviewed in ROBERTSON and MEYER 1992 ; SOLL 1992 , SOLL 1997 ; RAINEY et al. 1993 ; and MOXON et al. 1994 ). This phenotypic variation (also denoted adaptive evolution, intraclonal polymorphism, antigenic variation, and phase variation) plays a role in microbial adaptation to novel environments by generating genotypic and phenotypic diversity. For example, phenotypic variation plays an important role in bacterial pathogenesis by aiding in resistance to host defenses as well as in dissemination to and colonization of different ecological niches. The genotypic and phenotypic diversity generated by phenotypic variation can be viewed as a form of cellular differentiation in microorganisms and as a stochastic method for regulating gene expression.

There are a variety of phenotypic variation mechanisms. Some mechanisms for phenotypic variation, such as silencing (PILLUS and RINE 1989 ; GOTTSCHLING et al. 1990 ) and [PSI⁺] (TRUE and LINDQUIST 2000 ), are epigenetic. However, other common mechanisms for phenotypic variation involve changes at the DNA sequence level, such as the expansion-contraction of highly mutable simple repetitive sequences (reviewed in ROBERTSON and MEYER 1992 ; RAINEY et al. 1993 ; MOXON et al. 1994 ). Where there are changes at the DNA sequence level, specific sequences in phenotypically variable genes can be thought of as being at the high-frequency end of the mutational continuum; that is, specific sequences in phenotypically variable genes are mutational hotspots.

As shown by extensive progress with silencing and [PSI⁺], Saccharomyces cerevisiae is a particularly useful eukaryotic model for phenotypic variation. In addition to more recent studies with [PSI⁺] and silencing, phenotypic variation has also been seen classically in S. cerevisiae (WINGE 1944 ; LINDEGREN 1956 ; SKOVSTED 1956 ; SCHEDA and YARROW 1966 , SCHEDA and YARROW 1968 ) but the descriptions were largely phenomenological, involving, for the most part, subjective assays, such as distinguishing among multiple colony morphologies in one strain. To avoid the phenomenology inherent in subjective assays, objective phenotypes, such as the conditional growth assays used to study silencing and [PSI⁺], are extremely advantageous.

Phenotypic variation has been noted in clinically derived S. cerevisiae strains (MCCUSKER and DAVIS 1991 ) and, given the relevance of bacterial phenotypic variation to pathogenesis, was an intriguing area for further investigation. This work focuses on one case of phenotypic variation in a clinically derived S. cerevisiae genetic background, which, in addition to having an objective conditional growth assay, displays intriguing effects on respiration and sporulation and presents an interesting genetic puzzle.

We show that this case of phenotypic variation is due to a cox15 ochre mutation, which acts as a reporter and reverts to sense at a low frequency; it is suppressed by tyrosine-inserting ochre suppressors, which arise at a very high frequency. We also describe a novel phenotype associated with ochre suppressors, namely substantially reduced PGK1 promoter activity; there is significant phenotypic variability in the effects of the tRNA-Tyr gene family on PGK1 promoter activity. Therefore, this work provides an additional example of the pleiotropic impact of translational misreading on phenotype and of the relevance of translational misreading to phenotypic variation. Finally and most interestingly, within the eight-member tRNA-Tyr gene family there is a strong position effect on mutation frequency.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Strains:
All S. cerevisiae strains used in this study are shown in Table 1. Strains with names beginning with an "S" were isogenic with S288c, while strains with names beginning with a "V1-" were isogenic with the clinically derived strain YJM421 (MCCUSKER et al. 1994 ).

Media:
YEPD (SHERMAN et al. 1974 ) and YEPD containing G418 (Geneticin, GIBCO BRL, Gaithersburg, MD), hygromycin B (Calbiochem-Novabiochem, La Jolla, CA), and nourseothricin (clonNAT, Hans-Knoll Institute fur Naturstoff-Forschung, Jena, Germany) have been described previously (WACH et al. 1994 ; GOLDSTEIN and MCCUSKER 1999 ). Osmotic sensitivity was tested on YEPD containing 1.5 or 1.75 M KCl. YEPEG contained 1% succinic acid, 1% yeast extract, 2% bacto-peptone, 2% glycerol (mixed, adjusted to pH 5.5 and autoclaved), and 2% ethanol (added after autoclaving). Strains carrying cox15-421 were grown in medium containing antimycin A (Sigma, St. Louis; added after autoclaving to a final concentration of 1 µg/ml) when propagated long enough to potentially exhaust dextrose and select for ochre suppressors. Synthetic dropout media have been described previously (ROSE et al. 1990 ). All types of solid media contained, 2% bacto-agar (Difco, Detroit).

Fcy1⁺ Fcy1MX4-containing transformants were selected on synthetic minimal medium containing 1.7 g/liter yeast nitrogen base without amino acids or (NH₄)₂SO₄ (Difco), 20 g/liter dextrose (Sigma), 20 g/liter agar, and 1 mM cytosine (Sigma) as the sole nitrogen source (ERBS et al. 1997 ); medium used to select for loss of the Fcy1MX4 cassette was similar but lacked cytosine and contained 5 g/liter (NH₄)₂SO₄ and 10^-4 M 5-fluorocytosine (Sigma). Cytosine and 5-fluorocytosine were added to the medium after autoclaving.

Preparation of genomic and plasmid DNA:
Yeast genomic DNA was prepared as described previously (AUSUBEL et al. 2002 ). Plasmid DNA was isolated using QIAGEN (Valencia, CA) mini- and midi-kits. Operon (Alameda, CA) or IDT (Cornwallis, IL) synthesized oligonucleotides. Sequence analysis was done by the Duke University Sequencing Core using an ABI prism sequencer.

PCR-mediated gene deletion construction:
Using primers shown in Table 2, MX3 and MX4 cassettes were amplified by PCR for construction of insertion-deletion mutations as described previously (WACH et al. 1994 ; GOLDSTEIN and MCCUSKER 1999 ; GOLDSTEIN et al. 1999 ). Colony PCR was used to confirm insertion-deletion mutations as described previously (NIEDENTHAL et al. 1996 ; GOLDSTEIN and MCCUSKER 2001 ).

Transformation of Escherichia coli and S. cerevisiae:
DNA was introduced into DH10ß cells (GIBCO BRL) via electroporation (DOWER et al. 1988 ) with a Gene Pulser II (Bio-Rad). URA3-containing plasmids (0.1 µg) and PCR products [1.0 and 0.1 µg for PCR products with short (30–40 bp) and long (300–500 bp) regions of homology, respectively] were introduced into S. cerevisiae strains as described previously (GIETZ et al. 1995 ). For PCR products with short regions of target homology, DMSO (10 µl) was added before heat-shocking the cells.

Cloning of COX15:
A YCp50-based S. cerevisiae library (ROSE et al. 1987 ), obtained from the American Type Culture Collection, was amplified and 0.5 µg of isolated plasmid DNA was transformed per reaction into the Pet^- strain V1-13-. Each of 14 transformation reactions was plated onto three synthetic uracil dropout (SDC-URA) + antimycin A plates, which were incubated for 2 days at 30°; these plates were then replica plated to YEPEG plates, which were incubated for 3 days at 30°.

Introduction of the cox15-421 mutation into the S288c background:
The cox15-421 ochre mutation was introduced into the S288c genetic background in a three-step process. First, the Fcy1MX4 cassette [the open reading frame (ORF) of the kanMX4 cassette (WACH et al. 1994 ) was replaced with the S. cerevisiae FCY1 ORF] in plasmid pPH37 was amplified with primers PH127 and PH128 (with 5' sequences up and downstream from the COX15 open reading frame). The resulting PCR product was transformed into the fcy1 strain S175, selecting for the ability to utilize cytosine as the sole nitrogen source. The resulting cox15::Fcy1MX4 mutation in S182 was confirmed by PCR (using primers PH164 and PH137 to assay for cox15::Fcy1MX4), Fcy1⁺ phenotype, and a ⁺ Pet^- phenotype.

Second, a PCR product containing the cox15-421 ochre mutation was amplified from YJM421 genomic DNA using primers PH133 and PH134. Finally, the cox15-421-containing PCR product was cotransformed with the URA3-containing plasmid pRS316 (SIKORSKI and HIETER 1989 ) into strain S182, selecting for Ura⁺. The resulting Ura⁺ colonies were replica plated to 5-fluorocytosine-containing medium to screen for replacement of cox15::Fcy1MX4 by cox15-421. Replacement of cox15:: Fcy1MX4 by cox15-421 was confirmed by PCR using primers PH136 and PH139 (to assay for the COX15 ORF) and primers PH164 and PH137. The ura3 fcy1 cox15-421 strain S183 was used for further analysis.

Library construction for cloning SUP7-o from YJM421:
Genomic DNA from YJM421, which was completely digested with HindIII or XbaI, was ligated into 0.5 µg of appropriately digested pRS316 (SIKORSKI and HIETER 1989 ). For each digest, three ligation reactions were incubated for 16 hr at 15° and then diluted to 100 µl with H₂O. Each ligation reaction was split into 50-µl aliquots for direct transformation into strain V1-13-. Each of 14 yeast transformation mixes was plated onto three SDC-URA + antimycin A plates. An estimated 140,000 and 60,000 Ura⁺ colonies were recovered from the HindIII- and XbaI-digested DNA ligations, respectively. After being grown for 2 days on SDC-URA + antimycin A, Ura⁺ transformants were replica plated to YEPEG to screen for Pet⁺.

Determination of mutation frequencies and rates:
To estimate mutation frequencies and rates, six independent colonies of each strain were grown overnight at 30° in 5 ml of YEPD containing 1 µg/ml of antimycin A. After overnight growth, cells were pelleted, washed twice with water, resuspended in water, and counted. Aliquots of 100 cells were inoculated into 50 ml of YEPD medium containing 5% dextrose and grown at 30°. The cell densities of all cultures were determined by cell counts to estimate the number of cell divisions each culture had undergone since inoculation for subsequent mutation rate determinations. Cells were then harvested in log phase and washed with water.

To measure the Pet⁺ mutation frequency in each cox15-421 culture, 2 x 10⁸ cells were plated onto each of two YEPEG plates. To measure forward mutation frequencies at the CAN1 and LYP1 loci in each culture, 3 x 10⁷ cells were plated onto synthetic arginine dropout medium containing 40 µg/ml canavanine (Sigma) and synthetic lysine dropout medium containing 100 µg/ml of S-2-aminoethyl-L-cysteine (Sigma), respectively. Simultaneously, appropriately diluted cells of each culture were plated onto YEPD to estimate the number of viable cells. The plates were incubated at 30° for 2–3 days after which colonies were counted. The mutation frequencies were calculated for each culture by dividing the average number of colonies on selective media by the number of cells plated; these frequencies were then averaged. Mutation rates were calculated using the formula , where µ is the mutation rate, f₁ is the initial mutation frequency in culture, f₂ is the final mutation frequency in culture, N₁ is the initial number of cells in culture, and N₂ is the final number of cells in culture (DRAKE 1991 ); these rates were then averaged.

Isolation of independent Pet⁺ mutants:
For each of the cox15-421 strains, one Pet^- colony was streaked (separately) for single colonies on YEPD. After incubation at 30° for four days, the YEPD plates, which contained multiple single Pet^- colonies, were replica plated to YEPEG. After incubation at 30° for 2–3 days, Pet⁺ papilli appeared from many (but not all) of the replica-plated Pet^- colonies. To ensure that Pet⁺ mutants were independent, a single Pet⁺ papilla was picked from each Pet^- colony and purified by streaking on YEPD for single colonies for subsequent genotyping purposes.

Genotyping of tRNA-Tyr:
The genotypes of tRNA-Tyr loci in Pet⁺ cox15-421 strains were first assayed by determining the presence (in sup⁺) or absence (in SUP-o) of the HpyCH4III site in the anticodon region of the tRNA-Tyr genes. Using the sets of primers shown in Table 2 and genomic DNA as templates, PCR products containing each of the eight tRNA-Tyr loci were obtained from each of the Pet⁺ strains. PCR products were ethanol precipitated, digested with the restriction enzyme HpyCH4III, and electrophoresed in 4% agarose gels. In Pet⁺ derivatives of haploid cox15-421 strains, the PCR products of a given tRNA-Tyr locus were either HpyCH4III sensitive or HpyCH4III resistant. In diploid cox15-421 strains, the PCR products of sup⁺/sup⁺ tRNA-Tyr loci were HpyCH4III sensitive while SUP-o/+ loci yielded a mixture of HpyCH4III-sensitive and -resistant PCR products.

For all putative SUP-o mutants (tRNA-Tyr loci with HpyCH4III-resistant PCR products), mutant-allele-specific amplification (MASA; TAKEDA et al. 1993 ; HASEGAWA et al. 1995 ) PCR was performed to confirm the SUP-o genotypes. Genomic DNAs were used as a template in conjunction with primers with different 3' sequences (Table 2, SA86–SA90) in combination with locus-specific primers (Table 2, SA2, SA4, SA5, SA8, SA9, SA12, SA14, and SA16). MASA-PCR reactions were initiated at 95° for 5 min followed by 35 amplification cycles (95° for 30 sec, 52° for 30 sec, and 72° for 2 min) and terminated with a 10-min extension at 72°. The reaction mixtures were then electrophoresed in 4% agarose gels. To be deemed a SUP-o mutant, a HpyCH4III-resistant tRNA-Tyr locus in Pet⁺ derivatives of the haploid cox15-421 strains V1-1- and S183 had to meet two criteria. First, in conjunction with locus-specific primers, a HpyCH4III-resistant tRNA-Tyr locus had to fail to produce MASA-PCR products with primers SA86 (3' end homologous to the wild-type tRNA-Tyr anticodon), SA88 (3' end homologous to tRNA-Tyr with an amber suppressor anticodon), and SA89 (3' end homologous to tRNA-Tyr with an opal suppressor anticodon). Second, in conjunction with locus-specific primers, a HpyCH4III-resistant tRNA-Tyr locus had to produce MASA-PCR products with both primer SA90 (which, at its 3' end, encompasses the two bases of the HpyCH4III site outside of the anticodon) and primer SA87 (3' end homologous to tRNA-Tyr with an ochre suppressor anticodon).

YJM421 (SUP7-o/+) and Pet⁺ derivatives of the cox15-421 diploid strain YSA3 contained tRNA-Tyr loci with both HpyCH4III-sensitive and -resistant PCR products. Therefore, the criteria for identifying an ochre suppressor in these SUP-o/+ strains differed slightly from the criteria used for haploid strains in that a MASA-PCR product was expected with both primer SA86 and primer SA87.

Introduction of SUP2-o and SUP8-o mutations into the S288c background:
Spontaneous SUP2-o and SUP8-o mutants were not isolated in the S288c genetic background. To introduce SUP2-o and SUP8-o into the S288c genetic background, SUP2-o and SUP8-o were first amplified from SUP-o-containing, V1-1-derived strains using primers SA1 and SA2 (SUP2-o) and primers SA13 and SA14 (SUP8-o). The resulting SUP2-o- and SUP8-o-containing PCR products were then transformed, separately, into the ura3 cox15-421 strain S183, together with the kanMX4-containing CEN plasmid pSA15. G418^r transformants were selected; screening for Pet⁺ identified putative SUP-o-containing cotransformants. Replacement of sup⁺ with SUP-o at the SUP2 or SUP8 locus in G418^r Pet⁺ transformants was examined by the destruction of the HpyCH4III recognition sequence and then confirmed by MASA-PCR, as described above. Finally, derivatives of the SUP2-o- and SUP8-o-containing strains that had lost pSA15 were selected by demanding growth on 5-fluoroorotic acid-containing medium. The plasmid-less strains S1260 and S1269 (SUP2-o) and S1258 and S1259 (SUP8-o) were used for further analysis.

Determination of suppressor efficiency:
Ochre suppressor efficiency was determined by quantification of the suppression of the ochre mutation in the plasmid-borne ß-galactosidase gene of pUKC817 (URA3 lacZ-ochre) relative to the control plasmid pUKC815 (URA3 lacZ⁺), as described previously (FIROOZAN et al. 1991 ; STANSFIELD et al. 1995 ). For each SUP-o locus, except for SUP3-o, two or three independently isolated SUP-o-containing strains were assayed for ß-galactosidase. For each experimental measurement, three independent transformants of each SUP-o-containing Pet⁺ strain were assayed for ß-galactosidase. Assays from each culture were performed in duplicate.

Oligonucleotide-mediated transformation of the cox15-421 ochre mutation to sense:
Oligonucleotide-mediated transformation was performed using a modification of a previously described (MOERSCHELL et al. 1991 ) procedure. Seven single base-pair substitution mutations can convert an ochre codon to one of seven sense codons, representing six amino acid substitutions. To determine which of these six amino acid substitutions might result in a functional Cox15p, strain S183 was cotransformed with (i) plasmid pSA11 (CEN natMX4) and (ii) one of seven oligonucleotides (SA98–SA104; using 10 µl of 500 µM solution of each oligonucleotide), which could convert the ochre codon of cox15-421 to one of the seven sense codons. After 2 days growth at 30°, nourseothricin-resistant transformants were screened for Pet⁺ by replica plating to YEPEG. The nourseothricin-resistant Pet⁺ transformants were tested for osmotic sensitivity to exclude all of the tyrosine-inserting SUP-o except SUP11-o. The SUP11 genotype of Pet⁺ Osm⁺ transformants was determined as described above. For sup11⁺ Pet⁺ Osm⁺ transformants, the region around codon 39 of COX15 was amplified (using primers PH132 and PH158), cloned into pCR2.1-TOPO (Invitrogen, San Diego), and sequenced. For each sup11⁺ Pet⁺ Osm⁺ transformant, two independent plasmid clones were sequenced on both strands using M13 and reverse primers. To exclude the presence of [PSI⁺], four sup11⁺ Pet⁺ Osm⁺ cox15-421 isolates were grown under [psi^-]-inducing conditions (LUND and COX 1981 ; TUITE et al. 1981 ); that is, they were streaked for single colonies on YEPD + 5 mM guanidine hydrochloride followed by replica plating twice to YEPD + 5 mM guanidine hydrochloride. After growth under these [psi^-]-inducing conditions, the Pet⁺ phenotypes of the strains were retested.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The unusual genetic behavior of YJM421:
In spite of being HO and self-diploidized, tetrads of the clinically derived Spo⁺ strain YJM421 (MCCUSKER et al. 1994 ) dissected onto YEPD showed Mendelian (2:2) segregation for colony size, suggesting heterozygosity at one locus. Small colony size is frequently indicative of a respiration defect. Therefore, segregants were tested for their ability to respire by replica plating to YEPEG; 2 Pet⁺:2 Pet^- segregation was observed with, unexpectedly, all of the large colonies being Pet^- and all of the small colonies being Pet⁺ (Fig 1A and Fig B). There was no difference in cell size between the isogenic Pet⁺ and Pet^- segregants, which suggested that the difference in colony size was due to a difference in growth rate. As expected for an HO strain, all segregants were nonmating but, unexpectedly, all segregants were sporulation deficient (Spo^-).

View larger version (130K): In this window In a new window Download PPT slide   Figure 1. Tetrads of YJM421 showing (A) colony size on YEPD, (B) growth on YEPEG, and (C) growth on YEPD + 1.5 M KCl.

The Pet^- phenotype of YJM421 segregants was unstable. All Pet^- segregants produced Pet⁺ papilli after prolonged incubation on YEPD and all Pet^- segregants produced abundant Pet⁺ papilli when replica plated from YEPD to YEPEG medium (Fig 2). The Pet^- Pet⁺ variants were Spo⁺ and the original segregation pattern of YJM421 was recapitulated in these Pet^- Pet⁺ variants; that is, after sporulation and dissection, all Pet^- Pet⁺ variants produced tetrads with 2 Spo^- Pet^- large colony:2 Spo^- Pet⁺ small colony segregation.

View larger version (158K): In this window In a new window Download PPT slide   Figure 2. Tetrad of YJM421 on YEPEG—frequent Pet+ papilli from Pet- segregants.

The results were consistent with a formal model where the unusual behavior and phenotypic variability in the YJM421 genetic background was due to a phenotypic variability locus (PHV1), with two alleles (PHV1-1 and PHV1-2) that had pleiotropic effects on respiration and sporulation. Formally, the PHV1-1 allele would be necessary for respiration and dominant to PHV1-2 with respect to the Pet⁺ phenotype. However, since heterozygosity was required for sporulation, PHV1-1 and PHV1-2 would be either codominant or dosage dependent.

Genetic behavior of the Pet^- phenotype in crosses with S288c background strains:
To aid in the design of a cloning strategy for the hypothetical PHV1 locus, haploid Pet^- ho YJM421 background strains were crossed with laboratory S288c background strains to further analyze the Pet^- phenotype. YJM421 background ho Pet^- strains crossed with S288c background strains yielded Pet⁺ diploids. In addition, tetrads from these crosses showed 2 Pet⁺:2 Pet^- segregation with the Pet^- segregants being capable of producing Pet⁺ papilli (data not shown). Because the Pet^- phenotype was complemented in crosses with laboratory strains and showed single gene segregation in meiosis, it seemed likely that we would be able to complement the Pet^- phenotype with a clone from a laboratory strain library.

Complementation of the Pet^- phenotype:
To complement the unstable Pet^- phenotype of YJM421-derived strains, a YCp50-based (CEN URA3) S. cerevisiae library (ROSE et al. 1987 ) was transformed into the ura3 Pet^- strain V1-13-. Over 26,000 Ura⁺ colonies were screened and 9 Pet⁺ colonies were isolated. The Pet⁺ phenotype of three of these isolates was determined to be plasmid dependent; each of these plasmids had overlapping restriction fragment sizes after EcoRI and HindIII digestion. The insert ends of these three plasmids were sequenced, revealing that the Pet⁺-complementing region spanned the 450,009–462,741 region of chromosome 5.

The presumed sequence for one clone (7A), derived from the Saccharomyces Genome Database, was used to identify available restriction sites and thereby design and construct a series of deletions within the insert. These insert-deletion-containing plasmids were transformed into the Pet^- strain V1-13-. Analysis of these plasmids demonstrated that the COX15 (YER141w) region was required to complement the Pet^- phenotype in YJM421-derived strains. COX15 encodes a protein required for cytochrome oxidase assembly and cox15 mutations have been shown to result in a Pet^- phenotype (GLERUM et al. 1997 ). Crosses of COX15 and cox15 S288c background strains with Pet^- YJM421-derived strains showed that COX15 was required to complement the Pet^- phenotype. Therefore, alteration in COX15 expression or Cox15p function was solely responsible for the variable Pet^- phenotype of YJM421 background strains.

Sequence analysis of COX15 from Pet^- and Pet⁺ YJM421 background strains:
Two alternative hypotheses for the variable Pet^- phenotype would be that COX15 expression or Cox15p function differed between the Pet^- and Pet⁺ variants because of a sequence change (i) at COX15 or (ii) at a locus unlinked to COX15. To distinguish between these two hypotheses, we cloned COX15 by gap repair (ROTHSTEIN 1991 ) from both Pet⁺ and Pet^- YJM421 background strains. A PvuII deletion of clone 7A (7A-p), which removed the COX15 ORF as well as the flanking sequence 2000 bp upstream and 800 bp downstream, was used to gap repair clone COX15 and surrounding sequences from the Pet^- strain V1-13- and its Pet⁺ variant V1-13+. Gap repaired plasmids from both Pet⁺ and Pet^- variants were rescued and sequenced. Sequence analysis of the YJM421-derived COX15 from both Pet⁺ and Pet^- strains showed a C T transition at nucleotide 115 of the 1461-bp open reading frame, which changed the CAA (encoding glutamine) of codon 39 to a TAA (ochre) codon; sequence analysis of the corresponding region of the S288c-derived COX15 showed the expected C at position 115. Since there was no sequence variation at cox15 in Pet⁺ and Pet^- variants, the basis for Pet⁺ variants must lie elsewhere, presumably in the formation of ochre suppressors.

Identification of the ochre suppressor in YJM421:
Libraries containing YJM421-derived DNA that were propagated and amplified in an E. coli host failed to yield yeast transformants with a plasmid-dependent Pet⁺ phenotype (data not shown). This result suggested that a sequence near the putative ochre suppressor was unstable in or deleterious to E. coli. A library not propagated in E. coli, but instead directly transformed into yeast, would presumably contain sequences that would be unstable in or deleterious to E. coli. Therefore, two libraries containing genomic DNA from YJM421 were constructed and directly transformed into a cox15-421-containing S. cerevisiae strain to clone the presumed ochre suppressor.

Twenty-seven Pet⁺ colonies from the HindIII library had a plasmid-dependent Pet⁺ phenotype. Four out of the first eight of these plasmids recovered from the yeast transformants were able to propagate in E. coli and these were amplified for further analysis. All four of these plasmids complemented (or suppressed) the cox15-421 mutation. Although all four plasmids contained multiple HindIII fragments, all had a 10-kb HindIII fragment in common. Six of the Pet⁺ transformants from the XbaI library had a plasmid-dependent Pet⁺ phenotype, only one of which could be propagated in E. coli. This plasmid isolate, named X2, contained a single insert of 15 kb, which, like the plasmids from the HindIII library, contained an internal HindIII fragment of 10 kb. We sequenced the ends of the X2 insert; the resulting sequences corresponded to chromosome 10 positions 342,491–357,278 (Saccharomyces Genome Database). This 14,787-bp fragment contained three tRNA sequences, one of which, tY(GUA)J1, had previously been determined to be SUP7 (HAWTHORNE and MORTIMER 1968 ; OLSON et al. 1977 , OLSON et al. 1979 ).

The cloned tRNAs could be separated into distinct EcoRI fragments, one containing the tyrosine-inserting tRNA tY(GUA)J1 (SUP7-o) and the other fragment containing the other two tRNAs. These fragments were subcloned into the TRP1-containing plasmid pRS314 (SIKORSKI and HIETER 1989 ) and individually transformed into strain V1-22-. Only the plasmid containing the tY(GUA)J1 region conferred Trp⁺ Pet⁺ growth on V1-22-. Sequence analysis of this smaller clone showed that tY(GUA)J1 contained a single G T mutation in the anticodon (ACTGTAA ACTTTAA), which converted it into the ochre suppressor SUP7-o.

Genotyping of tY(GUA)J1/SUP7-o in YJM421:
The genetic (2 Pet⁺ small colony:2 Pet^- large colony segregation) and molecular analysis (the cox15-421 ochre mutation and SUP7-o ochre suppressor) of YJM421 was consistent with the hypothesis that the genotype of YJM421 was cox15-421/cox15-421 SUP7-o/+. To test the hypothesis that the YJM421 genotype was SUP7-o/+, we took advantage of the fact that wild-type tRNA-Tyr loci contain an HpyCH4III site (ACNGT), which overlaps the anticodon (ACTGTA) and is destroyed when the anticodon is mutated to form an ochre suppressor (ACTTTA).

The eight tRNA-Tyr loci of YJM421 and the control S288c background strain S1 (Southern analysis showed that both genetic backgrounds had eight tRNA-Tyr loci; data not shown) were amplified, separately, to yield 200- to 300-bp PCR products. Agarose gel electrophoresis analysis of the undigested PCR products of all tRNA-Tyr loci from both S1 and YJM421 showed the expected band sizes. The HpyCH4III-digested PCR products of all tRNA-Tyr loci from S1 and, with the exception of tY(GUA)J1/SUP7, all tRNA-Tyr loci from YJM421 showed two bands of the expected sizes; that is, none of these loci were ochre suppressors. In contrast, the HpyCH4III-digested PCR product of tY(GUA)J1/SUP7 from YJM421 showed three bands, one corresponding to the undigested or HpyCH4III-resistant PCR product and two bands corresponding to the HpyCH4III-sensitive PCR product (Fig 3A). The results were consistent with the hypothesis that the genotype of YJM421 was SUP7-o/+.

View larger version (50K): In this window In a new window Download PPT slide   Figure 3. Genotyping of YJM421 and one tetrad from YJM421 by (A) HpyCH4III digestion and (B) MASA PCR.

Ochre suppressors are known to confer osmotic sensitivity (SINGH 1977 ). Consistent with the ochre suppressor hypothesis, tetrads from YJM421 showed 2 Osm^s Pet⁺:2 Osm⁺ Pet^- segregation (Fig 1B and Fig C). As a final test of the hypothesis that the genotype of YJM421 was SUP7-o/+ and that SUP7-o was the sole suppressor in YJM421, 14 YJM421 tetrads were phenotyped for Pet and their tY(GUA)J1/SUP7-o PCR products were genotyped using HpyCH4III. Consistent with the SUP7-o/+ hypothesis, the 14 tetrads showed 2 HpyCH4III-sensitive (2 bands) Pet^-:2 HpyCH4III-resistant (one band) Pet⁺ segregation (Fig 3A) and 2:2 segregation in the MASA genotyping assay (Fig 3B).

Mutation frequencies and rates in the YJM421 and S288c genetic backgrounds:
Mutation frequencies and rates were determined for the CAN1 and LYP1 loci and for Pet^- Pet⁺ in the cox15-421 strains V1-1- (isogenic with YJM421) and S183 (isogenic with S288c). To avoid selecting or enriching for Pet⁺, cultures were grown in YEPD containing 5% dextrose (instead of the usual 2% dextrose) and were harvested in log phase. As shown in Table 3, the CAN1, LYP1, and Pet^- Pet⁺ mutation rates and frequencies in the YJM421 and S288c genetic backgrounds were similar, differing at most by approximately twofold.

Genotypes of spontaneous Pet⁺ mutants:
Spontaneous Pet⁺ mutants from three strains (YSA3, V1-1-, and S183) were genotyped (the tRNA-Tyr loci by HpyCH4III digestion of PCR products and MASA and, in cases where all the tRNA-Tyr loci were wild type, COX15 by sequencing) to determine which loci had mutated. Of 129 spontaneous, independently isolated Pet⁺ mutants, 126 (42 in each of the three strains) had mutated one of the tRNA-Tyr loci (in the case of Pet⁺ derivatives of the diploid strain YSA3, one of the two copies of a given locus) to form SUP-o and only three had wild-type tRNA-Tyr at all eight loci. Of the three Pet⁺ mutants with all wild-type tRNA-Tyr loci, the one in S183 had mutated the TAA codon of cox15-421 to CAA, the wild-type glutamine; the other two Pet⁺ mutants, both in V1-1-, had mutated the TAA codon of cox15-421 to TAC (tyrosine) codons, the same amino acid inserted by the tRNA-Tyr SUP-o. Therefore, in spite of the fact that seven base-pair changes could result in a functional Cox15p (TAA CAA, AAA, GAA, TCA, TTA, TAC, or TAT; see below), the cox15-421 ochre codon had a mutation frequency per base pair substantially lower than that of the average tRNA-Tyr locus anticodon mutating to form an ochre suppressor. In addition, only tRNA-Tyr loci mutated to suppress cox15-421.

Mutation frequencies to SUP-o within the tRNA-Tyr gene family:
Aside from a single base-pair polymorphism within the intron, all members of the tRNA-Tyr gene family have identical sequences. Given their identical sequences, one hypothesis would be that the eight tRNA-Tyr loci would be equally likely to mutate to ochre suppressors. The alternative hypothesis would be that members of the dispersed tRNA-Tyr gene family would differ in their ability to mutate to ochre suppressors.

To test these two hypotheses, the data in Table 4 were analyzed by a ² test (SOKAL and ROHLF 2000 ) for 7 d.f. This analysis showed that the biased distribution of SUP-o mutations at the tRNA-Tyr loci was significant for YSA3 and highly significant for and . A G-statistic analysis (SOKAL and ROHLF 2000 ) showed that there were no significant differences (P = 0.05) in the distribution of mutants between the three strains. Therefore, ploidy, mating type (YSA3 vs. V1-1-), and genetic background (V1-1- vs. S183) had no significant effect on the relative mutation frequencies of members of the tRNA-Tyr gene family to ochre suppressors. Since there were no significant differences in the distributions of SUP-o mutants between the three strains, the SUP-o data for all three strains were combined. Analysis of the combined SUP-o mutant data by a ² test showed a highly significant deviation from the expected distribution , again indicating that the members of the dispersed tRNA-Tyr gene family differed significantly in their frequency of mutation to ochre suppressors.

To determine which of the tRNA-Tyr loci mutated to ochre suppressors at frequencies higher or lower than expected, the combined SUP-o mutant data for the three strains were modeled (SOKAL and ROHLF 2000 ); that is, the three tRNA-Tyr loci with the highest and lowest observed mutation frequencies (sup6⁺, sup2⁺, and sup8⁺) that contributed the most to the highly significant ² value were lowered (SUP6-o) and raised (SUP2-o and SUP8-o) to the mean value of the entire set of the eight loci (mean = 15.75). When the modeled data were analyzed by a ² test, the ² statistic was reduced to a nonsignificant value (P = 0.05), indicating that sup6⁺ mutated at a frequency higher than expected while sup2⁺ and sup8⁺ mutated at frequencies lower than expected.

Examination of SUP-o suppression efficiencies for possible correlation with mutation frequencies in the tRNA-Tyr gene family:
Because a correlation between suppressor efficiencies and locus-specific mutation frequencies might offer insight into the mechanism responsible for the locus-specific mutation frequencies, the suppressor efficiencies of S183-derived SUP-o mutants were characterized. The correlation coefficient (SOKAL and ROHLF 2000 ) between mutation frequencies for S183 (Table 4) and suppressor efficiencies for S183-derived SUP-o mutants (Table 5) was found to be not significant .

Locus- and dosage-dependent SUP-o mutant phenotypes:
Because such phenotypes would be relevant to phenotypic variation, the SUP-o mutants were examined to determine if there were locus- and/or dosage-specific phenotypes. In addition to their suppression of nonsense mutations, some SUP-o mutations have an osmotic sensitivity phenotype (Osm^s; SINGH 1977 ). With the exception of the SUP11-o mutants, all of the SUP-o mutants isolated in the haploid strains V1-1- and S183 were Osm^s. All of the 42 SUP-o/+ mutants isolated in the diploid strain YSA3 were Osm⁺; similarly, when 19 SUP-o mutants isolated in the haploid strain S183 were crossed with an isogenic sup⁺ strain, the resulting diploids were Osm⁺. Therefore, osmotic sensitivity was recessive, or dosage dependent, for the SUP-o mutants. The suppression efficiency of S1205/S1228 (SUP11-o/+ cox15-421/cox15-421) was substantially lower than that of haploid SUP11-o strains (Table 5), consistent with suppression efficiency being dosage dependent.

Of all the SUP-o mutants, only the SUP11-o mutants were Osm⁺. One hypothesis for the Osm⁺ phenotype of SUP11-o would be that SUP11-o expression was reduced or abolished under high-osmolarity conditions. However, all SUP11-o cox15-421 strains grew on YEPEG containing 1.5 M KCl. In addition, the suppression efficiency of S1205/S1228 (SUP11-o/+ cox15-421/cox15-421) was not reduced in medium containing 1.5 M KCl (Table 5). Therefore, similar to the Osm⁺ phenotype of SUP-o/+ strains, the Osm⁺ phenotype of SUP11-o was probably due to the suppression efficiency of SUP11-o, the least efficient of the tRNA-Tyr SUP-o mutants, being below a critical threshold for conferring the Osm^s phenotype.

Measuring the suppressor efficiencies of SUP-o mutants required the use of the control plasmid pUKC815, which contains wild-type lacZ⁺ under the control of the PGK1 promoter (FIROOZAN et al. 1991 ; STANSFIELD et al. 1995 ), in both sup⁺- and SUP-o-containing strains. For pUKC815-containing strains grown in the absence of antimycin A, the average ß-galactosidase level of sup⁺ COX15 strains (285.7 Miller units) was considerably higher than that for SUP-o cox5-421 (Pet⁺) strains (Table 5). Therefore, the SUP-o mutants as a class had substantially reduced PGK1 promoter activity.

The effect of the SUP-o mutations as a class on PGK1 promoter activity suggested a correlation between suppressor efficiency, which was SUP-o locus specific, and PGK1 promoter activity. The correlation coefficient (SOKAL and ROHLF 2000 ) between PGK1 promoter activity and suppressor efficiency (Table 5) was highly significant : that is, the more efficient the suppressor, the lower the PGK1 promoter activity.

The low frequency of intragenic COX15 revertants and restricted spectrum of SUP-o mutations:
The low frequency of intragenic cox15-421 revertants could be partially dependent upon the amino acid substitutions at position 39 of Cox15p, which would result in a functional gene product. Similarly, the restricted spectrum of extragenic suppressors of cox15-421 could be dependent upon the amino acid substitutions at position 39 of Cox15p, which would result in a functional gene product, and/or upon the low suppression efficiency of non-tRNA-Tyr SUP-o. Alternatively, relative to the average mutation frequency of the tRNA-Tyr loci, intragenic cox15-421 revertants and ochre suppressors at other tRNA loci might be formed at very low frequencies.

We used oligonucleotide cotransformation (MOERSCHELL et al. 1991 ) to address the question of permissible amino acid substitutions at position 39 of Cox15p. Seven single base-pair substitution mutations can convert a TAA ochre codon to one of seven sense codons: CAA (Gln), TAC (Tyr), TAT (Tyr), AAA (Lys), GAA (Glu), TCA (Ser), and TTA (Leu). Therefore, we transformed the cox15-421 strain S183 with (i) a plasmid containing a natMX4 selectable marker and (ii) oligonucleotides that would convert the ochre codon of cox15-421 to one of the seven sense codons. Nourseothricin-resistant (plasmid-containing) transformants were selected and then screened for Pet⁺. After excluding tRNA-Tyr SUP-o-containing Pet⁺ strains, the region around codon 39 of COX15 of the remaining plasmid-containing Pet⁺ transformants was sequenced.

As expected, wild-type glutamine (oligonucleotide SA98) and tyrosine (inserted by tRNA-Tyr SUP-o; oligonucleotides SA99 and SA100) amino acid substitutions at amino acid 39 of Cox15p were recovered (2, 1, and 3 cotransformants, respectively). In addition, lysine (oligonucleotide SA101), glutamate (oligonucleotide SA102), serine (oligonucleotide SA103), and leucine (oligonucleotide SA104) substitutions were recovered (3, 1, 1, and 1 cotransformants, respectively). Since all of these amino acids resulted in a functional Cox15p, all possible seven intragenic single base-pair substitution mutations of cox15-421 occurred at low frequencies relative to the average tRNA-Tyr locus mutating to SUP-o. In addition, the results suggested that the failure to isolate non-tRNA-Tyr tRNA SUP-o mutants was not due to the inserted amino acid at position 39 resulting in a nonfunctional Cox15p.

Four Pet⁺ Osm⁺ sup11⁺ cox15-421 mutants were also isolated. In principle, the Pet⁺ phenotype of these mutants could have been due to [PSI⁺] or to non-tRNA-Tyr SUP-o mutations. However, after being grown in [psi^-]-inducing conditions, these Osm⁺ sup11⁺ cox15-421 mutants retained their Pet⁺ phenotypes; that is, the Pet⁺ phenotypes of these Osm⁺ sup11⁺ cox15-421 mutants were not due to [PSI⁺]. These results suggested that, at a low frequency relative to the average tRNA-Tyr locus, some tRNA genes other than the tRNA-Tyr loci could apparently mutate to SUP-o and suppress cox15-421.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The high-frequency formation of tRNA-Tyr SUP-o mutants, along with the presence of the cox15-421 reporter mutation, is responsible for the phenotypic variation described in this work. In addition, the analysis of YJM421 and other cox15-421-containing strains explains the more subtle nuances of this intriguing case of phenotypic variation. First, although dominant in terms of their suppressor function, ochre suppressors confer a recessive sporulation defect (ROTHSTEIN et al. 1977 ), which explains why the Pet⁺ Spo⁺ strain YJM421 (HO/HO MATa/MAT cox15-421/cox15-421 SUP7-o/+) produces Pet⁺ Spo^- segregants (HO/HO MATa/MAT cox15-421/cox15-421 SUP7-o/SUP7-o). Second, ochre suppressors have been described as being detrimental to cell growth (SHERMAN 1982 ), which explains why Pet⁺ (SUP7-o) segregants of YJM421 form colonies smaller than those of Pet^- (sup7⁺) segregants. Finally, the high frequency of tyrosine-inserting ochre suppressor formation (relative to cox15-421 COX15) explains the phenotypic instability of the Pet^- segregants and the recapitulation by the resulting Pet⁺ Spo⁺ SUP-o/+ variants of the original behavior of YJM421.

Diversity in SUP-o mutant phenotypes—implications for phenotypic variation:
Although phenotypically indistinguishable on the basis of their Pet⁺ phenotypes, the SUP-o mutants in this study differ clearly in their suppression efficiencies. In turn, the different suppression efficiencies of the SUP-o mutants result in clear differences in other phenotypes. For example, the severity of the sporulation defect of SUP-o mutants correlates with suppression efficiency (ROTHSTEIN et al. 1977 ). All of the tRNA-Tyr SUP-o mutants in this study are osmotic sensitive, except for SUP11-o, which is the least efficient of the tyrosine-inserting ochre suppressors. Finally, the different tRNA-Tyr SUP-o mutations have substantial effects on PGK1 promoter activity that correlates with suppressor efficiency. Since the suppression efficiency of SUP mutants is dosage sensitive and SUP mutants exhibit locus-specific suppression efficiencies, SUP mutant formation is a modulatable phenotypic variation mechanism. Clearly, phenotypic variants that are phenotypically indistinguishable at the gross level may arise from mutational events at different loci; it is only upon closer examination that these variants may have different phenotypes. This SUP-o locus-specific and -dependent diversity and the resulting phenotypic diversity have implications for the study of other phenotypic variations in S. cerevisiae and for the study of phenotypic variation in other yeasts and fungi.

Genetic diversity and heterogeneity—implications for phenotypic variation:
Naturally occurring nonsense mutations have been described in different S. cerevisiae genetic backgrounds, such as suc⁰ (GONZALBO and HOHMANN 1989 , GONZALBO and HOHMANN 1990 ), flo8 (LIU et al. 1996 ), and a "delayed homothallism" ho mutation (TANI et al. 1994 ; EKINO et al. 1999 ). The cox15-421 mutation is another example of a naturally occurring nonsense mutation. In the context of this work, naturally occurring nonsense mutations are genetic-background-specific nonsense suppressor and phenotypic variation reporters. Viewed from the perspective of SUP-o-mediated variation, phenotypic variants in different genetic backgrounds may arise from mutational events in the same gene but will have profoundly different phenotypes due to the different nonsense mutations in each genetic background; a similar argument has been made with respect to [PSI⁺]-mediated phenotypic variation (TRUE and LINDQUIST 2000 ). Therefore, what may appear to be different phenotypic variation systems in unrelated members of the same species could in fact be due to mutational or switching events in the same gene.

Mutation frequencies and rates:
A high frequency of mutators has been reported in specific ecotypes (DENAMUR et al. 2002 ) and in clinical isolates of pathogenic bacteria and these mutators have been suggested to be important for adaptation to a pathogenic life style (LECLERC et al. 1996 ). Given its clinical origin, the possibility that a mutator might play a role in the phenotypic variability seen in YJM421 was intriguing. However, the mutation frequencies and rates in the S288c and YJM421 genetic backgrounds are similar, which argues strongly against the mutator hypothesis.

Base-pair substitution frequencies—the average tRNA- Tyr locus anticodon vs. the cox15-421 ochre codon:
One base-pair substitution mutation in the anticodon (GTA to TTA) can convert any of the eight tRNA-Tyr loci to ochre suppressors. In contrast, seven single base-pair substitution mutations can convert the TAA ochre codon of cox15-421 to one of seven sense codons resulting in a functional Cox15p. Given the gene copy numbers and the number of permissible base substitution mutations per gene, one might expect comparable numbers of tRNA-Tyr SUP-o mutants and intragenic (nonsense to sense) COX15 mutants. However, in our sample size of 129 spontaneous Pet⁺ mutants, we find 126 tyrosine-inserting SUP-o mutants but only 3 intragenic COX15 revertants. Therefore, relative to the ochre codon of cox15-421 mutating to sense, the average tRNA-Tyr locus mutates to SUP-o at a very high frequency.

As argued for [PSI⁺] (TRUE and LINDQUIST 2000 ), translational misreading allows the potential value of nonsense mutations to be assessed. Viewed from the perspective of phenotypic variation, the high mutation frequency of the average tRNA-Tyr locus from wild type to SUP-o, combined with the tRNA-Tyr gene copy number, would facilitate both the retention of naturally occurring nonsense mutations, such as cox15-421, and the testing of these nonsense mutations for potentially beneficial effects.

The position effect on tRNA-Tyr mutation frequency:
One reasonable hypothesis would be that the eight tRNA-Tyr loci, which aside from a 1-bp intron polymorphism are identical in sequence, would mutate to form ochre suppressors at equal frequencies. However, as shown in this work, members of the tRNA-Tyr gene family mutate to form ochre suppressors at significantly different frequencies. To the best of our knowledge, no previous studies have examined the effect of gene position in the genome on mutation frequency. Now that a very substantial position effect on mutation frequency has been found, the challenge is to deduce a mechanism. Toward this end, we examined the association between tRNA-Tyr locus-specific mutation frequencies and multiple factors.

The eight tRNA-Tyr genes have a very short, simple structure: exon 1 (base pairs 1–39), an intron (base pairs 40–53), and exon 2 (base pairs 54–89). Although flanking sequences can affect transcription efficiency, the tRNA-Tyr promoter is intragenic (reviewed in GEIDUSCHEK and TOCCHINI-VALENTINI 1988 ; PAULE and WHITE 2000 ). The sequences of exon 1 and exon 2 are identical in all eight members of the tRNA-Tyr gene family. However, there is a 1-bp sequence polymorphism in the intron. At the polymorphic intron position (base 44) of the two lowest frequency tRNA-Tyr loci, there is a T at sup2⁺ and a C at sup8⁺. Similarly, sup5⁺, sup7⁺, sup11⁺ (three of the four average frequency loci), and sup6⁺ (the highest frequency locus) all have a C at position 44. Therefore, there is no association between the one polymorphism in the tRNA-Tyr loci and their mutation frequencies.

Is there any correlation between tRNA-Tyr gene-centromere or -telomere distances and locus-specific mutation frequencies? The distances (in ascending order) between the tRNA-Tyr genes and their centromeres are sup11 (19 kb), sup3 (39 kb), sup6 (63 kb), sup7 (82 kb), sup5 (99 kb), sup4 (107 kb), sup2 (500 kb), and sup8 (570 kb). Similarly, the distances (in ascending order) between the tRNA-Tyr genes and their telomeres are sup6 (59 kb), sup8 (86 kb), sup11 (103 kb), sup5 (169 kb), sup4 (203 kb), sup3 (288 kb), sup7 (354 kb), and sup2 (586 kb). There is no significant correlation between tRNA-Tyr mutation frequencies and distances to centromeres or telomeres .

Using the Saccharomyces Genome Database, the regions flanking the tRNA-Tyr genes were examined for transposon-related or transcriptional characteristics that might associate with mutation frequency. No Ty1 or Ty3 elements are in the regions immediately flanking any of the tRNA-Tyr genes. A sigma element is near (17 bp) sup2 but no sigma elements are near any of the other tRNA-Tyr genes. The number and location of delta elements relative to the tRNA-Tyr genes are sup2 [one delta element (639 bp) in a 6.5-kb region], sup3 (no delta elements in a 9.8-kb region), sup4 [five delta elements (the nearest being 404 bp) in a 7.3-kb region], sup5 [one delta element (114 bp) in a 6.4-kb region], sup6 [one delta element (3.5 kb) in a 9.4-kb region], sup7 [two delta elements (the nearer being 206 bp) in a 10-kb region], sup8 [one delta element (164 bp) in a 8.8-kb region], and sup11 [one delta element (5.3 kb) in a 10.4-kb region]. Finally, all of the RNA polymerase II transcribed genes near the tRNA-Tyr genes are transcribed at low levels (for most genes, less than one transcript per cell). Therefore, there is no obvious association between Ty elements, delta/sigma elements, or RNA polymerase II transcription levels and tRNA-Tyr mutation frequencies.

One model for the highly skewed mutation frequencies would be that recovery of mutants in specific tRNA-Tyr genes is biased by suppressor efficiency. For example, there might be reduced recovery of more efficient ochre suppressors when their high suppression efficiency is extremely deleterious to the cell. Alternatively, there might be reduced recovery of inefficient ochre suppressors when their low suppression efficiency is incapable of producing sufficient Cox15p. However, there is no significant correlation between suppressor efficiency and mutation frequency. In addition, the spectrum of tRNA-Tyr mutants in isogenic haploid and diploid strains (all SUP-o/+ heterozygotes, which, due to dosage, have reduced suppression efficiency relative to SUP-o-containing haploids) is indistinguishable. These results argue strongly against the skewed mutation frequencies being due to differences in suppressor efficiency.

Another model for the highly skewed mutation frequencies in the eight tRNA-Tyr genes would be that the level of transcription of these genes influences the mutation frequencies. Although RNA polymerase III-transcribed tRNA genes have not been examined, there is evidence in S. cerevisiae for a correlation between transcription of RNA polymerase II-transcribed genes and mutation frequency (DATTA and JINKS-ROBERTSON 1995 ; MOREY et al. 2000 ). However, the lack of correlation between suppressor efficiency, which is presumably a measure of transcription efficiency of these eight identical tRNA-Tyr genes, and mutation frequency argues strongly against the transcription hypothesis.

One aspect of DNA replication that might be relevant to the position effect on mutation frequency is the timing of replication during S phase, which has recently been determined for the entire S. cerevisiae genome. The times of replication of the tRNA-Tyr loci are sup2 (35 min), sup3 (19 min), sup4 (18 min), sup5 (20 min), sup6 (17 min), sup7 (29 min), sup8 (27 min), and sup11 (18 min; data derived from RAGHURAMAN et al. 2001 ). However, there is no significant correlation between the times of replication of the tRNA-Tyr loci and their frequencies of mutation to SUP-o .

Another aspect of DNA replication that might be relevant to the position effect on mutation frequency is the rate of fork movement during replication of the tRNA-Tyr loci, which might influence replication fidelity or DNA repair. The rates of fork movement (in ascending order) of the members of the tRNA-Tyr gene family are sup2 (1.3 kb/min), sup6 (1.4 kb/min), sup7 (1.7 kb/min), sup5 (1.8 kb/min), sup4 (2.1 kb/min), sup3 (2.3 kb/min), sup11 (2.4 kb/min), and sup8 (2.7 kb/min; data derived from RAGHURAMAN et al. 2001 ). One caveat with respect to these fork rate movements is that they were determined for relatively large (compared to the size of tRNA-Tyr loci) regions and local fork rate movements may differ. Indeed, DNA replication fork pause sites have been localized to tRNA loci that are transcribed in opposition to replication forks (DESHPANDE and NEWLON 1996 ). However, with this caveat in mind, there is no significant correlation between the rate of replication and locus-specific mutation frequencies .

The final aspect of DNA replication that might be relevant to the position effect on mutation frequency is gene orientation relative to the nearest origin(s) of replication. Gene origin of replication orientation effects on mutation that are attributable to differences in fidelity in leading- vs. lagging-strand DNA synthesis have been described in E. coli (FIJALKOWSKA et al. 1998 ; MALISZEWSKA-TKACZYK et al. 2000 ). In S. cerevisiae, the effect of gene orientation relative to the origin of replication on the distribution and frequency of mutations has been examined using a plasmid-borne SUP4-o forward mutation detection system where there is only a single origin of replication. While the spontaneous sup4^- mutation frequency is orientation independent, the distribution of sup4^- mutations is influenced by gene orientation (KARTHIKEYAN et al. 2000 ).

The gene origin of replication orientation hypothesis has specific predictions: (1) bidirectionally replicated tRNA-Tyr loci (e.g., roughly equidistant between two efficient origins) should have average mutation frequencies and (2) unidirectionally replicated tRNA-Tyr loci (e.g., physically close to one efficient origin), depending upon the gene orientation, should have either high or low mutation frequencies. The locations of origins of replication, as well as probable replication termini, have recently been determined across the entire S. cerevisiae genome (RAGHURAMAN et al. 2001 ). An examination of origin locations and efficiencies, as well as likely replication termini relative to tRNA-Tyr gene locations, suggests that all of the tRNA-Tyr genes are replicated unidirectionally but the predicted bimodal (high or low, depending upon gene orientation) distribution in mutation frequencies is not observed. However, it is interesting to note that sup2, sup3, and sup8, the three loci with the lowest mutation frequencies, are all transcribed in the same direction as the most likely replication forks. In contrast, sup4, sup5, sup7, and sup11, the four loci with average mutation frequencies, as well as sup6, the locus with the highest mutation frequency, are all transcribed in the opposite direction to the most likely replication forks. This suggests that gene orientation relative to the direction of replication may be a factor in locus-specific mutation frequencies. However, if gene origin of replication orientation is a factor, the distribution of mutation frequencies suggests that one or more additional factors contribute to locus-specific mutation frequencies.

In addition to the analysis of position effects on mutation frequency in this study, the tRNA-Tyr gene family has been used to examine the effect of genomic position on gene conversion. Gene conversion frequencies have been determined for all eight tRNA-Tyr loci in SUP/+ diploids: that is, the same base pair, in the same immediate sequence context, in the same dispersed gene family as this study. There are clear differences between the position effects on gene conversion frequencies and the position effects on mutation frequencies in this study; specifically, gene conversion frequencies for seven of the eight tRNA-Tyr loci did not differ significantly, ranging from 1.5 to 5.3%. However, it is interesting to note that SUP6, which had the highest mutation frequency in this study, had a significantly higher gene conversion frequency of 21% (MORTIMER and MCKEY 1969 ). While it is not clear whether there is any relationship between the position effects on gene conversion and mutation, this gene conversion study provides a clear precedent for significantly different behavior of the same sequence, in this case members of the dispersed tRNA-Tyr gene family, in different genomic locations.

Conclusion:
Like [PSI⁺], tRNA-Tyr suppressors are a mechanism for generating phenotypic variation. There are clear differences between [PSI⁺] and tRNA-Tyr suppressors, such as the amino acid(s) inserted at nonsense codons (apparently unknown for [PSI⁺] but specific for tRNA SUP), suppression efficiency (low for [PSI⁺], high for tRNA-Tyr SUP), and specificity of suppression (all three types of nonsense codons for [PSI⁺] but specific to a single type of nonsense codon for tRNA SUP). However, there are many similarities between [PSI⁺] and tRNA-Tyr suppressors, such as suppression of nonsense mutations, modulatable suppression efficiency, and the frequency of formation and loss.

The finding in this study that members of the tRNA-Tyr gene family are agents of phenotypic variation has multiple implications. First, phenotypic variants can arise by multiple mechanisms; this may involve mutating different genes, such as different tRNA-Tyr loci, as well as mechanisms of loss of SUP-o. Second, variants that are indistinguishable at the gross phenotypic level may be due to mutations at different loci that, upon closer examination, differ phenotypically. Finally, in different backgrounds the same switch or mutational event can produce quite different phenotypes. All of these results have implications for the study of other phenotypic variations in S. cerevisiae and the study of phenotypic variation in other yeasts and fungi. Models for phenotypic variation must take into account not only the phenotypic diversity but also the genetic diversity, such as strain-to-strain differences that could be due to both strain-specific reporter(s) as well as different phenotypic variation mechanisms/systems.

For the first time to our knowledge, this work demonstrates that there is a strong position effect on mutation frequency. A position effect on mutation frequency has a variety of implications in the field of genetics. For example, different regions of the genome may vary in how quickly they diverge and in the amount of sequence diversity they exhibit and, in experimental systems, mutant recovery in different screens and selections may be influenced by genomic location.

With respect to mechanisms for the position effect on mutation frequency, most of the usual suspects appear to be excluded, at least as sole factors, from consideration; it is likely that multiple factors interact to produce the position effects on mutation frequency. Although the tRNA-Tyr gene family offers a powerful, naturally occurring system to examine the effect of genomic position on mutation frequency, the tRNA-Tyr system has two limitations: one can assay only a single base pair for one substitution mutation and one must assay eight loci in every mutant strain. Other mutation reporter systems will be considerably more flexible with respect to the selections for mutations that can be applied, the number and different types of mutations that can be assayed, the assay method(s), and the experimental placement of reporter gene(s) in different genomic locations. In the future, we will use other systems to more explicitly examine the effects of reporter gene position within the genome, including the effects of gene orientation and transcription, on both mutation frequency and mutational spectrum.

	ACKNOWLEDGMENTS

The authors thank J. Heitman and K. Kreuzer for comments and suggestions on this work. This work was funded by the National Institutes of Health (GM-58129).

Manuscript received February 4, 2002; Accepted for publication May 3, 2002.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

AUSUBEL, F. M., R. BRENT, R. E. KINGSTON, D. D. MOORE, J. A. SMITH et al. (Editors), 2002 Current Protocols in Molecular Biology. Wiley Interscience, New York.

DATTA, A. and S. JINKS-ROBERTSON, 1995  Association of increased spontaneous mutation rates with high levels of transcription in yeast. Science 268:1616-1619.[Abstract/Free Full Text]

DENAMUR, E., S. BONACORSI, A. GIRAUD, P. DURIEZ, and F. HILALI et al., 2002  High frequency of mutator strains among human uropathogenic Escherichia coli isolates. J. Bacteriol. 184:605-609.[Abstract/Free Full Text]

DESHPANDE, A. M. and C. S. NEWLON, 1996  DNA replication fork pause sites dependent on transcription. Science 272:1030-1033.[Abstract]

DOWER, W. J., J. F. MILLER, and C. W. RAGSDALE, 1988  High efficiency transformation of E. coli by high voltage electroporation. Nucleic Acids Res. 16:6127-6145.[Abstract/Free Full Text]

DRAKE, J. W., 1991  A constant rate of spontaneous mutation in DNA-based microbes. Proc. Natl. Acad. Sci. USA 88:7160-7164.[Abstract/Free Full Text]

EKINO, K., I. KWON, M. GOTO, S. YOSHINO, and K. FURUKAWA, 1999  Functional analysis of HO gene in delayed homothallism in Saccharomyces cerevisiae wy2. Yeast 15:451-458.[Medline]

ERBS, P., F. EXINGER, and R. JUND, 1997  Characterization of the Saccharomyces cerevisiae FCY1 gene encoding cytosine deaminase and its homologue FCA1 of Candida albicans.. Curr. Genet. 31:1-6.[Medline]

FIJALKOWSKA, I. J., P. JONCZYK, M. M. TKACZYK, M. BIALOSKORSKA, and R