Yeast Frameshift Suppressor Mutations in the Genes Coding for Transcription Factor Mbf1p and Ribosomal Protein S3: Evidence for Autoregulation of S3 Synthesis

Corresponding author: Michael R. Culbertson, R.M. Bock Labs, 1525 Linden Dr., University of Wisconsin, Madison, WI 53706., mrculber{at}facstaff.wisc.edu (E-mail)

Communicating editor: A. G. HINNEBUSCH

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The SUF13 and SUF14 genes were identified among extragenic suppressors of +1 frameshift mutations. SUF13 is synonymous with MBF1, a single-copy nonessential gene coding for a POLII transcription factor. The suf13-1 mutation is a two-nucleotide deletion in the SUF13/MBF1 coding region. A suf13::TRP1 null mutant suppresses +1 frameshift mutations, indicating that suppression is caused by loss of SUF13 function. The suf13-1 suppressor alters sensitivity to aminoglycoside antibiotics and reduces the accumulation of his4-713 mRNA, suggesting that suppression is mediated at the translational level. The SUF14 gene is synonymous with RPS3, a single-copy essential gene that codes for the ribosomal protein S3. The suf14-1 mutation is a missense substitution in the coding region. Increased expression of S3 limits the accumulation of SUF14 mRNA, suggesting that expression is autoregulated. A frameshift mutation in SUF14 that prevents full-length translation eliminated regulation, indicating that S3 is required for regulation. Using CUP1-SUF14 and SUF14-lacZ fusions, run-on transcription assays, and estimates of mRNA half-life, our results show that transcription plays a minor role if any in regulation and that the 5'-UTR is necessary but not sufficient for regulation. A change in mRNA decay rate may be the primary mechanism for regulation.

SUPPRESSORS of frameshift and nonsense mutations have long been used in bacteria, phage, and yeast to identify the RNAs and proteins important in the accuracy of translation. In the yeast Saccharomyces cerevisiae, frameshift suppressors distributed at 25 different loci (SUF1 through SUF25) were identified in previous studies in our laboratory (CULBERTSON et al. 1977 , CULBERTSON et al. 1980 , CULBERTSON et al. 1990 ; CUMMINS et al. 1980 ). Suppressors mapping at 22 of these loci exhibited allele-specific suppression of +1G:C insertions in glycine or proline codons (GABER et al. 1983 ; MATHISON and CULBERTSON 1985 ). Among these, 16 suppressor genes were shown to encode isoacceptors for glycine tRNA, including 2 for glycine tRNA^CCC, 3 for glycine tRNA^UCC, and 11 for glycine tRNA^GCC (GABER and CULBERTSON 1982 ; MENDENHALL et al. 1987 ; BALL et al. 1988 ; MENDENHALL and CULBERTSON 1988 ). Six suppressor genes were shown to encode isoacceptors for proline tRNA, including 4 for proline tRNA^UGG, and 2 for proline tRNA^IGG (CUMMINS et al. 1982 , CUMMINS et al. 1985 ; WINEY et al. 1989 ).

Detailed analyses of two of the yeast suppressors, SUF16 (glycine tRNA^IGG) and SUF8 (proline tRNA^UGG), revealed two different kinds of tRNA structural alterations that lead to suppression of +1 frameshift mutations. SUF16 glycine tRNA suppressors represent a class in which the tRNA contains an extra nucleotide in the anticodon loop such that the loop is extended from seven to eight unpaired nucleotides. By analyzing suppression using all possible combinations of four-base anticodons and four-base glycine codons, it was found that base-pairing at the fourth nucleotide is not required for suppression (GABER and CULBERTSON 1984 ). SUF8 proline tRNA suppressors result from base substitutions at positions 31 or 39 in the anticodon stem of SUF8 tRNA (CUMMINS et al. 1985 ; MATHISON et al. 1989 ). The mutations disrupt base-pairing of the last base pair of the anticodon stem, resulting in a mature tRNA with a novel secondary structure consisting of a four-base anticodon stem and a nine-base anticodon loop. A model based on alternate three-dimensional base stacking conformations of tRNA was originally proposed to explain how tRNAs with eight or nine unpaired nucleotides in the anticodon loop might cause four-base translocation on the ribosome (CURRAN and YARUS 1987 ; CULBERTSON et al. 1990 ). A more recent interpretation based on new information is that frameshift suppressor tRNAs may instead cause slippage of peptidyl-tRNA by a mechanism similar to that proposed for programmed +1 frameshifting (FARABAUGH 2000 ).

In Escherichia coli, suppressor mutations in ribosomal RNA have been found in four regions of the small ribosomal subunit that are associated with ribosomal proteins known to affect translational accuracy when mutated (S4, S5, and S12; OAKES et al. 1990 ; LODMELL and DAHLBERG 1997 ). Similar rRNA mutations have been created in S. cerevisiae, where studies indicate that translational accuracy in eukaryotes is a complex interplay between rRNA, ribosomal proteins, translation factors, and tRNAs that are brought together in three-dimensional space on the ribosome (CHERNOFF et al. 1994 ).

In S. cerevisiae, mutations in 3 of the original 25 frameshift suppressor genes (suf12, suf13, and suf14) exhibited patterns of suppression not typical for the tRNA suppressors described above (CULBERTSON et al. 1982 ). These suppressors were recessive or semidominant and were not allele-specific like the tRNA suppressors. Mutations in suf12 had the broadest spectrum of suppression, acting on +1 insertions in glycine and proline codons as well as on UAG and UGA nonsense alleles. The suf12 gene was found to code for a GTP-binding protein with significant structural similarity to the elongation factor EF-1A (WILSON and CULBERTSON 1988 ). suf12 is synonymous with sup35, which codes for translation termination factor eRF3 (ZHOURAVIEVA et al. 1995 ). Mutations in suf12 presumably suppress nonsense mutations by promoting readthrough of a premature stop codon. It is less clear how suf12 mutations cause frameshift suppression, which requires a change in decoding during elongation to correct the reading frame.

In S. cerevisiae, frameshift/nonsense suppressor mutations were also identified previously in the TEF2 gene, which encodes the elongation factor EF-1A (SANDBAKEN and CULBERTSON 1988 ). Similar mutations have been reported in E. coli (HUGHES et al. 1987 ; VIJGENBOOM and BOSCH 1989 ). EF-1A is directly involved in insuring translational accuracy, which may involve both kinetic proofreading (THOMPSON et al. 1986 ) and allosteric interactions between aminoacyl-tRNA in the ribosomal A site and E-site bound deacyl-tRNA (NIERHAUS 1990 ). Mutations in EF-1A affect both +1 frameshifting and amino acid misincorporation, possibly by affecting the rate of GTP hydrolysis of EF-1A:GTP:aminoacyl-tRNA ternary complexes, which is a critical feature of kinetic proofreading, or possibly by affecting the binding affinity of EF-1A to ribosomal components in the A site. Recessive mutations in yeast TEF2 were not identified in the screens that yielded the 25 frameshift suppressor genes described above because of genetic redundancy. EF-1A is encoded by two nontandem repeated genes, TEF1 and TEF2 (SCHIRMAIER and PHILIPPSEN 1984 ).

In this article we describe the suf13 and suf14 genes and show that both genes code for proteins. We also have determined the sequences of two frameshift mutations, leu2-3 and met2-1, both of which are suppressed by mutations in the SUF12, SUF13, and SUF14 genes (CULBERTSON et al. 1982 ). SUF13 is a single-copy nonessential gene that codes for the transcriptional activator Mbf1 (TAKEMARU et al. 1998 ), which has been shown to act as a bridge between transcription factors and TATA-binding protein. Despite its primary role in transcription, we present evidence that mutations in the SUF13 gene suppress frameshift mutations by a translational mechanism. SUF14 is a single-copy essential gene that codes for ribosomal protein S3 (FINGEN-EIGEN et al. 1996 ). Evidence is presented that mutations in SUF14 cause frameshift suppression by altering translational accuracy of ribosomes.

We examined the expression of the SUF13 and SUF14 genes using changes in gene dosage as an indicator of regulated expression. Whereas SUF13 failed to exhibit altered expression when the gene dosage was increased, we found that the expression of SUF14 was regulated. We present evidence that ribosomal protein S3, the product of SUF14, inhibits its own synthesis by limiting the accumulation of SUF14 mRNA. Using a SUF14lacZ promoter fusion, run-on transcription assays, and estimates of mRNA half-life, our results suggest that the 5' untranslated region (5'-UTR) is necessary but not sufficient for regulation. Transcription appears to play a minor role if any in regulation, whereas S3-mediated changes in the SUF14 mRNA decay rate may be the primary mechanism for regulation.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Strains, media, and reagents:
The S. cerevisiae strains used in this study are listed in Table 1. Plasmids are listed in Table 2. Standard yeast genetic techniques have been described (SHERMAN 1991 ). Yeast transformation was performed by electroporation (GREY and BRENDEL 1992 ). Media for the growth of S. cerevisiae are described by GABER and CULBERTSON 1982 . Nomenclature for media is as follows: SC, synthetic complete; SC-ura, synthetic complete-uracil omitted; SC-ura/leu, synthetic complete with uracil and leucine omitted. Growth was assayed by plating serial dilutions of cells from cell cultures as described by SHIRLEY et al. 1998 . Bacterial strains used for cloning were strain MC1066a (leuB600, trpC9830, pyrF74::Tn5, kan^r, ara, hsdS, hsdM⁺, recA13; SANDBAKEN and CULBERTSON 1988 ) and strain 6507 (leu^- pro^- pyr23::Tn5 recA r_K m_K KanR). Paromomycin sulfate, cycloheximide, hygromycin B, and streptomycin were obtained from Sigma (St. Louis). Geneticin (G418) was obtained from Boehringer Mannheim (Indianapolis). Enzymes were obtained from New England Biolabs (Beverly, MA) and Promega (Madison, WI).

DNA/RNA methods:
Plasmid DNA was isolated as described in MANIATIS et al. 1982 . Yeast genomic and plasmid DNA was prepared by the method of SHERMAN et al. 1982 . Southern analysis was performed as described in MANIATIS et al. 1982 as modified by VAN TOL et al. 1987 . DNA probes from isolated restriction fragments were made with a random primer kit (Pharmacia, Piscataway, NJ). Deoxyoligonucleotides were 5'-end labeled with polynucleotide kinase (New England Biolabs).

DNA was sequenced using the ABI sequenase kit (Perkin Elmer, Norwalk, CT) and an ABI 377 automated fluorescent sequencer. Universal sequencing primers were obtained from New England Biolabs. Additional oligonucleotides (Operon Technologies, Inc., Alameda, CA) were used as primers where necessary to complete sequencing. Sequence analysis and database searches for sequence similarities were performed using the National Center for Biotechnology and the Saccharomyces Genome Database.

To measure the accumulation of his4-713 mRNA by Northern analysis, cells were grown in SC-ura containing histidine and then washed and resuspended in SC-ura medium lacking histidine. RNA was isolated by hot phenol extraction (LEEDS et al. 1991 ) at the time of resuspension and after 3 hr of incubation in medium lacking histidine. The his4-713 mRNA was probed using a 2234-bp XhoI/XbaI DNA fragment that was radiolabeled using a random-primed DNA labeling kit (Boehringer Mannheim). A total of 15 µg of RNA was fractionated on a 1% agarose/formaldehyde gel and transferred to GeneScreen Plus (DuPont, Wilmington, DE; URSIC et al. 1997 ). The ACT1 (actin) transcript used as a loading control was detected with a radiolabeled actin-specific 1144-bp XhoI/KpnI DNA fragment as probe. All other Northern blotting experiments were carried out as described by MANIATIS et al. 1982 as modified by VAN TOL et al. 1987 . The actin transcript was detected using the oligonucleotide 5'-TGTTAATTCAGTAAATTTTC-3', which is complementary to a sequence in the ACT1 open reading frame (ORF). All Southern and Northern blots were analyzed on a Molecular Dynamics (Sunnyvale, CA) phosphorImager.

Run-on transcription assays were performed using permeabilized cells in the presence of [³²]UTP as described by WARNER 1991 . Labeled RNA transcripts from the whole-cell reactions were detected by hybridization to 2 µg of target DNA bound to GeneScreen Plus (DuPont) using hybridization conditions identical to those used for Northern blotting.

Decay rates for mRNA were determined using the methods of HERRICK et al. 1990 and PARKER et al. 1991 in which POLII transcription was blocked by shifting a strain carrying rpb1-1 to the restrictive temperature of 37°. The rate of mRNA decay was monitored by Northern blotting. Decay curves generated using Cricket Graph software were plotted as the percentage of RNA remaining compared to t = 0 vs. time. Lines were generated using a least-squares fit.

Analysis of the leu2-3 and met2-1 mutations:
Plasmid YIp26-LEU2, containing the yeast LEU2 and URA3 genes, was cleaved at a unique KpnI site within the LEU2 coding sequence. The linear plasmid was used to transform strain IEY73, which carries the leu2-3 mutation, to a Ura⁺ phenotype by site-directed integration at the LEU2 locus. All Ura⁺ transformants were Leu^- in phenotype, suggesting that each integrant had probably homogenotized through gene conversion, yielding two homologous leu2-3 alleles separated by vector sequences following transformation. Chromosomal DNA from one Ura⁺ Leu^- transformant was digested with BamHI, ligated, and used to transform E. coli 6507 to ampicillin resistance. A representative plasmid, YIp26-leu2-3, which had a restriction map identical to the parental YIp26-LEU2 plasmid, was used to determine the DNA sequence of the leu2-3 gene.

The wild-type MET2 gene was cloned by complementation of met2-1 in strain 1160 using a YEp24 library (CARLSON and BOTSTEIN 1982 ). Using an in vivo gene conversion mapping method (SANDBAKEN and CULBERTSON 1988 ), the met2-1 mutation showed 23% coconversion with repair of a double-stranded break at a XbaI restriction site in the 5' noncoding region and 37% coconversion with repair of a double-stranded break at a SacI site within the coding region. These results indicated that the met2-1 mutation resided between the XbaI and SacI restriction sites in the 5' one-third of the MET2 gene. DNA carrying the met2-1 mutation was cloned using gapped-duplex repair (ORR-WEAVER et al. 1983 ) by transforming the met2-1 strain 1160 with YCp50-MET2 linearized at both the XbaI and SacI restriction sites. The complete sequence of the XbaI-SacI restriction fragment in the YCp50-met2-1 plasmid was determined.

SUF13 gene:
A restriction map of the SUF13 region is shown in Fig 1A. SUF13 was cloned by complementation of suf13-1. A plasmid designated pJH13.1 was isolated from a YCp50 yeast DNA library that complemented the suf13-1 allele as indicated by a plasmid-dependent slow-growth phenotype on SC-ura/leu medium. YCpSUF13 contains a HindIII/HindIII fragment derived from pJH13.1, which was subcloned into pRS316. To construct YEpSUF13/1, a 3.0-kb fragment containing SUF13 was removed from pJH13.1 by cleavage at SphI and PvuII sites in the vector and inserted into YEp352. To construct YEpSUF13/2, which contains a unique BglII site in the SUF13 gene, an EcoRI fragment was removed from YEpSUF13/1 that contained a second BglII site. To construct YEpSUF13/3, a 1.42-kb HindIII fragment containing SUF13 was removed from YEpSUF13/2 and inserted into YEp351. YIpSUF13 was constructed by inserting the same 1.42-kb HindIII fragment into YIp5.

View larger version (15K): In this window In a new window Download PPT slide   Figure 1. (A) The SUF13 gene and surrounding DNA. The box represents yeast genomic DNA and the solid line represents the vector sequence in YEp352. The arrow represents the position of the SUF13 open reading frame and the direction of transcription. DNA from the leftward HindIII site to the BamHI site was sequenced. The site of insertion of the TRP1 gene at nucleotide +76 in suf13::TRP1 is indicated. The site of the 2-nt deletion in suf13-1 at positions +208–209 is shown. (B) The SUF14 gene and surrounding DNA. The box represents yeast genomic DNA. The arrow represents the position of the SUF14 open reading frame and the direction of transcription. The locations of restriction sites are indicated. DNA from the SphI site to the rightward KpnI site was sequenced. The site of insertion of the TRP1 gene at nucleotide +68 in suf14::TRP1 is indicated. The site of the G A transition mutation corresponding to suf14-1 at nucleotide +322 is indicated.

To generate a SUF13 disruption, a fragment carrying the TRP1 gene was inserted at the unique BglII site in SUF13. To accomplish this, the single-stranded ends of a 1.45-kb EcoRI fragment carrying TRP1 were filled in using T4 DNA polymerase. YEpSUF13/2 was digested with BglII, and the ends were filled in using T4 DNA polymerase. The two blunt-ended fragments were ligated, resulting in the plasmid YEpsuf13::TRP1, which carries the suf13::TRP1 disruption allele. One-step gene replacement was used to integrate suf13::TRP1 at the suf13 locus (ROTHSTEIN 1983 ). A 2.3-kb HindIII/BamHI fragment containing suf13::TRP1 was removed from YEpsuf13::TRP1 by codigestion with HindIII/BamHI and used to transform strain YPH274 to tryptophan prototrophy. The presence of suf13::TRP1 at the suf13 locus was verified by Southern blotting. The resulting diploid, JHY2, was sporulated and tetrads were analyzed to determine if haploid strains carrying the disruption were viable.

The suf13-1 mutation was cloned using gap repair (ORR-WEAVER et al. 1983 ). YCpSUF13 was partially digested with SacI followed by digesting to completion with HpaI. The digestion products were used to transform strain JHY1 to uracil prototrophy. Twenty-three out of 72 transformants grew on SC-ura/leu medium at the same rate as strain JHY1 transformed with a control vector, indicating that the plasmid conferred uracil prototrophy but no longer contained a wild-type SUF13 allele and most likely contained the suf13-1 allele instead. Plasmids from five transformants were rescued into E. coli strain 6507 (HOFFMAN and WINSTON 1987 ). Four of the plasmids had restriction patterns identical to YCpSUF13. These plasmids were retransformed into strain JHY1. Growth on SC-ura/leu was identical to that of strains carrying the suf13-1 allele. One of these plasmids was used to determine the DNA sequence of the suf13-1 gene.

SUF14 gene:
A restriction map of the SUF14 region is shown in Fig 1B. To construct a SUF14 disruption, plasmid YEp352R15 was constructed by digesting YEp352 with EcoRI followed by filling in of the recessed ends and religation. A 1.1-kb SphI/PvuII fragment carrying SUF14 was subcloned into YEp352R15, resulting in the plasmid YEpSUF14/2. A 1.45-kb EcoRI fragment carrying the yeast TRP1 gene was inserted in the unique EcoRI site located within the SUF14 coding region. One-step gene replacement was used to integrate the suf14::TRP1 allele at the suf14 locus (ROTHSTEIN 1983 ). To accomplish this, a fragment carrying suf14:TRP1 was removed from YEpSUF14/2 by digestion with SphI and used to transform strains YPH274 and YPH501 (SIKORSKI and HIETER 1989 ) to tryptophan prototrophy. The insertion of the disruption at the suf14 locus was verified by Southern blotting. The resulting diploids were sporulated and tetrads were analyzed to determine if haploid strains carrying the disruption were viable.

The suf14-1 mutation was cloned using gap repair (ORR-WEAVER et al. 1983 ). To accomplish this, the plasmid YEp352R1H was constructed by digesting YEp352 with HindIII followed by filling in of the recessed ends and religation. A 1.5-kb SphI/DraI fragment containing SUF14 was cloned into YEp352R1H, resulting in the plasmid YEpSUF14/4. The SUF14 coding region was removed from the plasmid by codigestion with EcoRI and HindIII, both of which cut at unique sites in the plasmid. The plasmid lacking SUF14 coding sequences was gel purified and used to transform strain JHY6 to uracil prototrophy. Twenty-two out of 23 transformants grew on SC-ura/leu medium at the same rate as strain JHY6 transformed with a control vector, indicating that the plasmid conferred uracil prototrophy but no longer contained a wild-type SUF14 allele, and most likely contained the suf14-1 allele instead. Plasmids from six transformants were rescued into E. coli strain 6507 (HOFFMAN and WINSTON 1987 ). All six plasmids had restriction patterns identical to YEpSUF14/1. These plasmids were retransformed into strain JHY6. Growth on SC-ura/leu was identical to that of strains carrying the suf14-1 allele. One of these plasmids was used to determine the DNA sequence of the suf14-1 gene.

The suf14fs allele was constructed by digesting YEpSUF14/4 with EcoRI, filling in the four base overhangs with DNA polymerase I large fragment and religating the plasmid with T4 DNA ligase, resulting in plasmid YEpsuf14fs/1. A 3.34-kb SphI/KpnI fragment containing suf14fs was removed and subcloned into YEp351, resulting in plasmid YEpsuf14fs/2.

The allele CUP1pSUF14, which is a fusion of the CUP1 promoter and the SUF14 coding region, was constructed as follows. Starting with plasmid YEpSUF14/4, a BamHI restriction site was inserted immediately 5' of the SUF14 initiator AUG methionine codon using inverse PCR, resulting in plasmid YEpSUF14B. The plasmid YpJ166 is a derivative of YEp352 that contains CUP1 promoter sequences from the plasmid pCUP1pgaIKCYC1 (obtained from D. Ecker) up to an EcoRI site located before the transcriptional start site of the CUP1 gene. To construct YEpCUP1pSUF14, YEpSUF14B was digested with BamHI and a SacI site in the vector sequence. The BamHI site was made blunt by filling in using Klenow and the 3' overhangs were removed by using T4 DNA polymerase. Plasmid YpJ166 was digested with EcoRI, and the site was filled in using Klenow. The blunt-ended fragment from YEpSUF14B was ligated to EcoRI-digested blunt-ended YpJ166. This resulted in a fusion of the CUP1 promoter to the SUF14 ORF. There are no AUG codons in the 5'-UTR of the CUP1-SUF14 fusion such that the first AUG in the fusion mRNA is the SUF14 AUG. A 2.2-kb BamHI fragment containing CUP1-SUF14 was removed from YEpCUP1-SUF14 and inserted into pRS314, pRS315, pRS424, and YEp351, resulting in plasmids named YCpCUP1-SUF14/1, YCpCUP1-SUF14/2, YEpCUP1-SUF14/1, and YEpCUP1-SUF14/2, respectively (see Table 2). Since CUP1pSUF14 and SUF14 mRNA contain different 5'-UTR sequences, it was possible to detect these mRNAs separately on Northern blots. SUF14 mRNA was detected using the oligomer oSUF145UTR (5'-ACCATGGATCAATTCGTTAC-3'), which is complementary to the 20-nucleotide (nt) long 5'-UTR of SUF14 mRNA (FINGEN-EIGEN et al. 1996 ). CUP1-SUF14 mRNA was detected using the oligomer oCUP1-SUF14 (5'-ACCATGGATCAATTCGTTAC-3'), which is complementary to 5'-UTR of CUP1-SUF14 mRNA.

The SUF14 promoter and 5'-UTR were fused to the E. coli lacZ ORF. To construct a multicopy plasmid carrying the fusion, the SUF14 ORF was first removed from YEpSUF14B by codigestion with BamHI and HindIII. The ends were made blunt by filling in the overhangs. A 3.0-kb SalI fragment containing the lacZ ORF was made blunt by using T4 DNA polymerase, and the two blunt-ended fragments were ligated together and inserted into linearized blunt-ended YEp352, resulting in the plasmid YEpSUF14placZ.

Growth in the presence of antibiotics:
A filter disk assay was used to determine the extent of growth in the presence of antibiotics (SINGH et al. 1979 ). Exponentially growing cells were either directly spread on plates (Fig 2A) or mixed with the appropriate molten medium containing 0.8% agar and overlaid onto the corresponding solid medium (Fig 2B and Fig C). Twenty-five microliters (Fig 2A) and 12.5 or 25 µl (Fig 2B and Fig C) of each drug were applied to filter disks (Schleicher & Schuell, Keene, NH; 3/8 inch) from water-soluble stock solutions as follows: 10 µg/ml cycloheximide; 20 mg/ml hygromycin B; 20 mg/ml Geneticin [G418]; 40 mg/ml paromomycin; 20 mg/ml streptomycin). The disks were then placed on plates preseeded with lawns of the strains to be tested. The plates were incubated at 30° for 3 days. The relative extent of growth inhibition by each antibiotic in isogenic mutant and wild-type strains was quantitated by comparing the zone of growth inhibition in millimeters after subtracting 9.5 mm for the diameter of the disk.

View larger version (58K): In this window In a new window Download PPT slide   Figure 2. Effects of antibiotics on the growth of strains carrying suf13 and suf14 alleles. (A) Filter discs containing 25 µl of each of five antibiotics were placed on lawns of cells (MATERIALS AND METHODS). S, streptomycin; C, cycloheximide; H, hygromycin B; G, geneticin (G418); and P, paromomycin. The diameter of the zone of growth inhibition was used as a quantitative measure of the sensitivity of each strain to each antibiotic (Table 5). (A) Lawns of cells were prepared from strains JHY3a (SUF13), JHY3b (suf13-1), and JHY3c (suf13::TRP1). (B) Lawns of cells were prepared from haploid strains JH85.9a [YCp50] (suf14-1) and JH85.9a [pPW14.1] (SUF14). (C) Lawns of cell were prepared from diploid strains 2033 (SUF14/SUF14) and 2033.2c3c (SUF14/suf14::TRP1). In B and C, discs at the top contain 12.5 µl antibiotic/disc. The discs in the bottom panels each contain 25 µl antibiotic.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Suppression of frameshift and nonsense mutations by suf13-1 and suf14-1:
Alleles of the SUF13 and SUF14 genes were originally identified as cosuppressors of the his4-713, leu2-3, and met2-1 mutations. To test the pattern of suppression of suf13-1 and suf14-1, haploid strains were constructed that contained each of the suppressors in combination with a wide variety of +1 frameshift mutations and all three types of nonsense mutations. We found that the patterns of suppression for suf13-1 and suf14-1 were identical (Table 3). Both mutations suppressed the +1 frameshift mutations his4-520, his4-507, his4-504, his4-713, leu2-3, and met2-1. Among nonsense mutations, only trp1-1 (UAG) was suppressed. Since the suppressible frameshifts are all +1 insertions but in different types of codons, altered tRNAs could not in theory suppress all of the mutations. This led us to suspect that SUF13 and SUF14 code for proteins rather than tRNAs. The suf13-1 and suf14-1 suppressors failed to suppress his4-305 and his4-306. These mutations change the initiator methionine codon from AUG to GUG and UUG, respectively (CASTILLO-VALAVICIUS et al. 1990 ).

The leu2-3 and met2-1 alleles had not been previously analyzed at the DNA sequence level. When DNA sequence analysis was performed, we found that leu2-3 and met2-1 are both +1 frameshift mutations (Table 4). The ICR-170-induced mutation corresponding to leu2-3 is a single base-pair insertion of G/C in a consecutive run of four G/C base pairs. The mutation causes a change in the wild-type amino acid sequence from Lys₈₄-Trp-Gly-Thr-Gly-Ser-Val to Lys₈₄-Trp-Gly-Tyr-Arg. The arginine codon is followed by a UAG stop codon that causes premature termination of translation. Likewise, the spontaneous mutation corresponding to met2-1 is a single base-pair insertion of G/C in a consecutive run of three G/C base pairs. The mutation causes a change in the wild-type amino acid sequence from Gly₈₅-Pro-Leu-Leu-Gly to Gly₈₅-Pro-Ser-Ser-Gly. The final glycine codon is followed by a UAA stop that causes premature termination of translation.

The SUF13 gene is synonymous with the MBF1 gene:
A heterozygous SUF13/suf13-1 leu2-3/leu2-3 diploid strain JHY1a (Table 1) grew on SC-leu but at a reduced rate compared to JHY1 (the suf13-1 leu2-3 parent to JHY1a). This result indicated that suf13-1 exhibits semidominant suppression of leu2-3 in the diploid. Since the extent of suppression was distinguishable from that observed in a haploid strain, it was possible to clone SUF13 by complementation of the mutant suf13-1 allele. To accomplish this, strain JHY1 was transformed to a Ura⁺ phenotype using a YCp50-based yeast genomic library containing yeast DNA inserts from a SUF13 strain. Transformants were screened to identify those having an intermediate growth rate on SC-ura/leu medium. Four out of 10,000 transformants displayed this phenotype. Plasmids were rescued from all four transformants into E. coli (HOFFMAN and WINSTON 1987 ). One of the four clones recovered in E. coli, designated pJH13.1, conferred the expected partial complementation upon retransformation of yeast strain JHY1.

To confirm that the clone was likely to contain the SUF13 gene, we showed that a plasmid called YIpSUF13 carrying URA3 and a 1.42-kb HindIII yeast DNA fragment from pJH13.1 (Fig 1A) integrated following transformation at a site closely linked to the suf13 locus on chromosome 15 (GABER et al. 1983 ; ORR-WEAVER et al. 1983 ). URA3 and suf13-1 failed to recombine in a cross. The presence of a DNA insertion resulting from integration into the SUF13 region was confirmed by Southern blotting (data not shown). DNA sequence analysis of the 1.42-kb HindIII fragment in YIpSUF13 revealed a 453-nt ORF within a DNA fragment that complemented suf13-1.

A search of the Saccharomyces Genome Database showed that the ORF was identical to the MBF1 gene, which codes for a known transcription factor (TAKEMARU et al. 1998 ). When an 850-bp BsaA1 fragment of SUF13 was used to probe a Northern blot, an 800-bp mRNA was detected (data not shown). The predicted protein product is 151 amino acids in length. It was shown previously that the MBF1 gene is a single-copy gene that is not essential for growth (TAKEMARU et al. 1998 ). We confirmed the copy number using Southern blotting (data not shown). When JHY2, a diploid heterozygous for the suf13:TRP1 disruption, was sporulated, all of the spores were viable and segregated 2:2 on SC-Trp, indicating that disruption of SUF13 function did not impair viability.

The suf13-1 mutation corresponds to a deletion of two nucleotides:
The suf13-1 allele was cloned (MATERIALS AND METHODS) and the DNA sequence was analyzed. The suf13-1 mutant differs from wild type by a two-base deletion at positions 208–209 in the coding region resulting in a frameshift in the reading frame (Fig 1A). The predicted peptide produced from suf13-1 is 80 amino acids long with the first 69 residues derived from wild-type sequence followed by 11 residues derived from the sequence corresponding to the -2 reading frame. Translation is predicted to terminate at an out-of-frame stop codon.

The suf13-1 mutant differs phenotypically from a complete loss-of-function mutant:
To examine the effects of suf13 mutants on translational fidelity, growth tests were performed in the presence of five antibiotics, including streptomycin, cycloheximide, hygromycin B, geneticin (G418), and paromomycin (Fig 2A; Table 5). The latter three induce changes in the fidelity of translation and increased misreading of the genetic code (SINGH et al. 1979 ). A strain carrying suf13::TRP1 exhibited increased sensitivity to cycloheximide compared to an otherwise isogenic wild-type strain, whereas the sensitivity of a strain carrying suf13-1 was comparable to that of wild type. Furthermore, a strain carrying suf13-1 differed from the strains carrying suf13::TRP1 and SUF13 in their sensitivity to the aminoglycoside antibiotics. The strain carrying suf13-1 was more resistant to paramomycin, G418, and hygromycin B than the strain carrying SUF13. In contrast, the suf13::TRP1 mutation caused the cells to be more sensitive to these drugs than the wild type.

Overall, cells in which the function of SUF13 was disrupted were more sensitive to the antibiotics than wild type, whereas cells containing the frameshift mutation were more resistant. These results suggest that SUF13 plays a role in maintaining the fidelity of translation. In addition, suf13-1 is not identical in phenotype to a null mutant. Despite the presence of a 2-nt deletion predicted to cause frameshifting, which should trigger nonsense-mediated mRNA decay, this allele may produce a truncated protein that is responsible for the differential sensitivities to the aminoglycoside antibiotics.

Suppression of his4-713 by suf13-1 is not due to an elevated his4-713 transcript level:
To test whether suf13-mediated frameshift suppression was due to increased abundance of a suppressible frameshift mRNA, we measured the accumulation of his4-713 mRNA by Northern blotting. We chose to examine his4-713 because its accumulation is not affected by nonsense-mediated mRNA decay (NMD) whereas the analyses of other suf13-suppressible mutations are potentially complicated because they are subject to NMD (LEEDS et al. 1991 , LEEDS et al. 1992 ). In the presence of histidine, his4-713 mRNA accumulated at a threefold higher level in a wild-type SUF13 strain compared to a suf13-1 strain (Table 6). When SUF13 cells were shifted to medium lacking histidine for 3 hr, transcription was derepressed as indicated by about a fourfold increase in the level of his4-713 mRNA. The same extent of derepression was observed for suf13-1 cells 3 hr following starvation for histidine. However, the overall level of his4-713 mRNA was substantially reduced in both the presence and absence of histidine. These results suggest that suppression of his4-713 by suf13-1 is not due to increased abundance of the frameshift mRNA above that observed in an isogenic wild-type SUF13 strain. Suppression is therefore most likely due to a change in the fidelity of translation of the his4-713 message (see DISCUSSION).

The SUF14 gene codes for the ribosomal protein S3:
A heterozygous SUF14/suf14-1 leu2-3/leu2-3 his4-713/his4-713 diploid strain grew on SC-leu/his medium but at a reduced rate compared to haploid strains carrying suf14-1 leu2-3 his4-713. This result indicated that suf14-1 exhibits semidominant suppression of leu2-3 and his4-713 in the diploid. To clone SUF14, strain JHY6 carrying leu2-3, his4-713, and ura3-52 was transformed to a Ura⁺ phenotype with a YCp50 plasmid library (ROSE et al. 1987 ). One out of 10,000 transformants harbored a plasmid that complemented the suf14-1 allele as indicated by a plasmid-dependent slow-growth phenotype on SC-his/leu/ura medium. A plasmid designated pPW14.1 was rescued in E. coli. YepSUF14/1, which contains a 3.34-kb SphI/KpnI yeast DNA fragment from pPW14.1, conferred the expected partial complementation of suf14-1 upon retransformation of strain JHY6 (Fig 1B). A 720-nt open reading frame corresponding to suf14-1 complementing activity was identified by DNA sequence analysis. A search of the Saccharomyces Genome Database showed that the ORF was identical to the RPS3 gene on chromosome 14, which encodes ribosomal protein S3 (OTAKA et al. 1984 ). The suf14 locus had been mapped previously to this region (GABER et al. 1983 ). A 1.0-kb SUF14 transcript was detected by Northern blotting (data not shown).

It was shown previously that RPS3 is a single-copy essential gene (FINGEN-EIGEN et al. 1996 ). We confirmed the copy number using Southern blotting (data not shown). A SUF14::TRP1 disruption was constructed (MATERIALS AND METHODS) and used to confirm that SUF14 is essential for viability. When diploid strains JHY14 and JHY15 that were heterozygous for the disruption were sporulated, all tetrads derived from these diploids segregated 2:2 for lethality. The two viable spores in each tetrad failed to grow on SC-trp and were shown by Southern blotting to carry the wild-type SUF14 allele and not the SUF14::TRP1 disruption (data not shown).

The suf14-1 mutation corresponds to a single nucleotide substitution:
The entire open reading frame for RPS3 and surrounding DNA from a clone carrying suf14-1 were examined by DNA sequence analysis (see MATERIALS AND METHODS). A single A to G transition was found at position 322 in the open reading frame (Fig 1B). This results in a nonconservative amino acid substitution of positively charged lysine with negatively charged glutamic acid.

Mutant suf14 alleles confer increased resistance to aminoglycoside antibiotics:
We determined the relative sensitivity to aminoglycoside antibiotics of haploid strains that carry suf14-1 and diploid strains that carry SUF14/suf14::TRP1 or SUF14/SUF14 (Fig 2B and Fig C; Table 5). All of the strains were resistant to streptomycin. There was no difference in sensitivity to hygromycin B among the haploid strains that carry suf14 or SUF14. However, the suf14-1 strain was twice as resistant to paromomycin and G418 as the SUF14 strain using two different concentrations of the antibiotics applied to the filter disks (twice the amount of drug was needed to produce the same killing zone). Conversely, the suf14-1 strain was twice as sensitive to cycloheximide as the SUF14 strain.

When a SUF14/SUF14 homozygous diploid was compared to a SUF14/suf14::TRP1 diploid, cycloheximide and hygromycin B inhibited growth to the same extent, unlike the suf14-1 haploid strain that was hypersensitive to cycloheximide. At both of the two antibiotic concentrations tested, the SUF14/suf14::TRP1 strain was slightly more resistant to G418 and more than twice as resistant to paromomycin than the SUF14/SUF14 strain. Overall, these results resemble the effects of ribosomal mutations that alter translational fidelity, suggesting that ribosomal protein S3 affects translational accuracy.

Effect of gene dosage on the expression of SUF13 and SUF14:
Regulated genes often fail to be expressed in proportion to gene dosage. To examine the effects of gene dosage on expression, we compared the levels of expression of SUF13 and SUF14 in strains carrying a single chromosomal copy of each gene with strains carrying multiple copies on 2µ plasmids. Southern and Northern blotting were used to quantitate and compare gene copy number with levels of mRNA accumulation (Fig 3).

View larger version (38K): In this window In a new window Download PPT slide   Figure 3. Comparison of gene dosage and mRNA accumulation for SUF13 and SUF14. Strains JHY21a, JHY21b, and JHY21e are transformants of JHY21 that contain the plasmids pRS315 (vector only control), multicopy YEpSUF14/5 (wild-type SUF14), and multicopy YEpSUF13/3 (wild-type SUF13), respectively. (A) A representative Southern blot to determine gene copy number and a Northern blot to determine the levels of mRNA accumulation. A 0.85-kb Bsa1 DNA fragment was used as the probe for SUF13. A 0.65-kb EcoRI/HindIII DNA fragment was used as the probe for SUF14 (see Fig 1A and Fig 3A). ACT1 (actin) mRNA served as a loading control (not shown). DNA used for Southern blotting was digested with EcoRI (SUF14) and HindIII (SUF13). (B) A histogram of the SUF13 and SUF14 gene copy number in JHY21a, JHY21b, and JHY21e compared to the accumulation of SUF14 and SUF13 mRNA in these strains. Whiskers indicate the standard deviation at n = 6.

Strain JHY21e containing the multicopy plasmid YEpSUF13/3 was found by Southern blotting to contain 11.0 ± 1.0 copies of the SUF13 gene compared to untransformed JHY21, which contains a single chromosomal gene copy. When RNA accumulation was assessed by Northern blotting, the level of accumulation in JHY21e was increased 10 ± 0.5-fold. These results suggest that the SUF13 gene is expressed in proportion to the number of gene copies. Strain JHY21b containing the multicopy plasmid YEpSUF14/5 was found by Southern blotting to contain 7.2 ± 0.6 copies of the SUF14 gene compared to the single chromosomal copy present in JHY21. However, the level of RNA accumulation increased by 2.8 ± 0.3-fold. Although this represents a modest increase, it was not proportional to the number of gene copies. We repeated this experiment using transformants of strains JHY16e and JHY16f, which carry a single-copy centromeric plasmid YCpSUF14/2 and the multicopy plasmid YEpSUF14/5. In both strains, the accumulation of SUF14 mRNA on Northern blots was significantly less than expected compared to the number of gene copies determined by Southern blotting.

To test whether some of the promoter sequences required for transcription of SUF14 might be missing in the YEpSUF14/5 plasmid, which contains 0.4 kb of upstream sequences, we performed the same experiment described in Fig 4 using strain JHY21c. This strain contains YEpSUF14/6, which includes 1.3 kb of upstream sequences and is therefore more likely to contain all required promoter sequences. The same lack of proportion of mRNA accumulation relative to gene dosage was observed (not shown). We compared single-copy vs. multicopy expression of the URA3 and LEU2 genes. In both cases, the levels of mRNA accumulation correlated well with the number of gene copies (not shown). Since the dosage independence of SUF14 appeared to be unique, we examined SUF14 expression further to see if this was indicative of a regulatory mechanism controlling expression.

View larger version (44K): In this window In a new window Download PPT slide   Figure 4. Dosage-dependent expression of a frameshift allele of SUF14. Strains JHY21a, JHY21b, and JHY21e are transformants of JHY21 that contain the plasmids pRS315 (vector only control), multicopy YEpSUF14/5 (wild-type SUF14), and multicopy YEpsuf14fs (a frameshift allele of SUF14), respectively. (A) A representative Southern blot to determine gene copy number and a Northern blot to determine the levels of mRNA accumulation. The blots were probed simultaneously with a 0.65-kb EcoRI/HindIII SUF14 fragment and a 0.85-kb BsaA1 SUF13 fragment. ACT1 (actin) mRNA served as a loading control for the Northern blots (not shown). DNA used for Southern blotting was codigested with EcoRI and HindIII. (B) A histogram of the SUF14 gene copy number in strains JHY21a, JHY21b, and JHY21e compared to the accumulation of SUF14 mRNA in these strains. Whiskers indicate the standard deviation at n = 6.

Ribosomal protein S3 is required for dosage-dependent expression of SUF14:
Since three other yeast ribosomal proteins have been shown to autoregulate their own synthesis (ENG and WARNER 1991 ; DABEVA and WARNER 1993 ; LI et al. 1995 ; PRESUTTI et al. 1995 ; FEWELL and WOOLFORD 1999 ), we tested whether ribosomal protein S3 autoregulates expression of the SUF14 gene. If this occurred, it might explain why SUF14 exhibits gene dosage-independent expression.

To accomplish this, a frameshift mutation consisting of a four-base insertion at the EcoRI site in the SUF14 ORF was constructed (see MATERIALS AND METHODS and Fig 1B). This mutation causes premature termination of translation at a downstream out-of-frame UGA codon and is predicted to produce a truncated protein 30 amino acids long. The suf14fs allele is nonfunctional based on its inability to complement the suf14::TRP1 null allele. To avoid destabilization of suf14fs mRNA due to NMD (LEEDS et al. 1991 ), we performed our experiments with the frameshift allele in transformants of strain PLY102. This strain carries the upf1::URA3 allele (LEEDS et al. 1992 ), which inactivates the NMD pathway. The NMD pathway has no effect on expression of wild-type SUF14 (LELIVELT and CULBERTSON 1999 ), indicating that suf14fs mRNA and wild-type SUF14 mRNA have the same stability in strain PLY102.

To assess the effects of multiple nonfunctional copies of SUF14 on mRNA accumulation, we compared the number of gene copies and the extent of mRNA accumulation in strains JHY21a, JHY21b, and JHY21d, all of which carry a chromosomal upf1::URA3 gene disruption (Fig 4). These isogenic strains carry one chromosomal copy of the wild-type SUF14 gene, and carry either an empty vector (JHY21a), multiple plasmid copies of wild-type SUF14 (JHY21b), or one chromosomal copy of SUF14 and multiple plasmid copies of the nonfunctional suf14fs frameshift allele (JHY21d). mRNA levels in JHY21d were 9.0 ± 0.8-fold higher compared to JHY21a and 3- to 4-fold higher than JHY21b. These results indicate that mRNA accumulation in JHY21d was commensurate with gene copy number. Since the only difference between JHY21b and JHY21d is that the former strain produces a functional SUF14 product whereas the latter does not, this result suggests that ribosomal protein S3, the protein product of SUF14, is necessary for gene dosage-independent expression.

SUF14 5' noncoding sequences are required for dosage-dependent expression:
To test whether the SUF14 5' noncoding sequences are required for gene dosage-independent expression, we fused the CUP1 promoter immediately upstream of the AUG initiation codon of SUF14 to create the allele CUP1-SUF14, which lacks the sequences corresponding to the SUF14 promoter and 5'-UTR (MATERIALS AND METHODS). CUP1-SUF14 produces a functional product based on its ability to confer growth in strain JHY16c, which carries a chromosomal suf14::TRP1 disruption and the CUP1-SUF14 allele on plasmid YCpCUP1-SUF14/2. In the presence of 0.2–0.6 mM exogenous copper, mRNA transcribed from CUP1-SUF14 was induced to a level fivefold higher than that of the wild-type SUF14 transcript. Some CUP1-SUF14 transcript can be detected without added copper, presumably because the growth medium already contains a small amount of copper (see Fig 6).

View larger version (55K): In this window In a new window Download PPT slide   Figure 5. Dosage-dependent expression of CUP1-SUF14. Strains JHY21a and JHY21b are transformants of JHY21 that contain the plasmids pRS315 (vector only control) and YEpSUF14/5 (wild-type SUF14), respectively. Strains JHY23b and JHY23c are transformants of JHY23 that contain single-copy plasmid YCpCUP1-SUF14/1 and multicopy plasmid YEp-CUP1-SUF14/1 (CUP1 promoter fused to the SUF14 ORF), respectively. (A) Representative Southern blots to determine gene copy number and Northern blots to determine the levels of mRNA accumulation. DNA used for Southern blotting was digested with EcoRI and HindIII. Northern blots were stripped and reprobed using an oligomer that hybridizes to ACT1 (actin) mRNA, which served as a loading control (not shown). To detect SUF14 mRNA in JHY21a and JHY21b, the blots were probed with a 0.65-kb EcoRI/HindIII DNA fragment from the SUF14 coding region. To detect CUP1-SUF14 mRNA in JHY23b and JHY23c, blots were probed with a 20-mer that anneals to the CUP1 5'-UTR in CUP1-SUF14 mRNA (this probe does not anneal to wild-type SUF14 mRNA). (B) A histogram comparing gene copy number with mRNA accumulation. For strain JHY21b, the gene copy number and mRNA accumulation level were calculated relative to the single chromosomal copy of SUF14 and the corresponding transcript in JHY21a. For strain JHY23c, the gene copy number and mRNA accumulation level were calculated relative to the single centromeric plasmid copy of CUP1-SUF14 and the corresponding transcript in JHY23b. Whiskers indicate the standard deviation at n = 4.

View larger version (39K): In this window In a new window Download PPT slide   Figure 6. Copper-mediated induction of CUP1-SUF14 reduces accumulation of endogenous SUF14 mRNA. Strains JHY23a and JHY23b each carry a single chromosomal copy of the SUF14 gene. JHY23a contains the vector pRS314 and JHY23b contains YCpCUP1-SUF14, which carries the copper-responsive CUP1-SUF14 gene. (A) Northern blots used to determine the relative level of mRNA accumulation for SUF14 mRNA derived from the endogenous SUF14 gene and for CUP1-SUF14 mRNA. Blots were first probed with the oligomer oSUF145UTR, which hybridizes to the 5'-UTR of SUF14 mRNA but fails to anneal to CUP1-SUF14 mRNA. The blots were stripped and reprobed with the oligomer oCUP1-SUF14, which anneals only to the CUP1-SUF14 5'-UTR. ACT1 (actin) mRNA served as a loading control (not shown). The apparent band shift in the top right is an artifact of electrophoresis and does not indicate altered size of SUF14 mRNA. (B) The histograms show the relative levels of accumulation of SUF14 mRNA and CUP1-SUF14 mRNA at increasing copper concentrations. The open and solid boxes in the top show endogenous wild-type SUF14 mRNA levels in a strain lacking SUF14-CUP1 and a strain carrying SUF14-CUP1, respectively. The bottom shows the level of accumulation of the CUP1-SUF14 mRNA in response to added copper.

The accumulation of SUF14 mRNA was examined when either wild-type SUF14 or CUP1-SUF14 was expressed from single-copy or multicopy plasmids (Fig 5). The results shown for wild-type SUF14 mRNA resemble those shown in Fig 4 where mRNA accumulation is not commensurate with gene copy number. However, when expression of CUP1-SUF14 from single- and multicopy plasmids was compared, mRNA accumulation was increased 10.2 ± 0.7-fold in strain JHY23c, which carries the multicopy plasmid. This increase was mirrored by a similar increase in gene copy number, suggesting that the mRNA accumulates in a dosage-dependent manner. This result indicates that 5' noncoding sequences including the SUF14 promoter and/or the wild-type 5'-UTR of SUF14 mRNA are required for dosage-independent accumulation of SUF14 mRNA.

Induction of CUP1-SUF14 reduces accumulation of SUF14 mRNA:
The expression of CUP1-SUF14 was analyzed to assess what effect increased synthesis of ribosomal protein S3 might have on the accumulation of mRNA derived from the endogenous SUF14 gene. Strains JHY23a containing the pRS314 vector and JHY23b containing YCpCUP1-SUF14 were grown in the presence of 0.0–1.0 mM exogenous copper added to SC-Trp medium. The accumulation of CUP1-SUF14 mRNA and mRNA derived from the chromosomal SUF14 gene was monitored separately by Northern blotting (Fig 6).

As the copper concentration was increased, the accumulation of mRNA derived from CUP1-SUF14 increased up to 0.6 mM copper, after which a slight decrease in accumulation was observed. In strain JHY23a containing the pRS314 vector, the accumulation of endogenous SUF14 mRNA derived from the chromosomal SUF14 gene was unaffected by copper concentrations ranging from 0.0 to 0.6 mM after which the level began to decrease. In strain JHY23b containing CUP1-SUF14, the accumulation of endogenous SUF14 mRNA was reduced even when no exogenous copper was added. The reduction in mRNA accumulation in the absence of added copper is probably due to the presence of residual copper in the medium that causes some induction of CUP1-SUF14. The addition of copper to the medium caused a further decrease in endogenous SUF14 mRNA accumulation ultimately to 30–50% of the level of accumulation observed in the absence of the YCpCUP1-SUF14 plasmid. These results indicate that increased synthesis of ribosomal protein S3 derived from CUP1-SUF14 causes reduced accumulation of endogenous SUF14 mRNA.

Effect of copy number on transcription and decay of SUF14 mRNA:
We assessed whether the lower than expected levels of SUF14 mRNA accumulation that occur when the SUF14 gene copy number is increased are due to a change in transcription or mRNA decay. Run-on transcription assays (WARNER 1991 ) were performed to determine the relative rates of synthesis of newly synthesized transcripts in permeabilized cells from strains JHY21a and JHY21b, which carry single and multiple copies of the SUF14 gene. We also examined strain JHY21d, which carries the frameshift allele suf14fs on a multicopy plasmid. All derivatives of JHY21 carry upf1::URA3, which inactivates the NMD pathway such that suf14fs mRNA has the same stability as SUF14 mRNA.

The rates of SUF14 transcription for the three strains were compared to the levels of SUF14 mRNA accumulation as determined by Northern blotting and the SUF14 gene copy number as determined by Southern blotting (Fig 7). JHY21b and JHY21d, which carry SUF14 or suf14fs on a multicopy plasmid, showed a 7.4 ± 0.3-fold and 10.5 ± 0.8-fold increase in gene copy number, respectively. The levels of mRNA accumulation in these strains were increased 3.0 ± 0.2-fold and 9.5 ± 0.7-fold, respectively. These results are consistent with previous experiments showing that increases in mRNA levels were less than expected when SUF14 gene copy number was increased, but were approximately commensurate with increases in gene copy number when the SUF14 gene contained a frameshift mutation that blocked synthesis of S3 protein. By comparison, the transcript levels produced by run-on transcription increased 6.3 ± 0.9-fold in JHY21b and 8.2 ± 1.0-fold in JHY21d. While the increases in the levels of run-on transcripts are somewhat less than the increases in gene copy number, these experiments suggest that the lower than expected level of mRNA accumulation observed in JHY21b cannot be explained solely by a mechanism involving inhibition of transcription. Transcriptional inhibition due to an increased concentration of ribosomal protein S3, if it occurs, is modest.

View larger version (33K): In this window In a new window Download PPT slide   Figure 7. Transcription of SUF14. Transcriptional run-on assays were performed to assess the effects of increased gene copy number on the rate of transcription. JHY21a contains one chromosomal copy of the SUF14 gene and the vector YEp351. JHY21b contains the multicopy plasmid YEpSUF14/5. JHY21d contains the multicopy plasmid YEpsuf14fs/2. Each strain carries upf1::URA3 such that the frameshift mRNA in JHY21d is unaffected by the NMD pathway. (A) A representative run-on transcription experiment to detect labeled RNA from permeabilized cells that was hybridized to a nylon membrane to which 2 µg of target DNA was bound (see MATERIALS AND METHODS). The target DNAs included plasmid DNAs carrying the SUF14 gene and the ACT1 gene. pUC19 plasmid DNA (2 µg) served as a negative control. (B) A histogram of strains JHY21a, JHY21b, and JHY21d comparing mRNA accumulation levels detected by Northern blotting, gene copy number detected by Southern blotting (data from Fig 4), and newly synthesized transcripts detected by run-on transcription. Whiskers indicate the standard deviation at n = 4.

To further examine the potential role of transcription in S3-mediated autoregulation, we constructed fusions between the SUF14 promoter and 5'-UTR to the E. coli lacZ ORF. The accumulation of SUF14-lacZ fusion mRNA was monitored by Northern blotting in strains JHY22b and JHY22d, which carry the SUF14 gene as a single chromosomal copy or as multiple copies on a plasmid (Fig 8). The results indicate that the accumulation of SUF14-lacZ mRNA does not vary when the SUF14 gene copy number is increased. These results suggest that ribosomal protein S3 does not regulate its own synthesis when the SUF14 ORF and 3'-UTR are replaced with lacZ sequences even though the SUF14 promoter and 5'-UTR are present in the fusion transcript. Although the SUF14 5'-UTR appears to play a role in regulation as shown in previous experiments using a CUP1 promoter fusion, it is not sufficient by itself to mediate regulation.

View larger version (50K): In this window In a new window Download PPT slide   Figure 8. Accumulation of SUF14-lacZ promoter fusion mRNA is unaffected by changes in SUF14 gene copy number. Strains JHY22b and JHY22d contain the plasmid YEpSUF14-lacZ. Strain JHY22d also contains the SUF14 gene on the multicopy plasmid YEpSUF14/5. The accumulation of SUF14 mRNA and SUF14-lacZ mRNA was determined by Northern blotting using the probe oSUF145UTR (MATERIALS AND METHODS), which anneals to both transcripts. ACT1 (actin) mRNA served as a loading control (not shown). (A) Northern blot of RNA from three single-colony isolates of JHY22b (lanes 1, 2, and 3) and three of JHY22d (lanes 4, 5, and 6). (B) The histogram shows the relative levels of accumulation of SUF14 mRNA and SUF14-lacZ mRNA.

We measured the effect of changes in SUF14 gene copy number on the half-life of SUF14 mRNA. Decay rates were determined in strains JHY13a and JHY13b, which carry single and multiple copies of SUF14, respectively (Fig 9B). Both strains carry the mutation rpb1-1, which prevents transcription at the restrictive temperature of 37° (HERRICK et al. 1990 ). The decline in mRNA accumulation following temperature shift was determined by Northern blotting (Fig 9A). Decay rates determined from four separate experiments show that the SUF14 mRNA half-life was 15.9 ± 2.6 min in the single-copy SUF14 strain and 9.9 ± 1.3 min in the multicopy SUF14 strain. This represents a 40% relative decline in mRNA half-life when the SUF14 gene copy number is increased.

View larger version (32K): In this window In a new window Download PPT slide   Figure 9. The half-life of SUF14 mRNA decreases in the presence of multiple copies of the wild-type SUF14 gene. Strain JHY13a contains YEp351 (vector control) and a single chromosomal copy of the SUF14 gene. JHY13b contains the multicopy plasmid YEpSUF14/5. Both strains carry rpb1-1, which inhibits transcription at the restrictive temperature of 37°. (A) mRNA accumulation was determined by Northern blotting at the time intervals indicated (in minutes) following a shift to 37°. Blots were probed using oSUF145UTR (MATERIALS AND METHODS). (B) Data were plotted as the percentage RNA remaining vs. time following temperature shift. The lines were drawn using linear regression. Half-lives were determined from four separate experiments of which only one is shown.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Suppression of +1 frameshift mutations:
In addition to the five suppressible his4 frameshifts that are suppressed by suf13 and suf14 mutations (Table 4), there are seven other his4 +1 frameshifts that are not suppressed. These include his4-506 (CC CCC), his4-38 (GG GGG), his4-518 (CC CCC), his4-208 (GG GGG), his4-707 (CCC CCC), his4-206 (GGG GGGG), and his4-519 (GGG GGGG) (MATHISON and CULBERTSON 1985 ). There is no obvious correlation evident in the sequences that suggests why some of the frameshifts are suppressed while others are not suppressed. There is no similarity in the lengths of G or C runs, which range from three to five consecutive nucleotides, the nature of adjacent codons, the positions of stop codons brought into register by the +1 frameshift, or the position of the frameshift in the overall mRNA coding sequence. In addition, there is no common feature suggesting that all of the suppressible +1 frameshifts might be located in regions especially prone to reading frame correction through slippage (QIAN et al. 1998 ; FARABAUGH and BJORK 1999 ; SUNDARARAJAN et al. 1999 ; FARABAUGH 2000 ). It would be of interest in the future to determine whether suf13- or suf14-mediated suppression occurs by a slippage- or nonslippage-based mechanism as proposed by SUNDARARAJAN et al. 1999 .

SUF13 codes for the transcription factor Mbf1p:
In Drosophila melanogaster, it has been shown that Mbf1p functions as a bridging factor that recruits the TATA-binding protein (TBP) to the promoter binding site for the factor FTZ-F1, which results in transcriptional activation of the fushi tarazu gene (LI et al. 1994 ; TAKEMARU et al. 1997 ). In humans, two isoforms of Mbf1p exist, both of which bind to TBP and function in a manner similar to the Mbf1p in insects (KABE et al. 1999 ). hMbf1 localizes to the cytoplasm but is recruited to the nucleus when needed for transcriptional activation. In S. cerevisiae, the MBF1 gene is the functional homologue of the insect and human versions of the gene (TAKEMARU et al. 1998 ). In vivo and in vitro studies indicate that Mbf1p tethers TBP to the DNA-binding domain of the transcription factor Gcn4p. This promotes transcriptional activation of the HIS3 gene, which is one of many known transcriptional targets of regulation by Gcn4p.

In this article we show that suf13-1, which contains a 2-nt deletion at position +208-209 in the yeast SUF13(MBF1) ORF, causes suppression of +1 frameshift mutations in the HIS4, LEU2, and MET2 genes. The suf13-1 mutation is located in the middle of the domain that is required for binding of Mbf1p to Gcn4p and also brings a downstream out-of-frame stop codon in the MBF1 ORF into register upstream of the TBA binding region. Given this, the function of Mbf1p should be severely impaired (TAKEMARU et al. 1998 ). Furthermore, a complete null allele of SUF13(MBF1) resulting from insertion of the TRP1 gene at position +76 upstream of the Mbf1p functional domains also causes suppression of frameshift mutations. These results indicate that suppression is caused by loss of function of Mbf1p. However, the suf13-1 allele apparently retains some function since it confers differential resistance to aminoglycoside antibiotics as compared to suf13::TRP1.

Mechanism of suppression by transcription factor Mbf1p:
The HIS4, LEU2, and MET2 genes are regulated by Gcn4p, indicating that their expression levels should be reduced when Mbf1p function is impaired (TAKEMARU et al. 1998 ). We examined the level of accumulation of his4-713 mRNA in a suf13-1 strain and found that it was substantially reduced compared to the same mRNA in a his4-713 SUF13 strain. This makes it very unlikely that suppression is caused simply by overproduction of a frameshift mRNA in which the frameshift mutation can be read through to produce sufficient functional HIS4 product to confer growth in the absence of exogenous histidine.

It appears more likely that the fidelity of translation of the mRNA is affected. Our results from testing the sensitivity of suf13 alleles to antibiotics that induce translational misreading support this view. Our results suggest that impaired Mbf1p function may indirectly affect the translation of mRNAs. Whereas MBf1p and TBP function together in POLII transcription (TAKEMARU et al. 1998 ), TBP is a component of all three RNA polymerase complexes and may be involved in transcription of POLIII genes including tRNA genes (CORMACK and STRUHL 1992 ). If Mbf1p were to play a role in the expression of POLIII transcripts through binding to TBP such that tRNA transcription was reduced in suf13 mutants, then the reduced levels of tRNAs could induce increased rates of frameshift misreading leading to suppression (FARABAUGH 2000 ). This has been shown to occur for Maf1p, a nuclear protein that affects tRNA biosynthesis in yeast (K. PLUTA, N. C. MARTIN, J. WEISLAW, W. J. SMAGOWICZ, O. LEFEBVRE, D. R. STANFORD, S. R.