SPE3, Which Encodes Spermidine Synthase, Is Required for Full Repression Through NRE^DIT in Saccharomyces cerevisiae

Corresponding author: Jacqueline Segall, Department of Biochemistry, University of Toronto, Toronto, ON M5S 1A8 Canada., j.segall{at}utoronto.ca (E-mail).

Communicating editor: F. WINSTON

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

We previously identified a transcriptional regulatory element, which we call NRE^DIT, that is required for repression of the sporulation-specific genes, DIT1 and DIT2, during vegetative growth of Saccharomyces cerevisiae. Repression through this element is dependent on the Ssn6-Tup1 corepressor. In this study, we show that SIN4 contributes to NRE^DIT-mediated repression, suggesting that changes in chromatin structure are, at least in part, responsible for regulation of DIT gene expression. In a screen for additional genes that function in repression of DIT (FRD genes), we recovered alleles of TUP1, SSN6, SIN4, and ROX3 and identified mutations comprising eight complementation groups of FRD genes. Four of these FRD genes appeared to act specifically in NRE^DIT-mediated repression, and four appeared to be general regulators of gene expression. We cloned the gene complementing the frd3-1 phenotype and found that it was identical to SPE3, which encodes spermidine synthase. Mutant spe3 cells not only failed to support complete repression through NRE^DIT but also had modest defects in repression of some other genes. Addition of spermidine to the medium partially restored repression to spe3 cells, indicating that spermidine may play a role in vivo as a modulator of gene expression. We suggest various mechanisms by which spermidine could act to repress gene expression.

SPORULATION of the yeast Saccharomyces cerevisiae is a process of cellular differentiation that begins when MATa/MAT diploid cells are starved in the presence of a nonfermentable carbon source. As a cell progresses through the events of meiosis and spore wall formation, an ordered series of genetic and morphological changes generates a tetrad of dormant haploid spores that are resistant to environmental insults. A single round of DNA replication is followed by a lengthy prophase during which homologous chromosomes pair and undergo high levels of meiotic recombination. The two meiotic divisions, leading to segregation of homologous chromosomes and then sister chromatids, occur within the nucleus. Prospore membranes begin to form at the spindle pole bodies and expand to engulf each daughter nucleus, as well as some cytoplasm. Deposition of spore wall material then generates a multilayered spore wall, giving rise to four mature spores within the ascal sac (reviewed in KUPIEC et al. 1997 ). The process of spore formation is associated with expression of 4 temporally distinct classes of sporulation-specific genes, referred to as early, middle, mid-late, and late on the basis of their time of expression (reviewed in MITCHELL 1994 ; KUPIEC et al. 1997 ). Sporulation in S. cerevisiae, therefore, provides a useful model for studying the temporal control of gene expression during development.

The mid-late sporulation-specific genes are first activated around the time that the meiotic divisions are being completed and synthesis of the spore membrane has begun. The divergently transcribed genes, DIT1 and DIT2, are the only mid-late sporulation-specific genes thus far identified. These genes encode enzymes that are required for biosynthesis of the dityrosine precursor that is incorporated into the outermost layer of the spore wall (BRIZA et al. 1990 , BRIZA et al. 1994 ). DIT1 and DIT2 are repressed during vegetative growth via a common negative regulatory element, referred to as NRE^DIT (FRIESEN et al. 1997 ). Repression of the DIT1 and DIT2 genes during vegetative growth depends on the Ssn6-Tup1 corepressor acting through NRE^DIT (FRIESEN et al. 1997 ). We presume that a putative NRE^DIT-binding protein recruits the corepressor to the regulatory region of the DIT genes; it is possible, however, that the effect of Ssn6-Tup1 is indirect. NRE^DIT itself is bipartite in nature. One region has similarity to a middle sporulation element (MSE), an element that suffices for activation of middle sporulation-specific genes (HEPWORTH et al. 1995 ; OZSARAC et al. 1997 ) This MSE-like element from the DIT promoter is required for high levels of expression during sporulation in the context of the entire DIT promoter, but has no activity on its own. The adjacent region is essential for repression (FRIESEN et al. 1997 ). Regulation of expression of the DIT genes is complex; a high level of sporulation-specific gene expression requires at least two downstream elements in addition to NRE^DIT (FRIESEN et al. 1997 ).

Repression mediated in yeast by the Ssn6-Tup1 corepressor has been studied extensively. Ssn6 (Cyc8) and Tup1 are involved directly in the repression of genes regulated by glucose and by cell type and have been implicated in the direct repression of genes regulated by oxygen and by DNA damage, as well as genes involved in flocculation (MUKAI et al. 1991 ; KELEHER et al. 1992 ; ZHOU and ELLEDGE 1992 ; ZITOMER and LOWRY 1992 ; ELLEDGE et al. 1993 ; DERISI et al. 1997 ). Genetic and biochemical evidence indicates that Ssn6 and Tup1, neither of which binds to DNA, associate in a complex that is recruited to the promoters of coordinately regulated genes by pathway-specific DNA-binding proteins (WILLIAMS et al. 1991 ; KELEHER et al. 1992 ; TZAMARIAS and STRUHL 1994 ; SMITH et al. 1995 ; TREITEL and CARLSON 1995 ; TZAMARIAS and STRUHL 1995 ; VARANASI et al. 1996 ; REDD et al. 1997 ; reviewed in ROTH 1995 ; STRUHL 1995 ). Two models have been proposed for the mechanism of Ssn6-Tup1-mediated repression. In one model, Ssn6-Tup1 is thought to repress transcription by directing alterations in chromatin structure. In support of this model, Ssn6-Tup1-dependent repression is associated with positioned nucleosomes in the promoters of SUC2 (MATALLANA et al. 1992 ) and STE6 (COOPER et al. 1994 ). In addition, Tup1 has been shown to interact with histones H3 and H4 in vitro (EDMONDSON et al. 1996 ). Ssn6-Tup1 is also thought to mediate repression through effects on the general transcription machinery. In support of this model, partial Ssn6-Tup1-dependent repression of transcription can be recreated in vitro in reactions that contain naked DNA templates (HERSCHBACH et al. 1994 ; REDD et al. 1997 ). It is likely that Ssn6-Tup1 mediates repression in several ways, with direct and indirect mechanisms making different contributions at different promoters (HUANG et al. 1997 ).

In this article, we report the identification and preliminary characterization of genes that are required for complete repression through NRE^DIT, the Ssn6-Tup1-dependent operator controlling mid-late sporulation-specific gene expression. One of these genes is identical to SPE3, which encodes spermidine synthase. We found that cells that could not synthesize spermidine not only failed to support complete repression through NRE^DIT but also had modest defects in repression of other genes. Because addition of spermidine to the medium partially restored repression to spe3 cells, we suggest that spermidine may have a role in vivo as a modulator of gene expression.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Media, growth conditions and genetic methods:
Liquid and solid media have been described (HEPWORTH et al. 1995 ). Sporulation medium consisted of 1% potassium acetate supplemented with the required auxotrophic supplements. Synthetic medium (SD), also referred to as minimal medium, contained 0.7% yeast nitrogen base without amino acids, auxotrophic supplements [40 µg of adenine sulfate/ml, 20 µg of arginine (HCl)/ml, 20 µg of histidine/ml, 60 µg of leucine/ml, 30 µg of lysine (mono HCl)/ml, 20 µg of methionine/ml, 50 µg of phenylalanine/ml, 200 µg of threonine/ml, 40 µg of tryptophan/ml, 30 µg of tyrosine/ml, and 20 µg of uracil/ml], and 2% glucose. For sporulation, yeast strains were grown at 30° in minimal medium (SD) to midlog phase, and the cells were then harvested, washed, and transferred to sporulation medium at a density of ~2 x 10⁷ cells per ml. The time of transfer of cells to sporulation medium is referred to as 0 hr. Standard genetic methods were employed for mating, sporulation, and tetrad analysis (SHERMAN 1991 ). Yeast cells were transformed by the lithium acetate method (GIETZ et al. 1992 ).

Strains:
S. cerevisiae strains used in this study are listed in Table 1. EG123, EG123tup1, and EG123ssn6 were provided by A. JOHNSON and have been described (SCHULTZ et al. 1990 ; KELEHER et al. 1992 ). All other strains were derived from W303-1A and W303-1B. The a/ diploid strain obtained by mating W303-1A and W303-1B is referred to as LP112. DY1702 is a derivative of W303-1A in which the SIN4 gene has been replaced with the sin4::TRP1 allele (JIANG and STILLMAN 1992 ) and was provided by D. STILLMAN. The mutants whose isolation is described in this article are named after the defective allele; e.g., the mutant containing the frd3-1 allele is named Yfrd3-1. Homozygous mutant diploid strains, referred to as YYfrd, were obtained as follows: First, the Yfrd strains were mated with strain W303-1A containing pLG+NRE76, and diploids were selected on SD-Trp-Ura. These heterozygous diploids were sporulated, and Ura⁺ Trp^- colonies derived from MATa spores that contained the frd mutant allele were identified by their defect in repression of the reporter gene. The MATa frd mutants were mated back to the original MAT frd mutant strains, and homozygous diploids were selected on SD-Trp-Ura.

Yspe3::HIS3 was constructed in two steps. First, the wild-type diploid strain LP112 was transformed with a 10.3-kb XbaI-XbaI fragment that had been isolated from pG23Tn42 and that contained an spe3::HIS3 allele. Replacement in a His⁺ transformant of one copy of SPE3 by the spe3::HIS3 allele was confirmed by Southern blot analysis of DNA digested with BglII. The resultant strain was called LP112spe3::HIS3. The spe3::HIS3 allele, which had a Tn1000::HIS3 element (MORGAN et al. 1996 ) inserted 187 nt downstream of the ATG of the SPE3 gene, did not complement the frd3-1 mutation. Second, Yspe3::HIS3 was obtained by sporulation of cells of LP112spe3::HIS3 that had been transformed with pLG+ NRE76. Progeny derived from a haploid MAT spore that failed to fully repress the CYC1-NRE^DIT-lacZ reporter gene were grown in the presence of 5-fluoroorotic acid (5-FOA) (BOEKE et al. 1984 ) to select segregants that had lost pLG+NRE76, generating the strain Yspe3::HIS3.

WA-ROX3-LEU2 was constructed by transforming W303-1A with pRS305-ROX3 (see below) that had been digested with BglII. Integration at the ROX3 locus was confirmed by Southern blot analysis of DNA from Leu⁺ transformants. Yfrd3-1 and Yspe3::HIS3 strains containing an integrated CYC1-lacZ reporter gene or an integrated CYC1-NRE^DIT lacZ reporter gene were constructed by transformation with YIpLG312 and YIpLG+NRE76 (see below) that had been digested with StuI to target integration to the URA3 locus (KELEHER et al. 1992 ). Strains that had a single copy of the reporter gene were identified by Southern blot analysis.

The Escherichia coli strain DH5 was used for propagating plasmids. Strain MC1066 [pyrF74::Tn5(Km^r) leuB trp] was used to select for plasmids containing the yeast LEU2 marker (CASADABAN et al. 1983 ).

Plasmids:
Nonstandard plasmids used in this study are listed in Table 2. Throughout this work, we refer to pLG312(Bgl) (provided by A. MITCHELL), which is a derivative of pLG312 (GUARENTE and MASON 1983 ), as pLG312 and to the reporter gene on this plasmid as CYC1-lacZ. pLG312 contains a unique BglII site, flanked by SalI and XhoI sites, at nucleotide -178, which is located between the CYC1 UASs and the TATA box of the CYC1-lacZ fusion gene; a unique SmaI site is located at nucleotide -312, upstream of the upstream activating sequences (UASs). The plasmid pLG+NRE76, which contains the reporter gene referred to as CYC1-NRE^DIT-lacZ, has a 76-bp fragment containing NRE^DIT inserted into the BglII site of pLG312 (FRIESEN et al. 1997 ).

Plasmid pLG+NRE30 was constructed by annealing the oligonucleotides 5'-GATCCGGGTTCTCTTGCCAAGAAAAAATAAAAAGG-3' and 5'-GATCCCTTTTTATTTTTTCTTGGCAAGAGAACCCG-3' and cloning the double-stranded fragment into the BglII site of pLG312 (GUARENTE and MASON 1983 ).

pRS305-ROX3 was constructed by subcloning an ~2.7-kb HindIII-HindIII fragment containing ROX3 from YCp(33)-ROX3H (a gift from RICHARD ZITOMER) into the HindIII site of pRS305 (SIKORSKI and HIETER 1989 ). pSPE3 · LEU2 was constructed by cloning the 2.9-kb BglII-BglII fragment of pG23, which contains the SPE3 gene, into the BamHI site of pRS315 (SIKORSKI and HIETER 1989 ). YIpLG312 and YIpLG+NRE76 were constructed from pLG312 and pLG+NRE76 by digestion with HindIII and religation, resulting in the deletion of an ~2-kb fragment containing 2-µm sequences.

Isolation of frd mutants:
Strain W303-1BT (MAT) containing pLG+NRE76 was mutagenized to 76% survival with ethyl methanesulfonate (EMS) as described (LAWRENCE 1991 ). Mutagenized cells were plated on SD-Ura at a density of ~800 colony-forming units per plate and incubated at 30° for 2–3 days, at which time the colonies were overlaid with X-Gal-containing agar (see ß-galactosidase assays below). After an additional ~18 hr incubation, cells recovered from colonies that appeared blue were patched in duplicate onto SD-Ura. After growth at 30° for 1–3 days, one set of colonies was retested by the X-Gal overlay assay for derepression of the CYC1-NRE^DIT-lacZ reporter gene. Of ~60,000 colonies tested, 16 colonies that appeared blue on retesting were picked for further study.

Genetic analysis:
Mutants were placed into complementation groups using standard techniques (SHERMAN 1991 ). Allelism with SSN6 and TUP1 was assessed by mating mutants containing pLG+NRE76 with the isogenic strains EG123, EG123ssn6, and EG123tup1 and testing the resulting diploids for repression of the CYC1-NRE^DIT-lacZ reporter gene by the overlay assay. Allelism with SIN4 was assessed in the same way after mating mutants with W303-1A and the isogenic sin4 strain, DY1702. Allelism with ROX3 was assessed by mating Yfrd13-1 containing pLG+NRE76 with WA-ROX3-LEU2 and analyzing tetrads derived from the resulting diploid strain.

To monitor the relative level of expression of various lacZ reporter genes in the mutant strains, cells that had lost pLG+NRE76 were first selected on medium that contained 5-FOA (BOEKE et al. 1984 ). Ura^- derivatives of each mutant were then transformed with pLG312, pLG+NRE76, pLGSS, pLG+2op, and p(-537)DIT1-lacZ. Transformants were patched on SD-Ura plates and incubated at 30°. Patches that had been overlaid with X-Gal-containing agar were examined for relative blueness after 18 hr incubation at 30°.

ß-Galactosidase assays:
ß-Galactosidase activity was measured in extracts of cells as described (HEPWORTH et al. 1995 ). Cells were grown to late log phase in SD-Ura, and then diluted and grown for an additional three to four generations in the same medium before being harvested. The activities reported are averages obtained from three to six cultures. We repeated each experiment one to three times and consistently found that the relative levels of ß-galactosidase activity were similar from one experiment to the next. ß-Galactosidase activity is given in nanomoles of o-nitrophenyl-ß-D-galactopyranoside (ONPG) cleaved per min per mg protein at 28°.

The X-Gal overlay assay has been described previously (BARRAL et al. 1995 ). We used 0.2 mg X-Gal per ml in top agar in the screen for mutants and 0.4 mg X-Gal per ml in top agar for all subsequent experiments. We found that viable cells could be recovered from 80 to 95% of the overlaid colonies after 18 hr incubation; after 40 hr incubation, we could recover viable cells from less than 30% of the colonies (data not shown).

Cloning of FRD3:
Strain Yfrd3-1 containing pLG+NRE76 was transformed with a p366-based (CEN4 ARS1) yeast genomic library (ATCC, a gift of N. MACPHERSON and B. ANDREWS; described in ROSE and BROACH 1991 ). Twenty-four thousand transformants were plated on SD-Leu-Ura medium at a density of ~200 transformants per plate, and the plates were incubated for 3–4 days at 30°. After the colonies had been overlaid with X-Gal-containing agar (see ß-galactosidase assays above) and incubated for an additional 18 hr at 30°, cells were recovered from colonies that remained white. On retesting, two transformants were identified that repressed the CYC1-NRE^DIT-lacZ reporter gene and appeared to have no growth defect. Complementation by the p366-based library plasmids was confirmed by passaging them through MC1066, a leuB strain of E. coli (CASADABAN et al. 1983 ) and reintroducing the plasmids into Yfrd3-1.

DNA sequencing:
The junctions between vector and insert in pG23 and pG51, plasmids which complemented the frd3-1 mutation, were determined by dideoxy sequence analysis of double-stranded DNA (SANGER et al. 1977 ). The primers used, pBR-355T (5'-GGCGACCACACCCGTCCT-3') and pBR-394B (5'-GCGTCCGGCGTAGAGGAT-3'), flank the BamHI site of pBR322 and of its derivative, p366. The sequence was compared to the Saccharomyces Genome Database (http://genome-www.stanford.edu/Saccharomyces/) to identify the chromosomal region that was present in the complementing plasmids.

Transposon mutagenesis:
Tn1000 () transposon mutagenesis has been described (MORGAN et al. 1996 ; SEDGWICK and MORGAN 1994 ). Plasmid DNA was isolated from 24 colonies and transformed individually into Yfrd3-1 that contained pLG+NRE76. The site of insertion of the transposon was identified by sequencing from primers within the transposon as described (MORGAN et al. 1996 ; SEDGWICK and MORGAN 1994 ) and comparing the sequence with that in the Saccharomyces Genome Database.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

In a previous study, we found that multiple regulatory elements within the promoter region of the DIT1 gene of S. cerevisiae contribute to its sporulation-specific expression (FRIESEN et al. 1997 ). One of these elements, termed NRE^DIT, acts as an operator to prevent expression of the DIT1 gene in vegetatively growing cells. This element is also effective in preventing expression of a CYC1-lacZ reporter gene, with this effect being independent of the site of insertion of NRE^DIT and of its orientation (FRIESEN et al. 1997 ). In the current study, we have carried out a genetic screen to identify mutants that are defective in mediating NRE^DIT-dependent repression in vegetative cells. We anticipated that this screen would identify genes that are required specifically for repression through NRE^DIT and genes that serve a general role in repression of gene expression. In our experiments, we compared the efficiency of NRE^DIT-mediated repression in various strains by assessing the expression of two reporter genes, which we refer to as the CYC1-lacZ gene and the CYC1-NRE^DIT-lacZ gene. The CYC1-lacZ reporter gene contains the CYC1 UASs and TATA box and is present in pLG312; the CYC1-NRE^DIT-lacZ reporter gene contains a 76-bp fragment containing NRE^DIT inserted between the CYC1 UASs and TATA box of the CYC1-lacZ gene and is present in pLG+NRE76.

SIN4 contributes to NRE^DIT-mediated repression:
Because we had previously found that repression through NRE^DIT requires the Ssn6-Tup1 corepressor (FRIESEN et al. 1997 ) and because SIN4, which modulates expression of various genes (JIANG and STILLMAN 1992 ; CHEN et al. 1993 ; COVITZ et al. 1994 ), is required for full repression of several Ssn6-Tup1-regulated genes (CHEN et al. 1993 ; WAHI and JOHNSON 1995 ; SONG et al. 1996 ), we tested whether SIN4 contributed to NRE^DIT-mediated repression. Comparison of ß-galactosidase expression from plasmid-borne CYC1-lacZ and CYC1-NRE^DIT-lacZ reporter genes introduced into isogenic wild-type (W303-1A) and sin4 (DY1702) cells indicated that NRE^DIT-mediated repression was reduced 12-fold in sin4 cells (Figure 1); this effect is similar to the ninefold reduction in repression from the 2-Mcm1 operator in sin4 cells (WAHI and JOHNSON 1995 ). We conclude that SIN4 is required to achieve maximal repression through NRE^DIT.

View larger version (12K): In this window In a new window Download PPT slide   Figure 1. NREDIT-mediated repression of the CYC1-lacZ reporter gene is reduced in a sin4 strain. Cells of strains W303-1A (wild-type) and DY1702 (sin4) bearing either pLG312 or pLG+NRE76 were grown in SD-Ura, harvested, and assayed for ß-galactosidase activity. Units of ß-galactosidase are the averages of assays performed on at least three independent cultures. Fold repression refers to the ß-galactosidase activity obtained in the strain containing pLG312 divided by the ß-galactosidase activity obtained in the strain containing pLG+NRE76.

Isolation of mutants with defects in repression through NRE^DIT:
To identify additional genes that might be involved in mediating NRE^DIT-dependent repression, we monitored expression of a plasmid-borne CYC1-NRE^DIT-lacZ reporter gene in cells that had been exposed to the mutagen EMS. By using an overlay assay to detect ß-galactosidase activity in colonies of cells (BARRAL et al. 1995 ), we identified 16 strains from ~60,000 survivors of mutagenesis that expressed the reporter gene. Complementation analysis placed 15 of these mutants into 12 complementation groups, named frd1 through frd13, referring to the fact that the wild-type gene functions in repression of DIT (FRD) genes (Table 3; data not shown). Each mutant strain was named according to its defective allele; for example, Yfrd1-1 is a haploid mutant strain that contains the frd1-1 allele, and YYfrd1-1 is an a/ diploid homozygous for the frd1-1 allele. One mutant strain, which was completely defective in mating, and therefore could not be placed in a complementation group, was not characterized.

Because NRE^DIT-mediated repression requires the corepressor Ssn6-Tup1 (FRIESEN et al. 1997 ) and Sin4 (see above), we determined whether any of our strains contained mutant alleles of SSN6, TUP1, or SIN4. By testing diploid frdX/ssn6, frdX/tup1, and frdX/sin4 strains (JIANG and STILLMAN 1992 ; KELEHER et al. 1992 ) for expression of the CYC1-NRE^DIT-lacZ reporter gene with an X-Gal overlay assay, we found that three strains, Yfrd7-1, Yfrd7-2, and Yfrd7-3, contained mutant alleles of SSN6, one strain, Yfrd8-1, contained a mutant allele of TUP1, and two strains, Yfrd6-1 and Yfrd6-2, contained mutant alleles of SIN4 (Table 3). The identification of mutant alleles of SSN6, TUP1, and SIN4 indicated that this screen could indeed lead to the isolation of genes required for repression through NRE^DIT.

Expression of a CYC1-lacZ reporter gene lacking a UAS is elevated in class I mutants:
As the first step in the preliminary characterization of the Yfrd strains, we determined whether reduced repression through NRE^DIT could be accounted for by a defect in repression of basal transcription. We assessed basal transcription by monitoring expression of a plasmid-borne CYC1-lacZ reporter gene that lacks a UAS. This reporter gene was not expressed in the wild-type strain as monitored by an X-Gal overlay assay, but was expressed in Yfrd11-1, Yfrd12-1, and Yfrd13-1 and, as expected, in Yfrd6-1 and Yfrd6-2, strains that had mutant alleles of SIN4 (Table 3; pLGSS column). We refer to these strains as class I mutants.

We tested several genes that are known to have a role in repressing basal transcription for identity with class I genes. We found that a plasmid-borne version of ROX3/SSN7 complemented the frd13-1 allele. ROX3 is required for repression of other Ssn6-Tup1-regulated genes [CYC7 (ROSENBLUM-VOS et al. 1991 ); SUC2 (SONG et al. 1996 ); and MFA2 (WAHI and JOHNSON 1995 ; CARLSON 1997 )] and has recently been shown to encode a component of the mediator complex of RNA polymerase II holoenzyme (GUSTAFSSON et al. 1997 ). To confirm that frd13-1 was an allele of ROX3, we mated Yfrd13-1 with a wild-type strain that contained a LEU2 marker inserted adjacent to the ROX3 locus and analyzed tetrads derived from this diploid. In 13 of 14 tetrads analyzed, wild-type repression of the CYC1-NRE^DIT-lacZ reporter gene segregated with the LEU2 marker. The other candidate genes not allelic with Class I frd genes were: SPT4, SPT5, and SPT6 (CLARK-ADAMS and WINSTON 1987 ; NEIGEBORN et al. 1987 ; SWANSON et al. 1991 ; SWANSON and WINSTON 1992 ); SPT10 and SPT21 (NATSOULIS et al. 1991 ); SPT16 (MALONE et al. 1991 ); BUR1 (PRELICH and WINSTON 1993 ); RGR1 (SAKAI et al. 1990 ); GAL11 (FASSLER and WINSTON 1989 ; SAKURAI et al. 1993 ); MOT1 (DAVIS et al. 1992A ); genes encoding histones H2A and H2B (CLARK-ADAMS et al. 1988 ; HAN and GRUNSTEIN 1988 ; PRELICH and WINSTON 1993 ); and RPD3 (VIDAL and GABER 1991 ; VIDAL et al. 1991 ) (data not shown).

Class I and class II mutants are defective in repression through both NRE^DIT and the 2-Mcm1 operator:
We anticipated that reduced repression through NRE^DIT in some of the mutants that were isolated in our screen would be due to a general defect in operator-mediated repression, particularly in Ssn6-Tup1-dependent repression. To identify at least a subset of such mutants, we monitored expression of ß-galactosidase in Yfrd strains that harbored pLG+2op, a plasmid that contains the CYC1-lacZ reporter gene under the control of the 2-Mcm1 operator. Repression through this well-characterized operator, which occurs in MAT cells, requires the Ssn6-Tup1 corepressor (KELEHER et al. 1992 ) and several other gene products, including Sin4 and Rox3 (WAHI and JOHNSON 1995 ; CARLSON 1997 ). As expected, the Yfrd strains containing mutations in SSN6, TUP1, SIN4, or ROX3 expressed this reporter gene, as assessed by the X-Gal overlay assay (Table 3; pLG+2op column). Additionally, Yfrd10-1, Yfrd11-1, and Yfrd12-2 expressed the 2-Mcm1 operator-containing reporter gene. We refer to those mutants that were defective in repression through both NRE^DIT and the 2-Mcm1 operator, but that maintained repression of the reporter gene that lacked a UAS as class II mutants (see Table 3) and concluded that they had defects in general operator-mediated repression. WAHI and JOHNSON 1995 also noted that tup1 mutant cells maintain repression of basal transcription.

Candidate genes for FRD10, the only unidentified class II gene, included SRB8, SRB9, SRB10, and SRB11, which encode proteins that interact functionally with the carboxy terminal domain of RNA polymerase II (KIM et al. 1994 ; KOLESKE and YOUNG 1994 ; KUCHIN et al. 1995 ; MYERS et al. 1998 ) and have been shown to be required for repression of MFA2 (WAHI and JOHNSON 1995 ) or SUC2 (SONG et al. 1996 ), but not for repression of basal transcription. Plasmids containing these genes, however, failed to complement the frd10-1 allele, indicating that FRD10 was not one of these SRB genes (data not shown).

We found that the remaining five strains (Yfrd1-1, Yfrd2-1, Yfrd3-1, Yfrd4-1, and Yfrd5-1), which supported repression through the 2-Mcm1 operator (Table 1), also maintained repression of a CYC1-lacZ gene under the control of the URS1 operator (VERSHON et al. 1992 ; data not shown). These five strains, which we refer to as class III mutants, appeared to be specifically defective in repression through NRE^DIT.

In summary, the mutants that we identified on the basis of defects in repression through NRE^DIT were placed into three different classes. Class I and class II mutants were defective in repression through the NRE^DIT and 2-Mcm1 operators (Table 3). Class I mutants, which were also defective in maintaining repression of a gene that lacks a UAS, included two strains with mutations in SIN4, one strain with a mutation in ROX3, and two strains with mutations in unidentified genes. Class II mutants, which maintained repression of a gene that lacks a UAS, included three strains with mutations in SSN6, one strain with a mutation in TUP1, and one strain with a mutation in an unidentified gene (Table 3). By these preliminary criteria, the five mutants of class III, which maintained repression of a gene that lacks a UAS and were effective at mediating repression through the 2-Mcm1 operator, appeared to be specifically defective in repression through NRE^DIT. In further studies, however, we found that Yfrd3-1 and Yfrd4-1 grew slowly in synthetic medium (data not shown). This suggested that the FRD3 and FRD4 genes had roles in addition to their contribution to NRE^DIT-mediated repression.

Effect of frd mutations on expression of a DIT1-lacZ reporter gene:
Our preliminary analysis of the regulation of expression of the DIT1 gene had suggested that there might be a component of repression that is independent of NRE^DIT (FRIESEN et al. 1997 ). This apparent NRE^DIT-independent repression is mediated by the sequence between NRE^DIT and TATA^DIT (FRIESEN et al. 1997 ). We therefore tested the frd strains for their ability to maintain repression of a DIT1-lacZ translational fusion gene that contains DIT1 sequence from upstream of NRE^DIT to the initiator ATG. We have shown previously that this DIT1-lacZ fusion gene is repressed efficiently in wild-type cells during vegetative growth (FRIESEN et al. 1997 ; Table 3). We found that all class I mutants and, as expected, the three class II mutants that contained mutations in SSN6 (Yfrd7-1, Yfrd7-2, and Yfrd7-3) were defective in repressing DIT1-lacZ in vegetatively growing cells (Table 3; pDIT-lacZ column). Yfrd8-1, which contained a mutant allele of TUP1, did not appear to be defective in repressing the DIT1-lacZ reporter gene (Table 3), suggesting that the frd8-1 allele was not a null allele of TUP1. We previously noted that the region of DIT1 between NRE^DIT and TATA^DIT is able to mediate TUP1-dependent repression (FRIESEN et al. 1997 ). It is therefore possible that frd8-1 encodes a form of Tup1 that is more effective at mediating this latter repression than at mediating repression through NRE^DIT. DIT1-lacZ was also repressed in Yfrd10-1, the remaining class II mutant strain, and in the five class III mutant strains (Table 3). These latter mutants, therefore, are defective in components that contribute to NRE^DIT-mediated repression of a heterologous promoter, but that are not essential for repression mediated in the context of a 540-bp region from the promoter of the DIT1 gene.

Quantification of the repression defects in the class III mutants:
We next quantified the repression defects in the Class III mutants by monitoring ß-galactosidase activity in cells harboring pLG+NRE76, pLG+NRE30, or pLG+NRE76 x 2/S, which contain variants of a CYC1-lacZ reporter gene. pLG+NRE76, the plasmid that was used to isolate the mutants, has the 76-bp NRE^DIT-containing fragment (nucleotides -537 to -461 of DIT1) inserted between the CYC1 UAS and TATA box of the CYC1-lacZ reporter gene; pLG+NRE30 has a 30-bp fragment, which contains the downstream portion of NRE^DIT and lacks the MSE-like element (nt -493 to -464), inserted between the CYC1 UAS and TATA box of the CYC1-lacZ reporter gene; and pLG+NRE76 x 2/S has two copies of the 76-bp NRE^DIT fragment upstream of the CYC1 UAS of the CYC1-lacZ fusion gene (FRIESEN et al. 1997 ). The data are presented as a ratio (fold repression) of ß-galactosidase activity measured from the CYC1-lacZ gene, which contains no negative element, to the activity measured from the CYC1-NRE^DIT-lacZ reporter gene in the same strain (Figure 2).

View larger version (24K): In this window In a new window Download PPT slide   Figure 2. Effect of class III mutations on expression of various reporter genes. Fold repression of gene expression is expressed as the ratio of ß-galactosidase activity measured in cells containing the plasmid-borne CYC1-lacZ reporter gene to the ß-galactosidase activity measured in cells of the same strain containing the indicated plasmid-borne operator-containing reporter gene. The bars indicating fold repression represent the averages of assays performed on at least three independent cultures. Cells were grown overnight to late log phase in SD-Ura, diluted, and grown for three to four generations in SD-Ura, and then harvested and assayed for ß-galactosidase activity. Fold repression of gene expression (A): on insertion of NRE76 between the UASCYC1 and TATA box of the CYC1-lacZ reporter gene; (B) on insertion of NRE30 between the UASCYC1 and TATA box of the CYC1-lacZ reporter gene, and on insertion of two copies of NRE76 upstream of the UAS of the CYC1-lacZ reporter gene; and (C) on insertion of the 2-Mcm1 operator between the UASCYC1 and TATA box of the CYC1-lacZ reporter gene. The absolute values of the ß-galactosidase activities for the individual strains containing pLG312 are as follows: W303-1BT (WT), 2400 units; Yfrd1-1, 3600 units; Yfrd2-1, 1400 units; Yfrd3-1, 3100 units; Yfrd4-1, 4300 units; Yfrd5-1, 1900 units. We note that ß-galactosidase activities in different strains were not assayed on the same day and so are not directly comparable between strains.

In the wild-type strain, expression of the CYC1-NRE-lacZ gene contained in pLG+NRE76 was 500-fold lower than was expression of the CYC1-lacZ reporter gene contained in pLG312 (Figure 2A). The 30-bp fragment containing the downstream portion of NRE^DIT was a much less efficient repressor element than the full 76-mer; the 30-bp fragment reduced expression of ß-galactosidase 10-fold in wild-type cells (Figure 2B). As shown previously, ß-galactosidase expression from pLG+NRE76 x 2/S was repressed 40-fold relative to expression of the parental reporter gene in pLG312 (FRIESEN et al. 1997 ; Figure 2B).

Among the class III mutant strains, Yfrd1-1, Yfrd2-1, and Yfrd5-1 were the most defective in repression through the 76-bp NRE^DIT-containing fragment; repression was 14- to 40-fold less efficient than in the wild-type strain (Figure 2A). The mutations in the Yfrd3-1 and Yfrd4-1 strains were less deleterious, with repression through the 76-bp-containing fragment being only 9- and 5-fold less efficient, respectively, than in the wild-type strain. This same pattern was found in repression through the 30-bp fragment representing the downstream portion of the 76-bp fragment and in repression directed by the 76-bp fragment positioned upstream of the CYC1 UAS in the CYC1-lacZ reporter gene (Figure 2B). We conclude that the reduced ability of the class III mutants to mediate NRE^DIT-dependent repression reflects deficiencies in the contribution that the downstream portion of the 76-bp fragment makes to repression.

As a control, we also measured ß-galactosidase activity in cells containing pLG+2op. In wild-type cells, the presence of the 2-Mcm1 operator led to 500-fold repression of the reporter gene (Figure 2C). Four of the class III mutant strains (Yfrd1-1, Yfrd2-1, Yfrd4-1, and Yfrd5-1) maintained efficient repression of this reporter gene. Yfrd3-1, however, was 3-fold less efficient than the wild-type strain in mediating repression through the 2-Mcm1 operator (Figure 2C). This minor deficiency in repression through the 2-Mcm1 operator in Yfrd3-1 had escaped detection in the less sensitive X-Gal overlay assay (Table 3).

Mutation of FRD genes affects sporulation:
We next tested the Class III mutants for their ability to form spores. Although, to date, the NRE^DIT element has been identified only in the promoter region of the divergently transcribed DIT1 and DIT2 genes, we considered it likely that this element would also regulate other as-yet-to-be-identified, mid-late sporulation-specific genes. Although we did not detect derepression of the DIT1-lacZ reporter gene in the class III mutants (Table 3), we speculated that inappropriate expression of some of these other hypothetical mid-late sporulation-specific genes during vegetative growth or early sporulation might lead to defects in spore formation.

Homozygous mutant MATa/MAT frd/frd strains were transferred to sporulation medium, and ascus formation was monitored over a 5-day period. The efficiency of ascus formation in the wild-type strain was 62% after 40 hr in sporulation medium and 72% after 90 hr (Figure 3). The two mutant strains that grew slowly in synthetic medium, YYfrd3-1 and YYfrd4-1, were almost completely deficient in spore formation (<3% of the cells formed asci; Figure 3), and the other three class III mutants, YYfrd1-1, YYfrd2-1, and YYfrd5-1, showed a delay of ~10 hr in the onset of spore formation and about a twofold reduction in the efficiency of ascus formation (Figure 3). Thus, the mutations in the strains assigned to class III led to defects that affected progression through the sporulation program.

View larger version (19K): In this window In a new window Download PPT slide   Figure 3. Time course of ascus formation in wild-type and class III mutant strains. Wild-type diploid a/ cells (LP112) and diploid a/ cells homozygous for the indicated frd alleles were pregrown in SD-Ura, washed, and transferred to sporulation medium at t = 0. Samples were taken at various times, and the number of cells that had formed asci containing two or more spores was determined for >200 cells from each of two independent cultures for each mutant. Ascus formation is expressed as a percentage of total cells counted.

Cloning FRD3:
During our preliminary characterization of the class III mutants, which was carried out with cells grown on synthetic medium, we noticed that the Yfrd3-1 strain grew more slowly than did the wild-type strain (data not shown). We subsequently discovered that growing Yfrd3-1 on rich medium suppressed both its growth defect and its defect in NRE^DIT-mediated repression (data not shown). Growing YYfrd3-1 in rich medium (YEPA), rather than in synthetic medium, before transfer to sporulation medium also restored efficient spore formation (data not shown). The phenotypes of the other class III mutants were independent of the growth medium (data not shown). To gain insight into why the Yfrd3-1 strain had a defect in repression through NRE^DIT that was dependent on its growth medium, we proceeded to clone the FRD3 gene.

Plasmids containing the FRD3 gene were identified by transforming the original Yfrd3-1 strain with a yeast CEN4 LEU2-based genomic library and screening for restoration of repression of the CYC1-NRE^DIT-lacZ gene. Two plasmids, pG23 and pG51, that complemented both derepression of the CYC1-NRE^DIT-lacZ gene and the slow growth of Yfrd31 were isolated (Figure 4). Comparison of sequence obtained from the junctions of the genomic inserts with the Saccharomyces Genome Database revealed that the plasmids contained overlapping inserts from chromosome XVI.

View larger version (10K): In this window In a new window Download PPT slide   Figure 4. Cloning FRD3 by complementation. pG23 (containing sequences from nucleotide 678608–689892 of chromosome XVI, using the numbering of the Saccharomyces Genome Database) and pG51 (containing sequences from nucleotide 683257–694791), which were isolated from a yeast genomic DNA library, complemented both the derepression phenotype and the growth defect of Yfrd3-1. After pG23 had been mutagenized by Tn1000 transposon mutagenesis (SEDGWICK and MORGAN 1994 ; MORGAN et al. 1996 ), plasmids containing transposon insertions were introduced into Yfrd3-1. Three plasmids that were unable to complement the derepression phenotype and the growth defect were identified and partially sequenced to determine the sites of insertion of the transposons. pG23Tn40 contained a transposon insertion at nucleotide 684740, pG23Tn42 contained a transposon insertion at nucleotide 685245, and pG23Tn44 contained a transposon insertion at nucleotide 685062. These three insertions all disrupted the open reading frame (ORF) for the SPE3 gene, which extends from nucleotide 685432–684554. To confirm that the SPE3 gene was responsible for complementation of the mutant phenotype, an ~2.9-kb BglII-BglII fragment that extended from nucleotide 684020–686906 and contained SPE3 was subcloned into a CEN ARS1 plasmid. This plasmid, pSPE3 · LEU2, complemented both the derepression phenotype and the growth defect of Yfrd3-1. Open boxes, ORFs present in the portion of the yeast insert of pG51 that overlaps with the yeast insert present in pG23; B, BglII recognition site.

To determine which of the four ORFs present in the overlapping portions of the genomic inserts of pG23 and pG51 corresponded to FRD3, we subjected pG23 to transposon mutagenesis and identified three plasmids that could no longer complement Yfrd3-1 (see MATERIALS AND METHODS). Sequence analysis with primers that extended outward from the transposon (MORGAN et al. 1996 ) indicated that all three insertions disrupted the ORF designated YPR069c (Figure 4). This ORF has been recently identified as the SPE3 gene, which encodes spermidine synthase (HAMASAKI-KATAGIRI et al. 1997 ). Consistent with this assignment of FRD3 as SPE3, a low-copy plasmid that contained the SPE3 gene complemented both the derepression and the slow growth phenotypes of Yfrd3-1 (Figure 4).

To confirm that SPE3 was FRD3, and not a low-copy suppressor of the frd3-1 mutation, we disrupted the chromosomal copy of the SPE3 gene by integrative transformation with a DNA fragment that contained an spe3::HIS3 allele. This allele contained a Tn1000 transposon with the HIS3 gene inserted 187 nt downstream of the initiator ATG of the SPE3 gene (see MATERIALS AND METHODS). Both the haploid spe3::HIS3 strain and a diploid spe3::HIS3/frd3-1 strain were defective in NRE^DIT-mediated repression and growth on synthetic medium, suggesting that FRD3 was identical to SPE3. We next sporulated the spe3::HIS3/frd3-1 strain. Although we found that mutation of SPE3 reduced spore viability, some tetrads contained four viable spores. All the progeny of 7 such tetrads and of 12 tetrads that had 2 or 3 viable spores were defective in NRE^DIT-mediated repression and growth on synthetic medium. We conclude that spe3::HIS3 and frd3-1 are indeed allelic.

Addition of spermidine to synthetic medium partially suppresses the frd3 phenotype:
The biosynthetic pathway for polyamines in yeast and other organisms has been determined from biochemical and genetic studies (for review see TABOR and TABOR 1984 ; TABOR and TABOR 1985 ). Spermidine synthase, the product of the SPE3 gene, catalyzes the transfer of an aminopropyl group from decarboxylated S-adenosyl methionine to putrescine to give spermidine. In yeast, there is no SPE3-independent pathway for spermidine biosynthesis (COHN et al. 1978 ; HAMASAKI-KATAGIRI et al. 1997 ).

To test whether derepression of the CYC1-NRE^DIT-lacZ reporter gene in Yfrd3-1 cells grown in minimal medium was a direct effect of a deficiency of spermidine in this medium, we monitored repression of this reporter gene in cells grown in synthetic medium that had been supplemented with various concentrations of spermidine (Figure 5). Addition of spermidine to 10^-8 M increased repression of the CYC1-NRE^DIT-lacZ reporter gene ~2-fold; addition of spermidine to 10^-4 M, the highest concentration tested, increased repression of the CYC1-NRE^DIT-lacZ reporter gene ~10-fold. Higher concentrations of spermidine led to significant changes in the pH of the medium (data not shown) and were not tested for their effects on gene expression. Addition of spermidine to 10^-4 M to our presporulation synthetic medium also restored ascus formation in the YYfrd3-1 strain to the wild-type level (data not shown). Addition of spermidine to the sporulation medium only, however, did not permit efficient ascus formation (data not shown).

View larger version (26K): In this window In a new window Download PPT slide   Figure 5. Addition of spermidine to the growth medium suppresses the derepression phenotype of Yfrd3-1. Cells of strains W303-1BT (denoted WT) and Yfrd3-1 containing the indicated plasmids were grown overnight in SD-Ura containing various concentrations of spermidine, and then diluted in the same medium grown for three to four generations, harvested, and assayed for ß-galactosidase activity. ß-Galactosidase activities are averages of assays performed on at least three independent cultures. pLG+NRE76 and pLG+2op contain a CYC1-lacZ reporter gene with NREDIT and the 2-Mcm1 operator, respectively, inserted between the CYC1 UAS and TATA box.

These experiments clearly indicated that it was an absence of spermidine that led to deficient repression of the CYC1-NRE^DIT-lacZ reporter gene in the Yfrd3-1 strain and to the sporulation defect in YYfrd3-1. In contrast, the two- to threefold defect in repression through the 2-Mcm1 operator that we had observed in the Yfrd3-1 strain grown in minimal medium (Figure 2C) was not suppressed by spermidine (Figure 5). It is possible that exogenous spermidine was required at a concentration higher than 10^-4 M to correct for this latter defect.

Phenotype of a spe3::HIS3 allele:
We next compared the phenotype of Yfrd3-1 with the phenotype of Yspe3::HIS3, a strain that contained a disrupted spe3::HIS3 allele (see above). Yspe3::HIS3 was viable on minimal medium although, like Yfrd3-1, it grew more slowly than did the wild-type strain. The plasmid-borne CYC1-NRE^DIT-lacZ reporter gene was derepressed to the same extent in Yfrd3-1 and Yspe3::HIS3 grown in minimal medium; NRE^DIT-mediated repression was 43-fold in the mutant cells vs. 350-fold in wild-type cells (Figure 6A). Overall, therefore, the mutant strains were ~8-fold less efficient than the wild-type strain at mediating repression through NRE^DIT. Supplementing the medium with spermidine restored NRE^DIT-mediated repression in the mutant strains to within threefold of the level of repression observed in the wild-type strain (Figure 6A). Similar results were obtained on examination of expression of an integrated CYC1-NRE^DIT-lacZ reporter gene; NRE^DIT-mediated repression was 22- and 31-fold in Yfrd3-1 and Yspe3::HIS3, respectively, vs. 110-fold in the wild-type strain (Figure 6B). Addition of spermidine to the medium reduced expression of the integrated reporter gene in both mutant strains to the same low level as in the wild-type strain (Figure 6B). Because the extent of derepression of the CYC1-NRE^DIT-lacZ gene in Yspe3::HIS3 was no greater than in Yfrd3-1, we conclude that the frd3-1 allele was a null allele.

View larger version (43K): In this window In a new window Download PPT slide   Figure 6. Expression of CYC1-lacZ and CYC1-NREDIT-lacZ genes in wild-type and frd3 strains. (A) ß-Galactosidase activity was measured in cells of strains W303-1BT (WT), Yfrd3-1, and Yspe3::HIS3 harboring, as indicated, a CYC1-lacZ reporter gene on a high-copy plasmid or a CYC1-NREDIT-lacZ reporter gene on a high-copy plasmid. The data in the - column are from cells that had been grown overnight in SD-Ura with no exogenous spermidine, diluted, and grown for three to four generations in the same medium before being harvested and assayed for ß-galactosidase activity. The data in the + column are from cells grown as described above, but in SD-Ura that contained 10-4 M spermidine. Units of ß-galactosidase activity are given as the averages of assays performed on at least three independent cultures. Fold-repression refers to the effect of the NRE on the activity of the CYC1 UAS in the indicated strain and in the indicated growth medium; i.e., the ß-galactosidase activity of a strain containing pLG312 was divided by the ß-galactosidase activity of the same strain containing pLG+NRE76 and grown in the same medium. (B) ß-Galactosidase activity was measured as described above in cells of strains W303-1BT (WT), Yfrd3-1, and Yspe3::HIS3 that contained, as indicated, a single-copy, integrated version of the CYC1-lacZ reporter gene or a single-copy, integrated version of the CYC1-NREDIT-lacZ reporter gene.

In these experiments, we found that the frd3-1 and spe3::HIS3 alleles led to a modest increase in expression of our control CYC1-lacZ reporter gene in cells grown in minimal medium; this increase was suppressed by addition of spermidine to the medium (Figure 6). We note that throughout this study we have reported the efficiency of repression relative to expression of the control CYC1-lacZ reporter gene in the same strain; thus, the changes in repression that we present as fold-effects reflect changes in NRE^DIT activity only.

In summary, we uncovered SPE3 (FRD3) as a gene that is required for efficient repression through NRE^DIT, but is dispensable for repression of basal transcription. In our preliminary characterization of the mutant FRD strains, we had classified Yfrd3-1 as a class III mutant because it appeared to be specifically defective in NRE^DIT-mediated repression. We have reassigned Yfrd3-1 to the class II group of mutants, however, because we noted that Yfrd3-1, in addition to a conditional slow-growth phenotype, had general defects in gene expression. Mutation of SPE3 (FRD3) not only led to less efficient repression through NRE^DIT but also caused a two- to threefold reduction in repression through the 2-Mcm1 operator and a two- to threefold increase in expression of our control CYC1-lacZ reporter gene when cells were grown in minimal medium. Because some of these defects could be partially suppressed by the addition of spermidine to the medium, we conclude that one role of spermidine may be to modulate gene expression.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

In this study, we have further characterized NRE^DIT-mediated repression. This negative element directs Ssn6-Tup1-dependent repression of the sporulation-specific DIT1 and DIT2 genes in vegetative cells (FRIESEN et al. 1997 ). We have demonstrated that SIN4 is required to achieve full repression of a CYC1-NRE^DIT-lacZ reporter gene. NRE^DIT thus becomes the third Ssn6-Tup1-dependent element that is known to require SIN4 for full repression. SIN4, which encodes a component of the RNA polymerase II holoenzyme (LI et al. 1995 ), has been shown previously to contribute to the Ssn6-Tup1-dependent repression of MFA2 (CHEN et al. 1993 ) and SUC2 (SONG et al. 1996 ), as well as to repression and activation of a number of Ssn6-Tup1-independent genes (JIANG and STILLMAN 1992 ; CHEN et al. 1993 ; COVITZ et al. 1994 ). Because both Sin4 (JIANG and STILLMAN 1992 ; JIANG et al. 1995 ; MACATEE et al. 1997 ) and the Ssn6-Tup1 complex (ROTH et al. 1992 ; COOPER et al. 1994 ; GAVIN and SIMPSON 1997 ) have been implicated in modulation of chromatin structure, we consider it likely that NRE^DIT-mediated repression occurs, at least in part, by regulation of chromatin structure.

Three classes of frd mutants:
To gain further insight into the mechanism of NRE^DIT-mediated repression, we isolated mutants that were defective in repression of a CYC1-NRE^DIT-lacZ reporter gene. We tentatively assigned these FRD (function in repression of DIT) mutants, which represented 12 complementation groups, to three classes. We note that although some genes were isolated more than once, this screen was not saturating.

Class I mutants, in which basal transcription was increased, included strains with mutations in SIN4 and ROX3/SSN7 and two strains with mutations in unidentifed genes. ROX3, which encodes a component of the mediator complex of RNA polymerase II holoenzyme (GUSTAFSSON et al. 1997 ), has been shown to play a role in repression of three other genes regulated by Ssn6-Tup1: CYC7 (ROSENBLUM-VOS et al. 1991 ), SUC2 (SONG et al. 1996 ), and MFA2 (CARLSON et al. 1997). We note that previous studies of rox3/ssn7 mutants did not test for a defect in repression of basal transcription. Class II, which consisted of mutants that had defects in operator-mediated repression but maintained repression of basal transcription, included strains with mutations in SSN6 and TUP1 and one strain with a mutation in an unidentified gene.

Mutant strains that appeared to be specifically defective in NRE^DIT-mediated repression were assigned to class III. These strains are good candidates for having a mutation in a gene(s) encoding an NRE^DIT-binding protein(s). We found that mutation of the class III FRD genes caused only a partial loss of repression through NRE^DIT. It is possible that these genes encode proteins that do not have a key role in establishing a repression complex or that these frd alleles are not null alleles. The incomplete defects in repression seen for the class III FRD mutants could also reflect partial functional redundancies among the class III FRD gene products.

Identification of FRD3 as SPE3:
A major finding of this study was the demonstration that FRD3 is identical to SPE3, the gene encoding spermidine synthase. SPE3 has been cloned recently as a gene complementing the spermidine auxotrophy of a spe3-1 mutant strain (HAMASAKI-KATAGIRI et al. 1997 ). Although our preliminary characterization of Yfrd3-1 had suggested that the frd3-1 mutation specifically affected NRE^DIT-mediated repression of gene expression, further study indicated that Yfrd3-1 had additional deficiencies, including a conditional slow-growth phenotype and minor defects in expression of other genes. To our knowledge, SPE3 has never before been identified through its effects on gene expression.

In contrast to the report by HAMASAKI-KATAGIRI et al. 1997 that an spe3 mutant is unable to grow on synthetic medium to which no spermidine has been added, we found that an spe3::HIS3 mutant was able to grow, albeit slowly, on such medium. This discrepancy could be due to the presence of trace amounts of spermidine in our synthetic medium, but not in that used by HAMASAKI-KATAGIRI et al. 1997 (BALASUNDARAM et al. 1991 ). Alternatively, it is possible that our spe3::HIS3 allele allowed synthesis of a truncated, but partially active, enzyme.

Role for spermidine in modulating gene expression:
Spermidine is the predominant polyamine in yeast with intracellular concentrations in the millimolar range (COHN et al. 1978 ). Extensive studies have shown that polyamines are essential for optimal growth in all cell types and implicate them as contributors to processes such as DNA replication, transcription, translation, protein phosphorylation, and resistance to elevated temperature and oxygen toxicity, among other things (BALASUNDARAM et al. 1993 ; BALASUNDARAM et al. 1996 ; for review see TABOR and TABOR 1984 , TABOR and TABOR 1985 , DAVIS et al. 1992B ). Nonetheless, the molecular role of spermidine in vivo remains to be defined. In vitro, polyamines have been shown to bind to DNA and RNA (IGARASHI et al. 1982 ), to condense DNA (MARX and REYNOLDS 1982 ), and to enhance the binding of some proteins to DNA and to inhibit the binding of others (PANAGIOTIDIS et al. 1995 ).

We have found that growth in minimal medium of yeast cells that cannot synthesize spermidine leads to defects in gene expression. The most dramatic defect that we observed was in NRE^DIT-mediated repression: mutation of SPE3 (FRD3) led to an ~8-fold reduction in repression of a CYC1-NRE^DIT-lacZ gene reporter (Figure 2A, Figure 6A). Additionally, we found that spe3 (frd3) mutants expressed a CYC1-lacZ reporter gene at a two to threefold higher level than did wild-type cells and were two- to threefold less efficient than were wild-type cells in mediating repression through the 2-Mcm1 operator. Both the defect in repression through NRE^DIT and the overexpression of the CYC1-lacZ gene were partially suppressed by the addition of spermidine to the growth medium. Thus, the elevated expression of the CYC1-NRE^DIT-lacZ reporter gene in spe3 (frd3) cells grown in minimal medium may be the combined effect of a defect in repression through NRE^DIT and a defect in modulating the activity of the CYC1 UAS. We note that in this study we have reported the efficiency of NRE^DIT-mediated repression relative to expression of the control CYC1-lacZ reporter gene in the same strain; thus, the fold-effects that we refer to reflect changes in NRE^DIT activity only. Our data, therefore, clearly indicate that the predominant effect of spermidine on restoring repression to the CYC1-NRE^DIT-lacZ reporter gene in an spe3 strain is through its effects on NRE^DIT.

Spermidine could act to modulate gene expression in various ways. Its effect could be indirect; spermidine-induced changes in processes such as translational fidelity (BALASUNDARAM et al. 1994 ) might lead to differential synthesis of regulators of transcription. Spermidine could modulate gene expression directly by affecting the binding of sequence-specific DNA-binding proteins to their cognate sites on DNA. Indeed, PANAGIOTIDIS et al. 1995 demonstrated that in vitro spermidine enhances the binding of several proteins to DNA and inhibits the binding of others. It is also possible that spermidine promotes an interaction between the Ssn6-Tup1 corepressor and the NRE^DIT-binding protein. Future identification of the NRE^DIT-binding protein(s) will allow us to test for these potential roles of spermidine in regulating assembly of a repression complex at NRE^DIT.

Spermidine, which has a polybasic character similar to that of histones, could also modulate gene expression by promoting localized changes in DNA structure. Indeed, in vitro, spermidine binds to DNA and promotes its compaction (MARX and REYNOLDS 1982 ). It is possible, therefore, that in vivo spermidine acts in conjunction with nucleosomes to reduce differentially the accessibility of regions of DNA to regulators of transcription and to the general transcription machinery. This effect could prevent hyperactivation of positively regulated genes (such as CYC1), as well as lead to more efficient repression of negatively regulated genes (such as DIT1). In this case, the absence of spermidine would lead to higher levels of gene expression by allowing the transcriptional machinery readier access to promoter regions. In support of this model, polyamines have been found to increase the stability of nucleosome core particles in vitro (MORGAN et al. 1987 ). Furthermore, chromatin from HeLa cells depleted of polyamines by treatment with inhibitors shows increased accessibility to DNase (SNYDER 1989 ). It has been suggested that regulation of polyamine binding to DNA could be achieved through acetylation and deacetylation of polyamines (MATTHEWS 1993 ) in a manner similar to the way histone acetylation and deacetylation regulate the association of nucleosomes with DNA (for review see PAZIN and KADONAGA 1997 ; WOLFE 1997 ; STRUHL 1998