Regulated Antisense Transcription Controls Expression of Cell-Type-Specific Genes in Yeast

Abstract

Transcriptome profiling studies have recently uncovered a large number of noncoding RNA transcripts (ncRNAs) in eukaryotic organisms, and there is growing interest in their role in the cell. For example, in haploid Saccharomyces cerevisiae cells, the expression of an overlapping antisense ncRNA, referred to here as RME2 (Regulator of Meiosis 2), prevents IME4 expression. In diploid cells, the a1-α2 complex represses the transcription of RME2, allowing IME4 to be induced during meiosis. In this study we show that antisense transcription across the IME4 promoter region does not block transcription factors from binding and is not required for repression. Mutational analyses found that sequences within the IME4 open reading frame (ORF) are required for the repression mediated by RME2 transcription. These results support a model where transcription of RME2 blocks the elongation of the full-length IME4 transcript but not its initiation. We have found that another antisense transcript, called RME3, represses ZIP2 in a cell-type-specific manner. These results suggest that regulated antisense transcription may be a widespread mechanism for the control of gene expression and may account for the roles of some of the previously uncharacterized ncRNAs in yeast.

INTRODUCTION

One of the main paradigms for the control of gene expression is that regulatory proteins bind to the promoter regions of genes to activate or repress transcription. However, it is now clear that noncoding RNAs (ncRNAs) also play important roles in gene regulation. For example, RNA interference (RNAi)-mediated regulation controls gene expression in Caenorhabditis elegans, Arabidopsis thaliana, humans, and many other organisms (16, 24). However, a large number of ncRNAs do not appear to be involved in RNAi-mediated regulation. For example, more than 900 ncRNAs are expressed in the yeast Saccharomyces cerevisiae (11, 15, 33, 40, 45). Several of these ncRNAs act to regulate gene expression in yeast (2, 5, 18, 26). However, S. cerevisiae lacks the enzymes Dicer and Argonaute, which are required for RNAi, and therefore, it must utilize different mechanisms for ncRNA-mediated regulation (12). In this paper we investigate how two antisense ncRNAs regulate the expression of genes required for meiosis in yeast.

Under starvation conditions, diploid yeast undergoes meiosis and sporulation to form four haploid spores. This process involves the expression of more than 500 genes that are highly regulated in a coordinated manner (7, 28). Entry into the meiotic pathway is controlled by the expression of IME1, the master initiator of meiosis (20, 29). There are two signals that regulate IME1 expression (Fig. 1A). One signal relates to the nutritional status of the cell, activating IME1 expression when the cell is starved of both nitrogen and a fermentable carbon source (14, 37). The second signal operates through cell-type-specific regulation, which allows the expression of meiotic genes only in a/α diploid cells. Cell-type-specific regulation is controlled by the a1-α2 repressor complex, which regulates IME1 expression through two different pathways. One pathway involves the haploid-cell-specific repressor Rme1, which binds to the promoter of IME1, preventing its expression (10, 30). In diploid cells, the a1-α2 complex binds to the RME1 promoter, repressing its transcription and thereby relieving the repression of IME1. The second form of cell-type-specific control is mediated through Ime4, which is required for full expression of IME1 (38). IME4 is repressed in haploid cells, and the a1-α2 repressor complex is required for its expression in diploid cells. Rme1 does not regulate IME4, so it has been hypothesized that a different haploid-cell-specific repressor regulates IME4 (38).

Fig. 1.

Model for the regulation of IME1. (A) Cell-type-specific regulation of IME1 is mediated in part by the a1-α2 complex, which prevents the expression of RME1, a repressor of IME1. The IME4 gene is also required for the full activation of IME1. IME4 expression requires the a1-α2 complex, which represses the haploid-cell-specific ncRNA RME2 in diploid cells. (B) Expression of the haploid-cell-specific antisense RME2 transcript represses IME4. In diploid cells, the a1-α2 complex binds downstream of IME4 and prevents RME2 transcription, allowing the expression of IME4 under sporulation-inducing conditions. “X” indicates that the transcript is not expressed.

To identify a1-α2 target sites, and possibly the factor regulating IME4 expression, in the yeast genome, we used an algorithm that combined a1-α2 binding site preference data with cell-type-specific microarray data (34). One of the strong a1-α2 binding sites identified in the search is downstream of the IME4 open reading frame (ORF). This site has an indirect role in controlling the expression of IME4 through the regulation of an antisense ncRNA (17). This ncRNA, which we refer to here as RME2 (Regulator of Meiosis 2), is expressed in haploid cells and blocks the expression of IME4 (Fig. 1B). In diploid cells, the a1-α2 complex represses RME2 expression, allowing IME4 to be induced under starvation conditions. This system appears to function only in a cis configuration, since RME2 is unable to repress an adjacent copy of IME4 (17).

We show in this paper that the expression of another antisense ncRNA regulates the meiosis-specific ZIP2 gene in a manner similar to the regulation of IME4. This result suggests that regulated expression of antisense ncRNAs may be a conserved mechanism of gene regulation in yeast. Interestingly, there are specific elements within the IME4 ORF that are required for repression mediated by RME2. Antisense expression does not inhibit transcription factors from binding to the IME4 and ZIP2 promoters and therefore may block transcription elongation of the coding genes.

MATERIALS AND METHODS

Plasmid and strain construction.

Plasmid pBG1 contains a 600-bp PCR-generated fragment, consisting of bp +1400 to +2000 from the IME4 translation initiation site, cloned into the TOPO TA vector (Invitrogen). Site-directed mutagenesis was used to change four base pairs in the a1-α2 binding site (wild-type [WT] sequence, GTGTATTTTTTTACATCA; mutant [Mu] sequence, GTcgATTTTTTTACggCA) to produce plasmid pBG7. Plasmid pBG113 contains a 2.9-kb PCR fragment, consisting of bp −450 to +400 flanking the IME4 ORF, cloned into the TOPO TA vector (Invitrogen). This plasmid was first digested with XbaI and HindIII, and IME4 was cloned into the same sites in pRS415 and pRS405 to generate pBG112 and pBG129, respectively (39). The HOP1-urs1 mutant (pBG157) and 225-675 flip (pBG166) plasmids were generated by gap repair of pBG112 (32). All other IME4 mutants were generated by site-directed mutagenesis of pBG129. The rme2-s1 mutation changes bp +447 relative to the start site of the IME4 ORF from an A to a T and is silent in terms of coding for the IME4 protein. The rme2-s2 mutant changes bp −23 and −24 relative to the start site of the IME4 ORF from GA to TT.

Plasmid pJM532 contains a 3.1-kb PCR-generated fragment, consisting of bp −450 to +400 flanking ZIP2 genomic DNA, cloned into the TOPO TA vector (Invitrogen). This plasmid was digested with SpeI and ApaI, and ZIP2 was cloned into the same sites in pRS405 to generate pJM533. Site-directed mutagenesis was used to change four base pairs in the downstream a1-α2 binding site (as indicated above for the IME4 site) to produce pJM535.

A list of the strains used in this study is given in Table 1. Strains YBG111 and YBG112 were constructed by transforming the W303 derivative strains LNY315 and LNY316 with a PCR fragment amplified from pFA6a-KanMX6 with the KanMX cassette and 50 bp of flanking homology so as to delete bp −450 to +400 flanking the IME4 ORF (43). These strains were mated to produce an ime4Δ/ime4Δ diploid strain, YBG115. IME4 deletion and poly(A) terminator mutants were integrated at the LEU2 locus by digesting the plasmids listed above with XcmI and transforming the linearized DNA into strain YBG111. Transformants were selected on a medium lacking leucine and were confirmed by PCR.

Table 1.

Yeast strains

Strain	Genotype	Source
LNY315	MATaade2-1trp1-1his3-11,15can1-100ura3-1leu2-3,112	L. Neigeborn
LNY316	MATα ade2-1trp1-1his3-11,15can1-100ura3-1leu2-3,112	L. Neigeborn
LNY392	MATaade2-1TRP1his3-11,15can1-100ura3-1leu2-3,112	L. Neigeborn
LNY433	MATα ADE2trp1-1his3-11,15can1-100ura3-1leu2-3,112	L. Neigeborn
YBG111	MATaade2-1trp1-1his3-11,15can1-100ura3-1leu2-3,112ime4Δ::kanMX4	This study
YBG112	MATα ade2-1trp1-1his3-11,15can1-100ura3-1leu2-3,112ime4Δ::kanMX4	This study
YBG115	MATa/MATα ade2-1/ade2-1trp1-1/trp1-1his3-11,15/his3-11,15can1-100/can1-100ura3-1/ura3-1leu2-3,112/leu2-3,112ime4Δ::kanMX4/ime4Δ::kanMX4	This study
YBG144	MATa/MATα ade2-1/ADEtrp1-1/TRPhis3-11,15/his3-11,15can1-100/can1-100ura3-1/ura3-1leu2-3,112/leu2-3,112	This study
JMY076	LNY392 with IME4–a1-α2::CaURA3	This study
JMY077	LNY433 with IME4–a1-α2::CaURA3	This study
JMY081	LNY392 with IME4–a1-α2mut	This study
JMY082	LNY433 with IME4–a1-α2mut	This study
JMY084	YBG144 with IME4–a1-α2mut/IME4-a1-α2mut	This study
JMY104	LNY392 with zip2::kanMX4	This study
JMY105	LNY433 with zip2::kanMX4	This study
JMY108	LNY392 with zip2::kanMX4leu2::ZIP2	This study
JMY109	LNY433 with zip2::kanMX4leu2::ZIP2	This study
JMY110	YBG144 with zip2::kanMX4/zip2::kanMX4leu2::ZIP2/leu2::ZIP2	This study
JMY111	LNY392 with zip2::kanMX4leu2::ZIP2-a1-α2mut	This study
JMY112	LNY433 with zip2::kanMX4leu2::ZIP2-a1-α2mut	This study
JMY113	YBG144 with zip2::kanMX4leu2::ZIP2-a1-α2mut	This study
YBG145	YBG111 with leu2::ime4Δ3′	This study
YBG147	YBG111 with leu2::IME4	This study
YBG149	YBG115 with leu2::IME4/leu2::IME4	This study
YBG150	LNY315 with leu2::ime4Δ3′	This study
YBG158	YBG144 with zip2::kanMX4/zip2::kanMX4leu2::ZIP2-a1-α2mut/leu2::ZIP2	This study
YBG159	YBG111 with leu2::rme2-s1	This study
YBG160	YBG111 with leu2::ime4Δ1-900	This study
YBG161	YBG111 with leu2::ime4Δ1-450	This study
YBG162	YBG111 with leu2::ime4Δ451-900	This study
YBG183	YBG111 with leu2::ime4Δ1-224	This study
YBG184	YBG111 with leu2::ime4Δ676-900	This study
YBG202	YBG111 with leu2::rme2-s2	This study

Strains JMY076 and JMY077 were constructed by using a Candida albicans URA3 cassette from pGEM-CaURA3 to delete the a1-α2 binding site. These strains were transformed with a PCR product from pBG7 in the presence of 5-fluoroorotic acid (5-FOA) to recombine the mutant binding site at the wild-type locus, generating strains JMY081 and JMY082. JMY081 and JMY082 were mated to generate JMY084, a homozygous diploid with the a1-α2 mutation downstream of IME4.

The native ZIP2 gene was deleted by transformation with a KanMX PCR fragment amplified from the Yeast Deletion Strain collection (Research Genetics) in strains LNY392 and LNY433 to generate JMY104 and JMY105. JMY104 and JMY105 were transformed with XcmI-linearized pJM533 to generate JMY108 and JMY109, which were mated to produce JMY110. JMY104 and JMY105 were also transformed with XcmI-linearized pJM535 to generate JMY111 and JMY112, which were mated to produce JMY113. All genomic integrations were confirmed by PCR.

RT-PCR assays.

Expression of the RNA transcripts was assayed by reverse transcriptase PCR (RT-PCR), which was performed on two to four replicate samples. Single colonies of yeast strains (from separate transformations for plasmid-bearing strains) were grown under rich nutrient (yeast extract-peptone-dextrose [YEPD] or synthetic dextrose [SD] medium) or sporulation-inducing (sporulation medium [SPM]) conditions (3 h in SPM for IME4 assays; 5 h for ZIP2 assays), and total RNA was extracted by hot acid-phenol extraction, as described previously (1). Normalized RNA samples were treated with Turbo DNA-Free DNase (Ambion), and the DNase-treated RNA was amplified by PCR with the IME4/RME2 or ZIP2/RME3 primer set to verify the absence of contaminating DNA. cDNAs of the IME4, RME2, URA3, ZIP2, RME3, HSP26, YFL012W, and ACT1 genes were synthesized for each RNA sample with sense and antisense specific primers by using Omniscript RT (Qiagen). Different cDNA sample concentrations (1 to 4 μl) were assayed to verify that the reaction was in the linear range. PCRs using different concentrations of cDNA as a template were amplified in 50-μl reaction mixtures containing 10 pmol of each amplicon primer set, 1× AmpliTaq Taq Buffer II, and 2.5 U of AmpliTaq Taq polymerase (Applied Biosystems). The amplifications were carried out for 30 cycles of 94°C for 30 s, 52°C for 1 min, and 72°C for 30 s. Samples were run on 1.4% agarose-Tris-acetate-EDTA (TAE) gels, and images were photographed with a Fluorochem 8800 camera.

ChIP assays.

Chromatin immunoprecipitation (ChIP) assays were performed using a modified version of protocols described elsewhere (27, 34). Cultures of strains LNY392 and YBG144 or strains YBG111 and YBG115 carrying the HOP1pr::IME4 constructs indicated in Fig. 5 were grown to mid-log phase (optical density at 600 nm [OD₆₀₀], 0.500) in the appropriate medium (YEPD, SD-Leu, or SPM); 50 ml of the cultures was fixed with a final concentration of 1% formaldehyde for 15 min at 22°C, washed with 1 ml Tris-buffered saline (TBS), and frozen at −80°C for a minimum of 12 h. Cell pellets were suspended with 400 μl lysis buffer (50 mM HEPES [pH 7.4], 150 mM NaCl, 1 mM EDTA, 1% Triton X-100, 0.1% sodium deoxycholate) plus 1 mM phenylmethylsulfonyl fluoride (PMSF), 1× protease inhibitor cocktail from Roche (catalog no. 1873580), and 50 μl Sigma protease inhibitor cocktail (catalog no. P8215). To this, 200 μl of glass beads was added, and cells were lysed by vortexing at full speed for 40 min at 4°C. The lysate was then centrifuged for 5 min at 12,000 × g and 4°C. The supernatant was transferred to a new tube, and the beads were washed with 500 μl of FA lysis buffer; the supernatant from this wash was added to the original supernatant fraction. DNA was sonicated at 30% output for 6 5-s cycles to give an average chromatin fragment size of 500 bp. Sonicated lysates were centrifuged at 12,000 × g for 5 min, and 50 μl of the sonicated DNA was reserved as a total-chromatin (TC) sample. The remaining DNA was precleared by the addition of 25 μl of protein G agarose beads and was nutated for 1 h at 4°C, and the lysates were cleared by centrifugation at 12,000 × g for 5 min at 4°C. To immunoprecipitate TATA-binding protein (TBP)- or Abf1-bound DNA, 10 μl or 1 μl of the anti-TBP polyclonal antibody yN-20 (sc-26141; Santa Cruz) or the anti-Abf-1 polyclonal antibody yC-20 (sc-6679; Santa Cruz) was added, and the mixture was nutated for 16 h at 4°C. Centrifugation with protein G beads, washes, DNA elution, cross-linking reversal, and proteinase treatment were all performed as described previously (34, 36). Frozen TC samples were brought up to 500 μl in volume with elution buffer. A Qiagen PCR purification kit was used to purify the amplified DNA. Multiple quantities of DNA input were used for the PCR. Typically, 2 μl of TC samples (a 500-fold dilution and a 50-fold dilution) or immunoprecipitated (IP) samples (1 and 3 μl) was used in 50-μl reaction mixtures for 30 cycles. The PCR products were run on 1.5% agarose gels. Samples were quantified using ImageJ software and normalization to the ACT1 signal.

Fig. 5.

The HOP1-urs1 promoter is regulated by antisense transcription in a manner similar to the regulation of the native IME4 promoter. Shown are results of RT-PCR assays of IME4 and RME2 from haploid (1n) and diploid (2n) cells after 3 h in sporulation-inducing medium. IME4 (sense) transcription is driven either by the native IME4 promoter (LNY392 [lane 1] and YBG144 [lane 2]) or by a HOP1 promoter containing a mutation in the URS1 site (YBG111/pBG157 [lane 3] and YBG115/pBG157 [lane 4]). The expression of the constitutive HOP1-urs1 mutant promoter, combined with a mutant a1-α2 binding site downstream of the ORF, was assayed under the same conditions in haploid (YBG111/pBG167) (lane 5) and diploid (YBG115/pBG167) (lane 6) cells. Assays and controls are described in the legend to Fig. 2.

RESULTS

IME4 is repressed by cell-type-specific antisense transcription.

Chromatin immunoprecipitation (ChIP) assays had previously shown that the a1-α2 repressor complex is bound to a site downstream of IME4 (34). In agreement with previous reports, we found that a mutation in this site allows expression of the IME4 antisense transcript RME2 and prevents IME4 expression in diploid cells (17; also data not shown).

To determine whether RME2 is required for the repression of IME4, we constructed a mutation, ime4Δ3′, that deletes the RME2 promoter. Haploid cells with this mutation failed to express RME2, and IME4 was derepressed (Fig. 2B, lane 3). This suggests that RME2 expression is required for the repression of IME4.

Fig. 2.

(A) Cartoon illustrating the wild-type and ime4Δ3′ constructs. The ime4Δ3′ construct contains a deletion of the RME2 promoter region (+1809 to +2209), as indicated by the dashed line. “X” indicates that the transcript is not expressed from that copy of the IME4 locus. (B) Deletion of the RME2 promoter region allows the expression of IME4 in haploid cells, and RME2 expression fails to repress the ime4Δ3′ mutation in trans. RT-PCR assays of the IME4 and RME2 transcripts were performed on ime4Δ (YBG111) (lane 1), WT (YBG147) (lane 2), ime4Δ3′ (YBG145) (lane 3), and WT + ime4Δ3′ (YBG150) (lane 4) haploid strains under sporulation-inducing conditions. The −RT row shows PCR amplifications of DNase-treated RNA with the IME4 primer set to control for the contamination of genomic DNA. RT-PCR of ACT1 expression was used as a loading control. Genomic DNA (gDNA) (lane 5) was included as a control for PCR amplification.

Expression of RME2 from the native locus did not repress the transcription of IME4 from the ime4Δ3′ mutant integrated at the distant locus (Fig. 2B, lane 4). This indicates that RME2 transcription is able to repress IME4 only in a cis-acting manner. Our results are consistent with previous work that used a different mutation in the a1-α2 site and a different assay for expression of the sense and antisense genes (17).

Antisense transcription is a conserved model of gene regulation in yeast.

Since IME4 is regulated by transcription of an ncRNA in the antisense direction, we wanted to determine if other genes are regulated by a similar mechanism. We therefore reevaluated our previous data identifying a1-α2 sites in the yeast genome by relaxing the sequence requirements for a1-α2 sites and specifically searching for sites that are downstream of cell-type-specific genes (34). We then used the Yeast Transcriptome Database to search for the presence of antisense transcription of the gene upstream of the a1-α2 site (11). In addition to IME4, this search identified the HSP26, YFL012W, and ZIP2 genes as potential targets for antisense-mediated regulation. RT-PCR assays for the expression of sense and antisense transcripts from these genes showed that HSP26 and YFL012W are not regulated in a cell-type-specific manner by antisense transcription (data not shown). However, analysis of ZIP2, which was previously identified as a meiosis-specific component of the synaptonemal complex, suggested that the sense and antisense transcripts are regulated in a cell-type-specific manner (Fig. 3 A) (8). In agreement with gene expression profiling experiments during meiosis, the ZIP2 (sense) transcript was expressed only in diploid cells under sporulation conditions (Fig. 3B, lane 4) (7). In contrast, haploid cells under the same conditions expressed an ncRNA antisense to ZIP2, which we refer to as RME3 (Regulator of Meiosis 3) (Fig. 3B, lane 3).

Fig. 3.

Expression analyses of the ZIP2 gene and the antisense transcript RME3 show cell-type-specific regulation. (A) Schematic of the ZIP2 and RME3 transcripts and the relative position of the a1-α2 binding site downstream of the meiosis-specific gene ZIP2 and upstream of the non-cell-type-specific gene PDE1. (B) RT-PCR assays of ZIP2 and RME3 from haploid (LNY392) (lanes 1 and 3) and diploid (YBG144) (lanes 2 and 4) cells grown in YEPD (Veg) (lanes 1 and 2) or sporulation-inducing (Spo) (lanes 3 and 4) medium. (C) RT-PCR assays of ZIP2 and RME3 from haploid and diploid cells grown under sporulation conditions for 5 h. Either a wild-type (WT) haploid (JMY108) (lane 1) or diploid (JMY110) (lane 3) strain or an a1-α2 binding site mutant (Mu) haploid (JMY108) (lane 2) or diploid (JMY110) (lane 4) strain was used. (D) RT-PCR assays of ZIP2 and RME3 from wild-type (JMY110) (lanes 1 and 2) or heterozygous (YBG158) (lanes 3 and 4) diploid cells grown in Veg or Spo medium. Assays and controls are described in the legend to Fig. 2.

The expression pattern of ZIP2 and RME3 is similar to that observed for IME4 and RME2. We were therefore interested in determining whether the RME3 transcript was responsible for antisense-mediated regulated expression of ZIP2. Mutation of four bases in the a1-α2 site downstream of ZIP2 caused derepression of RME3 in diploid cells (Fig. 3C, lane 4). Expression of the ZIP2 transcript was inhibited in the mutant diploid strain (Fig. 3C, lane 4). This shows that, like IME4, ZIP2 is repressed by antisense transcription.

To test whether RME3 represses ZIP2 transcription in a cis- or a trans-acting manner, we constructed a diploid strain heterozygous for the ZIP2 locus. One of the ZIP2 alleles in this strain is wild type, while the other contains the mutation of four base pairs in the a1-α2 site that allows RME3 expression in diploid cells. If ZIP2 were repressed by RME3 in trans, we would have expected to see a lower level of ZIP2 in the heterozygote than in a wild-type homozygous strain, a mechanism similar to RNAi (9, 16). However, the level of ZIP2 expression was the same in the mutant heterozygous and wild-type homozygous diploid strains, indicating that expression of RME3 in trans did not inhibit ZIP2 transcription (Fig. 3D, lane 2 versus lane 4). This indicates that RME3 regulates ZIP2 in a cis-dependent configuration, a pattern similar to that of the IME4/RME2 regulatory system.

Changing the termination site of RME2 affects its ability to regulate IME4.

The observation that both IME4 and ZIP2 are repressed by antisense transcription in haploid cells suggests that this may be a common mechanism of gene regulation. We wanted to determine how the expression of an antisense transcript in cis prevents the expression of the sense transcript. One of the most economical models for this form of regulation is that antisense transcription through the sense promoter prevents transcription factor binding and activation. This model is similar to the mechanisms proposed for the SER3 and ADH1 genes, which are regulated by the upstream ncRNAs SRG1 and ZRR1, respectively (3, 25, 26). Transcription of the upstream ncRNAs through the promoter of the coding gene prevents the binding of the transcriptional activators required for expression. If IME4 is regulated by a mechanism similar to those for SER3 and ADH1, then premature termination of the RME2 transcript would allow the expression of IME4 in haploid cells. To test this model, two mutations, rme2-s1 and rme2-s2, that truncate the RME2 transcript by the introduction of a eukaryotic poly(A) signal sequence, AAUAAA, were constructed. The rme2-s1 mutation introduced a poly(A) site in the direction of RME2 at 447 bp downstream of the IME4 ATG; this change is silent with respect to the IME4 coding sequence. The rme2-s2 mutation introduced the same poly(A) site 23 bp upstream of the IME4 ATG; it does not overlap with the presumptive IME4 TATA box or alter the spacing of the promoter. Because transcription extends past poly(A) sites by roughly 100 bp, the rme2-s1 mutation shortens RME2 by 700 to 800 bp, and rme2-s2 shortens RME2 by 300 to 400 bp (Fig. 4 A) (4). A similar approach was used to truncate the Kcnq1ot1 transcript in mouse and to investigate its role in antisense-mediated regulation (19). To assay for premature termination of RME2, primer sets targeting regions within the IME4 promoter (Pr), the 5′ end of IME4 (5′), and the middle of the ORF (Mid) were used to differentially detect the wild-type, rme2-s1, and rme2-s2 transcripts (Fig. 4A). In wild-type cells, the RME2 transcript was detected by all three primer sets (Fig. 4B, lane 1). In contrast, the rme2-s1 transcript was detected only by the Mid primer set, indicating that the transcript was prematurely terminated before the 5′ end of the IME4 ORF. Similarly, rme2-s2 was detected only by the Mid and 5′ primer sets, not by the Pr primer set, indicating that transcription did not extend through the entire IME4 promoter and upstream region, as in the wild-type haploid strain (Fig. 4B, RME2, lane 3). Despite this change, the rme2-s2 mutation did not appear to affect the repression of IME4 (Fig. 4B, IME4, lane 3). In contrast, the truncated rme2-s1 transcript, which does not extend into the promoter or the 5′ end of IME4, was unable to repress the expression of IME4 (Fig. 4B, IME4, lane 2). In a/α diploid cells, the rme2-s1 and rme2-s2 mutants expressed IME4 at wild-type levels, indicating that they had no effect on TATA-binding protein (TBP) or polymerase binding at the IME4 promoter (data not shown). These results suggest that transcription of RME2 through the 5′ end of the IME4 ORF is essential for repression, but extension of the antisense transcript through the IME4 promoter region is not required.

Fig. 4.

Alteration of the polyadenylation and termination of the RME2 transcript alters its ability to repress IME4 in haploid cells. (A) Illustration of the three PCR amplicons (Pr, IME4 promoter; 5′, 5′ end of the sense ORF; Mid, middle of the ORF) used to detect both IME4 and RME2 transcripts upstream and downstream of the termination sites introduced in rme2-s1 and rme2-s2. Primers downstream (relative to RME2) of each amplicon were used to generate strand-specific cDNA. (B) RT-PCR assays of wild type (YBG147) (lane 1), rme2-s1 (YBG159) (lane 2), and rme2-s2 (YBG202) (lane 3) haploid strains. Assays and controls are described in the legend to Fig. 2. The −RT control was performed for all assay primer pairs.

RME2 transcription can repress a heterologous promoter.

Previous research showed that high-level expression of IME4 from the GAL1 promoter was able to override repression by RME2 in haploid cells (17). It is possible that the GAL1 promoter was not repressed by RME2 transcription due to differences in promoter specificity. For example, antisense transcription may disrupt the binding of specific transcription factors to the IME4 promoter. In contrast, the transcriptional activator of GAL1, the Gal4 protein and its cofactors, may be insensitive to this form of regulation. To test this model, we assayed for the ability of RME2 to repress a heterologous promoter with expression activity similar to that of the IME4 promoter. We constructed a strain in which the IME4 promoter was replaced with a derivative of the HOP1 promoter, HOP1-urs1, that is constitutively active in both haploid and diploid cells (42). Diploid cells grown under sporulation conditions expressed similar levels of IME4 from either the IME4 or the HOP1-urs1 promoter (Fig. 5, lanes 2 and 4). In haploid cells, both of the promoters were repressed by RME2 transcription (Fig. 5, lanes 1 and 3). To further test this result, we mutated the a1-α2 site downstream of IME4 in the context of the HOP1-urs1 promoter to allow the expression of RME2 in haploid and diploid cell types (Fig. 5, lanes 5 and 6). For the mutant, the HOP1-urs1 promoter was repressed by RME2 in diploid cells (Fig. 5, lane 6). This is consistent with previous work, where a similar a1-α2 site mutant was shown to permit RME2 expression and to prevent meiosis in diploid cells (17). These results indicate that RME2 transcription is able to repress a heterologous promoter with similar activity and that there are unlikely to be specific elements or factors bound at the IME4 promoter that make it sensitive to antisense transcription.

TBP binding at the sense and antisense promoters of IME4 and ZIP2.

The observation that the rme2-s2 mutant is able to repress the expression of IME4 suggests that antisense transcription across the promoter is not required for repression. Although RME3 is able to repress ZIP2 expression, transcriptome profiling experiments suggest that the RME3 transcript does not extend through the entire ORF and into the ZIP2 promoter (11). Even though rme2-s2 and RME3 do not extend through the promoters, they could still work through a mechanism that blocks transcription factors from binding and activating transcription. To test this model, we used ChIP assays to monitor transcription factor binding at both the sense and antisense promoter regions. The recruitment of TBP can be used to determine whether an ncRNA disrupts transcription factor binding at a promoter (25). TBP binding to the IME4 promoter (5′), the middle of the ORF (Mid), and the RME2 promoter (3′) was assayed in haploid and diploid cells. The RME2 promoter was bound by TBP in haploid cells with 16.9-fold higher affinity than in diploid cells (Fig. 6 A, 3′ IME4, lane 3 versus lane 4). As expected, TBP was bound to the IME4 promoter in diploid cells (Fig. 6A, 5′ IME4, lane 4). Surprisingly, despite the repression of IME4 by RME2, TBP was also bound to the IME4 promoter with almost equal affinity (0.90-fold difference) in haploid and diploid cells (Fig. 6A, 5′ IME4, lanes 3 and 4). In addition, Abf1, the HOP1 activator protein, and TBP were bound to the HOP1-urs1 IME4 promoter fusion even when IME4 was repressed by RME2 in haploid cells (data not shown) (13). These results indicate that RME2 transcription does not disrupt the occupancy of the IME4 promoter by transcription factors.

Fig. 6.

Antisense transcription does not disrupt TBP binding at the IME4 or ZIP2 promoter. (A) ChIP assays for TBP bound at both the sense (5′) and antisense (3′) promoters in haploid WT (LNY392) (lanes 1 and 3) and diploid WT (YBG144) (lanes 2 and 4) cells. Assays were performed after 3 h in SPM. The middle regions (Mid-) of the IME4 and ACT1 ORFs, which are not precipitated with an anti-TBP antibody, were included as negative controls, and the constitutive ACT1 promoter was included as a positive control for antibody binding. TC, total-chromatin sample (lanes 1 and 2); IP, DNA immunoprecipitated with the anti-TBP antibody yN-20 (lanes 3 and 4). The fold change in TBP binding in haploid versus diploid cells is given on the right. (B) Amplification of the ZIP2 and RME3 promoters from haploid (LNY392) (lanes 1 and 3) and diploid (YBG144) (lanes 2 and 4) cells. As in the IME4 ChIP assay for which results are shown in panel A, TBP binding was assayed with the 5′, Mid, and 3′ amplicons by using the concentrations of purified TC and IP DNA given in Materials and Methods. The native HOP1 promoter was tested as a meiosis-specific, cell-type-dependent promoter. ChIP assays were also performed on the ZIP2 and RME3 promoters in vegetative cultures of these haploid and diploid cells in order to confirm that these regions are not constitutively bound by TBP (Veg. 5′ ZIP2 and Veg. 3′ ZIP2, lanes 3 and 4).

We also examined the binding of TBP at the ZIP2 locus in order to determine if the mechanism of RME3-mediated repression is similar to that of RME2-mediated repression. Like the RME2 promoter, the RME3 promoter was bound by TBP in haploid cells with 16.1-fold higher affinity than in diploid cells (Fig. 6B, 3′ ZIP2, lane 3 versus lane 4). As with IME4, TBP bound to the ZIP2 promoter with roughly the same affinity in haploid and diploid cells (Fig. 6B, 5′ ZIP2, lanes 3 and 4). The wild-type HOP1 gene, which is activated at a time point in meiosis similar to that for ZIP2, was bound by TBP only in diploid cells (Fig. 6B, 5′ HOP1, lane 3 versus lane 4). This shows that, unlike the IME4 and ZIP2 promoters, the meiosis-specific HOP1 promoter is not bound by TBP when it is repressed according to cell type. These results suggest that RME2 and RME3 do not repress IME4 and ZIP2 through mechanisms that interfere with factors binding to their promoters.

A specific region within the IME4 ORF is required for antisense-mediated repression.

The observation that early termination of the rme2-s1 transcript blocked repression suggests that the IME4 ORF sequence may play a role in regulation by RME2. To determine whether regions within the IME4 ORF are required for antisense-mediated repression, we performed an internal deletion analysis of the gene. Constructs with specific deletions of the IME4 ORF were assayed for expression of the IME4 and RME2 transcripts in haploid cells. The ime4Δ1-900 deletion caused derepression of the IME4 transcript in haploid cells (Fig. 7, lane 3 versus lane 1). In contrast, deletion of bp 901 to 1800 had no effect on IME4 repression (data not shown). It was possible that the loss of IME4 transcriptional repression in the ime4Δ1-900 mutant was due to premature termination of the RME2 transcript. However, RT-PCR assays confirmed that RME2 was expressed and that the transcript extended across the IME4 promoter region (Fig. 7, lane 3). Taken together, these results suggest that the first 900 bp of the IME4 ORF are required for antisense-mediated repression.

Fig. 7.

Deletion analysis of IME4 reveals the region within the ORF that is required for antisense-mediated regulation. (Top) Cartoon showing the locations of the deletions and whether deletion of the hatched region leads to the loss of IME4 repression (Rep.). +, repression; −, loss of repression. (Bottom) RT-PCR assays of IME4 and RME2 expression from wild-type haploid (lane 1) and diploid (lane 2) cells and from haploid cells with the indicted base pairs of the IME4 ORF deleted (lanes 3 to 7), grown in sporulation-inducing medium. IME4 expression was monitored using an amplicon within the ORF, and RME2 expression was monitored using an amplicon in the IME4 promoter region. gDNA, genomic DNA. Deletion strains are listed in Table 1.

To further define the region required for the repression of IME4 by RME2, a series of smaller deletions were made within the first 900 bp of the IME4 ORF. The ime4Δ1-224 and ime4Δ676-900 deletions had no effect on the repression of IME4 in haploid cells (Fig. 7, lanes 6 and 7). In contrast, the ime4Δ1-450 and ime4Δ451-900 deletions caused derepression of IME4 (Fig. 7, lanes 4 and 5). This indicates that a DNA element within bp 225 to 675 is required for proper antisense-mediated repression.

The orientation of a portion of IME4 is involved in strand-specific expression.

Deletion analysis of the IME4 ORF showed that a region from bp 225 to 675 is required for RME2-mediated repression of IME4 in haploid cells. It is possible that transcription of this region in the antisense direction creates a chromatin structure that prevents effective extension of the full-length sense transcript. If this occurs, then there may be an orientation-specific requirement for this DNA element. To test for this, the segment of DNA from bp 225 to 675 was flipped to the opposite orientation within the context of the IME4 ORF (Fig. 8, top). This mutation had no effect on the expression of IME4 in diploid cells (Fig. 8, lane 3 versus lane 4), showing that it is silent with respect to the expression and stability of the IME4 transcript. However, in haploid cells, this mutation caused derepression of the IME4 transcript (Fig. 8, lane 1 versus lane 2). This result suggests that transcription of RME2 across this element may set up orientation-specific termination of IME4.

Fig. 8.

Reversing the orientation of the region of IME4 from bp 225 to bp 675 prevents proper regulation by RME2. Shown are results of RT-PCR assays of IME4 and RME2 from haploid (YBG111) (lanes 1 and 2) or diploid (YBG115) (lanes 3 and 4) cells carrying IME4 on a plasmid with bp 225 to 675 in either the wild-type (lanes 1 and 3) or the “flipped” (antiparallel relative to the WT) orientation (lanes 2 and 4). gDNA, genomic DNA.

DISCUSSION

Research on eukaryotic transcriptomes has uncovered the presence of large numbers of ncRNAs that are expressed from either intergenic regions or regions that overlap with coding genes in either the sense or the antisense direction (11, 21, 31, 35, 40). In higher eukaryotes, many of these ncRNAs regulate gene expression through trans-acting micro-RNA and small interfering RNA (siRNA) pathways that are dependent on Dicer and Argonaute (16, 24). Since the yeast S. cerevisiae lacks the genes coding for these proteins, it is unable to conduct RNAi-mediated regulation through mechanisms similar to those of higher eukaryotes (12). However, work by several labs has shown that some of the >900 ncRNAs in yeast do have regulatory functions. For example, trans-acting antisense regulatory ncRNAs function to silence the transcription and transposition of the Ty1 retrotransposon and the expression of PHO84 (2, 6). Expression of other ncRNAs, such as the SRG1 and ZRR1 ncRNAs, which are transcribed from upstream of the SER3 and ADH1 promoters, respectively, repress the expression of the coding genes in a cis-dependent manner (3, 25). Expression of an antisense ncRNA transcript, RME2, prevents IME4 expression in haploid cells (17) (Fig. 2B). We have now identified another cis-acting antisense ncRNA, RME3, that functions to repress the transcription of the meiosis-specific ZIP2 gene in haploid cells. This suggests that overlapping sense/antisense transcripts may be a common form of regulation in S. cerevisiae.

RME2 and RME3 appear to function strictly in cis; a single extra trans copy of the noncoding gene, expressed from a distant or adjacent locus, failed to repress sense transcription (Fig. 2B and 3D) (17). The SRG1 and ZRR1 transcripts are similarly nonfunctional when supplied in trans (3, 25). This contrasts with PHO84, where a second copy has a trans-acting effect on the native gene (6). Taken together, these cases suggest that while ncRNA-mediated repression may be prevalent in yeast, different sense-antisense gene pairs are regulated through at least two distinct mechanisms.

Previous work on the IME4 gene showed that the use of a GAL1 promoter fusion to express IME4 at a high level could overcome repression by RME2 in haploid cell types (17). The induced GAL1 promoter is therefore insensitive to antisense-mediated repression. One model for this finding is that there is a specific level of sense transcription that can be repressed by antisense transcription. In this model, the high level of IME4 expressed from the GAL1 promoter may not be repressed by antisense transcription from the weaker RME2 promoter. Another possible explanation for this result is that only specific promoters, such as IME4 and ZIP2, are sensitive to disruption by antisense transcription. However, a derivative of the HOP1 promoter is repressed by RME2 in a manner similar to that of the wild-type IME4 promoter. Therefore, the specificity of the promoter does not appear to be an integral part of the mechanism for repression by RME2, but the activity level of the repressed promoter may play a role.

One hypothesis for the mechanism of cis-acting antisense repression is that the transcription of ncRNA somehow acts to block the binding of the transcription factors required for the expression of the coding gene. This would be similar to the mechanism of regulation proposed for the SRG1 and ZRR1 ncRNAs, which are transcribed from upstream of the SER3 and ADH1 promoters, respectively, and repress the expression of the coding genes by inhibiting the binding of transcription factors to the downstream promoters (3, 25). Repression of the upstream ncRNA genes allows the factors to bind the downstream promoter to activate transcription of the coding genes. It is possible that RME2 regulates IME4 through a similar mechanism. However, expression of the rme2-s2 transcript, which terminates before reaching the IME4 promoter, is able to repress IME4. This shows that transcription through the IME4 promoter is not required for repression. Antisense transcription through the ZIP2 promoter also is not likely required for repression, because the RME3 transcript appears to terminate naturally in the ZIP2 ORF (11, 31). These results suggest that ncRNA-mediated repression of IME4 and ZIP2 does not occur through a mechanism of promoter interference such as that observed for SER3 and ADH1. Our results are similar to those of analysis of the Mus musculus gene Kcnq1 (Lit1), which is repressed by the expression of an internal antisense RNA, Kcnq1ot1 (41). When the Kcnq1ot1 transcript is shortened by the insertion of a premature polyadenylation site, the repression of Kcnq1 is reduced (19). One difference between the mouse and yeast genes that are regulated by antisense transcription is that large regions of the Kcnq1 locus are transcriptionally inactivated through chromatin modification. In contrast, transcriptome profiling experiments show that genes adjacent to IME4 and ZIP2 (HOS2/COX13 and RMR1/PDE1, respectively) are not corepressed when RME2 and RME3 are expressed (11). Further evidence against the promoter interference model was obtained in the ChIP assays for the presence of TBP at the IME4 and ZIP2 promoters (Fig. 6). If antisense transcription regulated through a mechanism of promoter interference, we would have expected to find TBP bound to the IME4 and ZIP2 promoters only in diploid cells, when RME2 and RME3 are repressed. However, we found that TBP remained bound to these promoters in both haploid and diploid cells. This result suggests that repression by RME2 and RME3 does not work through interference with transcription factors binding at the sense promoters. Our results are similar to those for regulation by the trans-acting PHO84 antisense transcript, which does not interfere with the binding of TBP at the sense promoter (5, 25). This further highlights the fact that the mechanisms of ncRNA-mediated regulation are different for different gene pairs.

A 450-bp region (bp 225 to 675) within IME4 is essential for RME2-mediated repression. It is possible that this region contains target sites for protein complexes that play a role in blocking the extension of the full-length IME4 transcript. RME2 transcription may expose these sites, allowing chromatin remodeling and/or modification that prevents full-length IME4 expression. It has been shown that directional chromatin remodeling represses antisense transcription from cryptic sites (44). Therefore, it is possible that deletion of this region of IME4 removes a target site for the required chromatin-remodeling/modifying enzymes, resulting in the expression of both the sense and antisense transcripts. It has been shown previously that expression of the internal cis antisense ncRNA Kcnq1ot1 alters the surrounding chromatin state to inhibit sense expression of the mouse gene Kcnq1 (41). RME2 and RME3 may remodel chromatin across the IME4 and ZIP2 loci to prevent full-length transcription of the coding genes.

Another possibility for the mechanism of antisense regulation by RME2 is that the region (bp 225 to 675) within IME4 that is essential for its repression is particularly sensitive to allowing transcription in one direction to disrupt transcription in the opposite direction. When this region of the IME4 ORF was deleted, both sense and antisense transcripts were expressed simultaneously (Fig. 7). This result is corroborated by the effect of the rme2-s1 mutation on regulation (Fig. 4). The truncated rme2-s1 transcript does not cross the entire bp 225-to-675 region of the ORF, and this truncation results in full-length transcription of IME4. This DNA sequence may therefore be transcribed only in a single direction at a given time. In contrast, other sequences permit bidirectional transcription. This appears to be the case for the nested antisense gene pair YGR031W and NAG1, two overlapping genes that appear to be transcribed simultaneously in vegetative medium (23).

We also observed that there is an orientation-specific requirement of IME4 bp 225 to 675 for antisense-mediated repression. Recent work has shown that abortive transcription upstream of IMD2 blocks the start site of the coding gene (18, 22). When a region of IMD2 required for the termination of these short transcripts is reversed in orientation, termination does not occur, and the IMD2 gene is no longer repressed. These experiments provide evidence that specific DNA architecture can play a role in gene regulation. Although RME2 represses the expression of the full-length IME4 transcript, short, incomplete IME4 transcripts have been detected (17; also data not shown). This suggests that the transcription of RME2 across this region of DNA causes premature termination of IME4.

There is no readily apparent homology between the bp 225-to-675 region of IME4 and the region of ZIP2 overlapped by RME3. It is therefore possible that these two gene pairs use different mechanisms for antisense-mediated repression. Alternatively, it is possible that degenerate sequences within each gene pair are required for transcriptional repression. If there are similarities in the mechanism of regulation between the IME4/RME2 and ZIP2/RME3 gene pairs, then it is possible that other overlapping antisense transcripts may also use a similar mechanism to regulate one another. Transcriptome profiling experiments have identified more than 350 ncRNAs expressed in an antisense orientation to coding genes (31). It is possible that some of these antisense ncRNAs regulate coding genes in a manner similar to that of RME2 and RME3. These gene pairs may explain the role of some of the >900 ncRNAs in yeast.