|Home | About | Journals | Submit | Contact Us | Français|
The cell fate decision leading to gametogenesis is essential for sexual reproduction. In S. cerevisiae, only diploid MATa/α but not haploid MATa or MATα cells undergo gametogenesis, known as sporulation. We find that transcription of two long non-coding RNAs (lncRNAs) mediates mating type control of sporulation. In MATa or MATα haploids expression of IME1, the central inducer of gametogenesis, is inhibited in cis by transcription of the lncRNA IRT1, located in the IME1 promoter. IRT1 transcription recruits the Set2 histone methyltransferase and the Set3 histone deacetylase complex to establish repressive chromatin at the IME1 promoter. Inhibiting expression of IRT1 and an antisense transcript that antagonizes the expression of the meiotic regulator IME4, allows cells expressing the haploid mating-type to sporulate with kinetics that are indistinguishable from that of MATa/α diploids. Conversely, expression of the two lncRNAs abolishes sporulation in MATa/α diploids. Thus, transcription of two lncRNAs governs mating type control of gametogenesis in yeast.
Gametogenesis, the process of gamete formation, is central to sexual reproduction. In multicellular organisms little is known about the molecular mechanisms whereby germ cells are induced to form gametes. Key determinants of this process have been identified in S. cerevisiae, making budding yeast an ideal model system to study entry into gametogenesis (reviewed in (van Werven and Amon, 2011)). In response to nutrient deprivation diploid budding yeast cells undergo gametogenesis to form four stress-resistant haploid gametes, called spores. This process is known as sporulation and is comprised of a specialized cell division, meiosis, to produce haploid gametes from a diploid precursor, and a developmental program that leads to the formation of spores.
Initiation of sporulation requires the convergence of multiple signals (reviewed in ((Honigberg and Purnapatre, 2003))). First, sporulation only occurs in cells of the diploid MATa/α mating type. Second, sporulation is only initiated under starvation conditions. Fermentable sugars and nitrogen sources must be absent and a non-fermentable carbon source must be present for sporulation to be initiated. Finally, cells must be able to respire. All these signals converge on the promoter of IME1, the master regulator of gametogenesis. IME1, inducer of meiosis 1, encodes a transcription factor that sets the sporulation program in motion (Kassir et al., 1988). When IME1 is transcribed, cells enter gametogenesis (Deng and Saunders, 2001; Kassir et al., 1988; Mitchell and Bowdish, 1992). Thus, IME1 gene expression regulation lies at the heart of gametogenesis control in budding yeast.
The IME1 promoter is over 2 kb in length, and one of the most regulated promoters in S. cerevisiae (reviewed in (Honigberg and Purnapatre, 2003; van Werven and Amon, 2011)). Little is known about the transcription factors that bring about nutritional and respiratory control of IME1 expression, but the mechanism that restricts IME1 expression to MATa/α diploid cells has been partially elucidated (Figure 1A). The transcription factor Rme1 binds to two RME1 binding sites in the IME1 promoter (~2 kb upstream of the translation start site) and inhibits IME1 expression in haploid cells (Covitz and Mitchell, 1993; Shimizu et al., 1998). In MATa/α diploid cells RME1 is not expressed. This is because the MATa locus encodes a1 and the MATα locus α2, which together form the a1–α2 repressor complex that inhibits RME1 expression (Figure 1A) (Covitz et al., 1991; Mitchell and Herskowitz, 1986). How Rme1 inhibits expression of IME1 in haploid cells is not understood.
IME1 is not the only inducer of sporulation whose expression is controlled by mating type. IME4 encodes a RNA methyl-transferase that is essential for initiation of sporulation in some strain backgrounds and contributes to efficient entry in others (Clancy et al., 2002; Hongay et al., 2006; Shah and Clancy, 1992). In MATa or MATα cells IME4 is not expressed because an antisense transcript (IME4-AS, also known as RME2), initiated from the 3’end of the IME4 locus, interferes with IME4 expression (Gelfand et al., 2011; Hongay et al., 2006). In MATa/α diploid cells, the a1–α2 complex inhibits the expression of the IME4 antisense RNA by directly binding to its promoter. Whether RME1 and IME4-AS are the sole mediators of mating type control of sporulation is not known.
Here we describe the mechanism whereby the cell’s mating type regulates IME1 expression and hence gametogenesis. We find that Rme1 induces the expression of a lncRNA in cells expressing the haploid MATa or MATα mating type but not in cells of the diploid MATa/α mating type. This lncRNA, termed IRT1, covers almost the entire IME1 promoter and functions in cis to prevent transcription factors from binding to the IME1 promoter. Interference with transcription factor binding is mediated by IRT1 transcription establishing a repressive chromatin state at the IME1 promoter. This requires the Set2 histone methyltransferase and the Set3 histone deacetylase complex (Set3C), indicating that co-transcriptional methylation of histones and recruitment of histone deacetylases is essential for IRT1 dependent silencing of the IME1 promoter. Furthermore, we define how the cell’s mating type regulates gametogenesis. Interfering with the expression of IRT1 and the antisense transcript at the IME4 locus is sufficient to allow cells expressing the haploid MATa or MATα mating type to sporulate as efficiently as MATa/α diploid cells. Conversely, expression of these two lncRNAs abolishes the ability of MATa/α diploid cells to sporulate. Our data demonstrate that transcription of two lncRNAs confers mating-type regulation of gametogenesis in budding yeast.
Recently a detailed map of non-coding RNAs in sporulating cells revealed transcriptional activity in the IME1 promoter (Figures 1B and S1; (Lardenois et al., 2011)). The IME1 gene itself is only expressed in cells of the MATa/α mating type and only under sporulation inducing conditions (Figure 1C, S1A). The gene is not expressed when nutrients are ample (Y). IME1 RNA begins to accumulate upon transfer of cells into sporulation-inducing medium (SPO medium; Figure 1C, 1D, S1A), increases during early stages of sporulation and declines thereafter.
Transcriptional activity was also detected in the IME1 promoter. A long promoter transcript, annotated as stable unannotated transcript 643 (SUT643) (Xu et al., 2009), is transcribed from the same strand as IME1 (Figure 1B). This transcript is weakly expressed in MATa/α diploid cells upon induction of sporulation, but highly expressed when MATα/α diploid cells are incubated in SPO medium (Figure S1B). Northern blot and quantitative RT-PCR analyses confirmed this result (Figure 1C, D). In MATa/a diploid cells and MATa haploid cells, SUT643 transcription is strongly induced in SPO medium and RNA levels remained high throughout the time course, despite the transcript being short-lived (Figure 1E, F). As expected, IME1 is not expressed (Figure 1C, D). This result shows that SUT643 and IME1 exhibit cell-type specific expression under sporulation-inducing conditions. In what follows, we show that SUT643 plays a key role in the control of IME1 expression. We therefore named the gene IRT1, for IME1 regulatory transcript 1. We detected a second, shorter transcript upstream of SUT643, designated as meiotic unannotated transcript 1573 (MUT1573), which was up-regulated during later stages of sporulation (Figure 1B, S1C). The significance of this transcript in IME1 regulation is presently unclear.
To define the relationship between IME1 and IRT1, we studied their expression in single cells. We measured IRT1 and IME1 RNA using RNA fluorescence in situ hybridization (FISH) in MATa/α diploid and MATa haploid cells upon transfer into sporulation-inducing conditions (Figure 2A, S2, S3) (Bumgarner et al., 2012; Raj et al., 2008). This analysis showed that 4 hours after transfer into SPO medium, IME1 is strongly expressed (average of ~44 transcripts per cell) and more than 90% of MATa/α cells harbor IME1 transcripts. In contrast, IRT1 RNA is barely detectable (Figure 2B, S3A).
In the MATa haploid strain we observed that upon induction of sporulation, ~80% of cells transiently expressed low levels of IME1, as defined by the presence of at least two IME1 RNA molecules in cells (0 – 60 minutes time points; Figure 2C (combine IME1 and IRT1/IME1); Figure S3B). The percentage of cells expressing IME1 decreased significantly at later time points. IRT1 expression was anti-correlated. The percentage of MATa cells expressing IRT1 was low upon transfer into SPO medium, but increased to ~80% within 2 hours (Figure 2C). We further observed that at times when IME1 RNA levels declined and IRT1 levels rose (30 to 60 minutes after transfer into SPO medium), cells harbored both IME1 and IRT1 transcripts (Figure 2C, S3B). This observation together with the finding that in other stages of sporulation, IME1 and IRT1 RNAs are mutually exclusive (Figure 2A, 2C), indicates that IME1 is transiently induced upon starvation even in cells that express the haploid MATa or MATα mating type but concomitantly with IRT1 induction, IME1 RNA levels decline in these cells.
The observation that IRT1 is expressed when IME1 is not, raises the possibility that IRT1 transcription mediates the repression of IME1 transcription. To test this we integrated the CYC1 transcriptional terminator 118 base pairs (bp) downstream of the transcription start-site of IRT1 (henceforth irt1-T). This led to the loss of full-length IRT1. Instead, a shorter IRT1 transcript was detected (marked with *; Figure 3A). Importantly, MATa/a diploid and MATa haploid cells harboring the irt1-T allele expressed IME1 (Figure 3A). A fraction of cells also underwent meiosis, which is lethal in haploid cells (Figure 3B, 3C). Thus, full-length IRT1 transcription is required for the repression of IME1 in cells expressing the haploid MATa or MATα mating type.
The transcription factor Rme1 is required for the repression of IME1 in cells of the haploid MATa or MATα mating type (Covitz and Mitchell, 1993). However, Rme1 does not behave like a classic transcriptional repressor. While Rme1 represses IME1 transcription, it functions as a transcriptional activator in the context of other promoters (Toone et al., 1995). The identification of IRT1 transcription as an inhibitor of IME1 expression raised the possibility that Rme1 activates IRT1 expression thereby inhibiting IME1 expression. Consistent with this hypothesis is the observation that the two Rme1 binding sites are located immediately upstream of the IRT1 transcription start site, and that their position within the IME1 promoter is highly conserved across Saccharomyces species (Figure 3D).
To test whether RME1 is required for IRT1 expression, we examined the consequences of deleting RME1. We found that IRT1 expression was lost in MATa rme1Δ haploid and MATa/a rme1Δ diploid cells or MATa haploid cells lacking the RME1 binding sites (Figure 3A, S4). IME1 was induced in all these strains (Figure 3A, S4) (Covitz and Mitchell, 1993). The degree of IME1 expression and degree of sporulation observed in the rme1Δ strain was remarkably similar to that of MATa/a cells expressing the prematurely terminated irt1-T allele (Figure 3A, 3B). Chromatin immunoprecipitation (ChIP) analysis further showed that Rme1 binding to the IRT1 promoter only occurs under conditions supporting IRT1 expression (Figure 3E). During vegetative growth and upon transfer into SPO medium Rme1 is not recruited to the IRT1 promoter and IRT1 is not expressed, but both events occur as cells enter the sporulation program. Our data show that Rme1 inhibits IME1 expression and hence sporulation in cells expressing the haploid MATa or MATα mating type through activation of IRT1 transcription. The observation that sporulation is not as efficient in rme1Δ or irt1-T MATa/a cells as in MATa/α cells further indicates that mating type control of sporulation must be mediated by additional factors.
The IRT1 transcript harbors several putative short open reading frames with the longest encoding a protein of 74 amino acids. If an IRT1-encoded protein was responsible for IME1 repression, the location of the IRT1 gene within the yeast genome should not affect the ability of IRT1 to inhibit IME1 expression. We performed two experiments to test this possibility. First, we created a haploid MATa strain in which the IRT1 locus is duplicated (Figure 4A). In this strain IME1 expression was inhibited (Figure 4B, S5A-C). However, when the IRT1 locus immediately upstream of IME1 harbored the CYC1 terminator (irt1-T allele), IME1 was expressed in the MATa haploid strain and cells underwent a lethal meiosis (Figure 4B, C).
The second way by which we tested the importance of IRT1 location with respect to IME1 regulation was by comparing the impact of constitutive expression of IRT1 from its native locus versus an ectopic locus. Expression of IRT1 from the constitutive GPD1 promoter (pGPD1- IRT1) was sufficient to prevent IME1 expression in MATa rme1Δ cells (Figure 4D). Furthermore, whereas MATa/a rme1Δ or MATa/α diploid cells readily sporulate, the same cells expressing pGPD1-IRT1 at the IRT1 locus showed poor sporulation (Figure S5D, E). Placing pGPD1-IRT1 upstream of an ectopic locus, a lacZ reporter gene integrated at URA3, did not affect the kinetics of entry into sporulation of MATa/a rme1Δ or MATa/α diploid cells, but lacZ expression was affected (Figure S5D-F). Our results show that IRT1 transcription represses IME1 in cis.
How does IRT1 transcription interfere with IME1 expression? IRT1 transcription could prevent the recruitment of IME1 transcriptional activators from binding the IME1 promoter. To test this possibility we examined the effects of IRT1 expression on the binding of known transcriptional activators to the IME1 promoter. In a screen to be described in detail elsewhere, we identified POG1 as being required for full IME1 expression. Pog1 activates CLN2 expression, and binds to the promoters of genes encoding cell cycle regulators (Horak et al., 2002; Leza and Elion, 1999). POG1 is also needed for wild-type level expression of IME1. In a pog1Δ strain IME1 expression is reduced and entry into and progression through sporulation are delayed (Figure 5A, B). Furthermore, Pog1 associates with the IME1 promoter in a region −750 and −1050 bp upstream of the translation start site. This binding is developmentally regulated, being low upon transfer into sporulation-inducing conditions, but increasing as cells progress through early stages of sporulation (3 hour time point; Figure 5C-E).
The identification of a direct activator of IME1 expression allowed us to assess the effects of IRT1 transcription on transcription factor binding at the IME1 promoter. In MATa/α diploid cells, Pog1 binding was induced under sporulation-inducing conditions (Figure 5E). In MATa haploid cells, Pog1 binding was also slightly elevated as cells entered the sporulation program (1 hour after transfer into SPO medium) but never increased to levels seen in MATa/α diploid cells (Figure 5E). Importantly, Pog1 binding at the IME1 promoter, was affected by IRT1. Pog1 was recruited to the IME1 promoter in haploid cells expressing the irt1-T allele but not in cells expressing full-length IRT1 (Figure 5F). These results indicate that at least one transcriptional activator of IME1 is differentially recruited to the IME1 promoter in MATa haploid and MATa/α diploid cells. Furthermore, our data demonstrate that IRT1 transcription inhibits transcriptional activators from being recruited to the IME1 promoter.
Transcription of IRT1 could antagonize IME1 expression via two not mutually exclusive mechanisms. Movement of the transcription machinery through the IME1 promoter could interfere with transcription factor binding. It is also possible that transcription through the IME1 promoter establishes a repressive chromatin state.
To determine whether IRT1 transcription establishes a repressive chromatin state at the IME1 promoter we examined nucleosome occupancy in MATa and MATa/α cells. Regions of low nucleosome occupancy, referred to as nucleosome free regions (NFRs), are found in promoters of transcriptionally active genes and are thought to allow transcription factors to bind to promoters. High nucleosome occupancy at promoters is indicative of repressive chromatin (reviewed in (Cairns, 2009)). We observed that nucleosome occupancy, as measured by histone H3 occupancy (Figure 5G-J), is differentially regulated between MATa haploid and MATa/α diploid cells. Nucleosome occupancy was high in both MATa haploid and MATa/α diploid cells during exponential growth when IME1 expression is low (Figure 5G). A NFR became apparent during starvation (saturated YPD and at the time of transfer into SPO medium) in both MATa haploid and MATa/α diploid cells, when IME1 is expressed at low levels in both cell types (compare Figures 5H and 5I with Figures 1 and and2).2). Shortly after transfer into SPO medium, high nucleosome occupancy was re-established in MATa haploid cells but not in MATa/α diploid cells (Figure 5J). These results show that nucleosome re-assembly at the IME1 promoter occurs in MATa cells at the time IRT1 is transcribed. Our data suggest that IRT1 transcription induces a repressive chromatin state, which prevents the recruitment of transcriptional activators to the IME1 promoter.
How does IRT1 transcription establish a repressive chromatin state at the IME1 promoter? Two previous studies have implicated the histone methyltransferase Set2 and the Set3 histone deacetylase complex in IME1 regulation. Deletion of either gene increases sporulation efficiency (Deutschbauer et al., 2002). SET3 was also shown to dampen IME1 expression in certain strain backgrounds (Pijnappel et al., 2001). Set2 and Set3 are directly involved in establishing repressive chromatin structures within transcribed regions (Carrozza et al., 2005; Keogh et al., 2005; Kim and Buratowski, 2009) and could thus be critical for repression of IME1 by IRT1 transcription.
Set1 and Set2 travel with RNA polymerase to deposit the repressive lysine 4 dimethylation (H3- K4-me2) and lysine 36 methylation (H3-K36-me) marks on histone H3, respectively (Carrozza et al., 2005; Keogh et al., 2005; Kim and Buratowski, 2009; Xiao et al., 2003). After 6 hours in SPO, when IRT1 is expressed in MATa haploid cells, both marks were significantly enriched in the IME1 promoter (Figure 6A-C) and, as expected, depended on SET1 and SET2 (Figure S6AC). We conclude that histone modifications characteristic of repressive chromatin are present in the IME1 promoter in cells expressing a haploid mating type.
To determine whether SET2 and SET3 were required for IRT1-mediated repression of IME1 we measured the expression of IME1 and IRT1 levels in MATa haploid cells lacking either SET2 or SET3 or both genes (note, unlike in other strain backgrounds (Krogan et al., 2003), deleting SET2 and SET3 did not lead to significant growth defects in SK1 cells). IRT1 expression was not affected in all three mutants, but IME1 expression was (Figure 6D, S6D). IME1 levels were somewhat elevated in the set2 and set3 single mutants but reached levels similar to that of cells lacking IRT1 transcription (irt1-T cells) in the double mutant (Figure 6D, S6E). Analysis of IME1 and IRT1 RNAs in single cells further showed that the two RNAs are co-expressed in set2 set3 double mutants (Figure 6E, F). The fraction of cells only expressing IRT1 (two transcripts or more per cell) decreased in the set2 and set3 single mutants and was the lowest in the set2 set3 double mutant (Figure 6F). The fraction of cells only expressing IME1 increased somewhat in all mutants, suggesting that SET2 and SET3 may be necessary for full IRT1 expression. Deleting SET2 and SET3 had the largest effect on the category of cells that co-express IRT1 and IME1. In the set2 set3 double mutant almost 50% of cells harbor both, IME1 and IRT1 transcripts. We conclude that repression of the IME1 promoter by IRT1 transcription is compromised in the set2 set3 double mutant.
To further study the role of Set2 and Set3 in IME1 expression, we analyzed the IME1 promoter architecture in set2 and set3 single and double mutants. In contrast to wild-type MATa cells, Pog1 is recruited to the IME1 promoter in the set2 set3 double mutant cells and also to some extent in the single mutants (Figure 6G). Furthermore, a nucleosome free region became apparent in the single and double mutants (Figure 6H).
Deleting SET2 and SET3 even allowed some sporulation to occur in cells expressing a haploid mating type. MATa/α set2 set3 mutants undergo sporulation with delayed kinetics presumably because the two genes are needed for other aspects of the sporulation program (Figure 6I). Deleting SET2 and SET3 however allowed a significant proportion of MATa/a cells to sporulate (Figure 6I), to produce viable spores (data not shown) and to induce a lethal meiosis in haploid cells (Figures S6F). These data demonstrate that IME1 repression by IRT1 transcription requires Set2 and Set3 to establish a repressive chromatin state in the IME1 promoter to prevent transcription factor recruitment. We propose that transcription of IRT1 deposits histone methylation marks, which recruit histone deacetylase complexes to repress the IME1 promoter (Figure 6J). At the 5’end of the IME1 promoter the histone H3 lysine 4 dimethylation mark directly recruits Set3 together with Set3C containing the histone deacetylases Hos2 and Hst1 (Kim and Buratowski, 2009). Consistent with this model is the observation of Set3 dependent recruitment of Hos2 to the IME1 promoter (Figure S6G). IRT1 transcription is also required for co-transcriptional Set2 dependent methylation of histone H3 at lysine 36. This mark recruits the histone deacetylase complex Rpd3C(S) (Carrozza et al., 2005; Keogh et al., 2005). Thus, IRT1 transcription represses the IME1 promoter by recruiting histone deacetylases.
Preventing IRT1 transcription allows MATa haploid and MATa/a diploid cells to induce IME1 and to enter sporulation. However, these cells do not sporulate with the same kinetics and efficiency as MATa/α diploids (Figure 3B). This observation indicates that other pathways exist that bring about mating type control of sporulation. IME4 regulation could be such a parallel pathway. In cells harboring only one mating type, expression of an IME4 antisense (IME4-AS) RNA prevents the expression of IME4 (Hongay et al., 2006). In MATa/α diploid cells, IME4-AS is repressed by the a1-α2 repressor and IME4 is expressed (Hongay et al., 2006).
To determine whether the IME4-AS and IRT1 transcripts collaborate to bring about mating type control of sporulation, we combined the irt1-T allele with an IME4 allele driven from the constitutive TEF1 promoter (pTEF1-IME4). Whereas each individual allele allowed 50% of MATa/a cells to sporulate with a delay, the combination of the two brought about sporulation efficiencies and kinetics seen in MATa/α diploid cells (Figure 7A).
We were also able to induce MATa/α levels of sporulation in MATa/a diploid cells by simply repressing transcription of IRT1 and IME4-AS. We constructed a strain carrying a TetR repressor fused to the transcription repressor Tup1 (TetR-Tup1; (Belli et al., 1998)). We then integrated tetO sites at the 5’end of the RME1 promoter (386 bp upstream of the RME1 translation startsite) and at the 3’ end of the IME4 gene (158 bp downstream from the IME4 stop codon) to replace the a1–α2 binding sites and hence a1–α2 regulation of RME1 and IME4-AS with that of the TetR-Tup1 fusion. MATa/a diploid cells that either harbor only tetO sites or express the TetR-Tup1 fusion in the absence of tetO sites did not sporulate (Figure S7A, B). When TetRTup1 was tethered either to the RME1 promoter or the IME4 3’end, a low percentage of cells sporulated (Figure S7A, B). However, when TetR-Tup1 was targeted to both sites simultaneously, MATa/a diploid cells formed spores with the same kinetics and efficiency as MATa/α diploids (Figure 7B). Similar results were obtained when the irt1-T allele was combined with the TetR-Tup1 repressible IME4-AS construct (Figure S7C). Our results show that inhibiting transcription of IRT1 and IME4-AS is sufficient to induce MATa/α levels of sporulation in MATa/a cells.
What are the effects of expressing IRT1 and IME4-AS in MATa/α cells? In MATa/α cells the a1–α2 repressor inhibits the transcription of the IRT1 transcription factor RME1 and IME4-AS. The RME1 promoter harbors two a1–α2 binding sites; the IME4-AS promoter has one (Figure 7C, S7D). We examined the consequences of deleting individual and combination of binding sites in MATa/α strains. Inactivating single a1–α2 sites in the RME1 promoter had little effect on sporulation (Figure 7D). Inactivating both a1–α2 binding sites in the RME1 promoter led to expression of RME1 in MATa/α cells similar to what is seen in MATa cells, indicating that the RME1 promoter is fully de-repressed (Figure S7E, compare MATa with 4). Consistent with this effect on RME1 expression, progression through meiosis and sporulation efficiency was significantly reduced in this mutant (Figure 7C, 7D, S7F). Deleting SET2 and SET3 suppressed the sporulation defect of cells with deletions of the a1–α2 binding sites in the RME1 promoter (Figure 7E, F), further confirming that SET2 and SET3 are required for IRT1 dependent repression of IME1.
Finally, we combined mutations in the a1–α2 binding sites in the RME1 promoter with a deletion of the a1–α2 binding site in the IME4-AS promoter. Deleting the IME4-AS a1–α2 binding site dramatically reduced sporulation in MATa/α cells (Figures 7C, 7D and S7G) (Hongay et al., 2006), but inactivation of all three a1–α2 binding sites obliterated sporulation (Figure 7D, S7G; strain number 8). We conclude that transcription of two long non-coding RNAs, IRT1 and IME4-AS, is the sole mediator of mating type control of sporulation in budding yeast.
The decision of whether or not to enter the developmental program that leads to gamete formation is governed by multiple extracellular and intracellular signals. Here we describe how the cell’s mating type regulates gametogenesis. The control is remarkably simple: transcription of two non-coding RNAs prevents, via distinct mechanisms, the expression of two central regulators of the sporulation program in cells expressing the MATa or MATα haploid mating type.
Understanding how the expression of IME1 is controlled lies at the heart of gamete formation and serves as a model to understand signal integration at promoters. We have unraveled the mechanism whereby the cell’s mating type controls IME1 expression. Several lines of evidence indicate that IRT1 transcription interferes with IME1 expression by preventing transcription factors from binding the IME1 promoter. First, full-length transcription of IRT1 through the IME1 promoter is needed for IME1 repression. Second, IRT1 functions in cis to inhibit the expression of downstream genes. This repressive cis-acting function of IRT1 is observed at the native locus and at an ectopic site. Third, Rme1 dependent repression of IME1 requires two components of the RNA polymerase mediator complex, RGR1 and SIN4 (Covitz et al., 1994; Shimizu et al., 1997). Finally, we observe that an activator of IME1, Pog1, is displaced from its binding site when full-length but not a truncated version of IRT1 is expressed.
How does IRT1 inhibit IME1 expression? The IRT1 RNA itself is unlikely to contribute to the repression of IME1 expression. IRT1 RNA is highly unstable and RNA FISH analysis showed that IRT1 transcripts do not localize to one region of the nucleus but are found throughout the cells. Furthermore, in the set2Δ set3Δ double mutant, IRT1 RNA is present in cells at levels seen in wild-type cells, yet IME1 is efficiently transcribed. Whether movement of the transcription apparatus through the IME1 promoter interferes with transcription factor binding is not yet known, but our data support a role for co-transcriptional chromatin modifications in establishing a repressive chromatin state at the IME1 promoter. IRT1 transcription is associated with an increase in nucleosome density and the repressive histone H3-K4-me2 and H3-K36-me marks at the IME1 promoter.
The inactive chromatin state at the IME1 promoter requires the Set2 histone methyltransferase and the Set3 histone deacetylase complex. Previous studies showed that the Set2/Rpd3C(S) pathway is essential for repression of cryptic transcription within long genes (Carrozza et al., 2005; Keogh et al., 2005; Li et al., 2007). Set3C is required for the repression of histone acetylation at the 5’end of genes (Kim and Buratowski, 2009). We propose that in the context of the IME1 promoter, these functions are employed to regulate expression of a downstream gene via lncRNA transcription. In cells expressing a haploid mating type, IRT1 transcription recruits the Set1 and Set2 histone methyltransferases. At the 5’end of the IME1 promoter Set1-mediated histone H3 lysine 4 dimethylation recruits the Set3 complex containing the histone deacetylases Hos2 and Hst1 (Kim and Buratowski, 2009) (Figure 6J). IRT1 transcription also promotes cotranscriptional Set2 dependent methylation of histone H3 at lysine 36. This mark recruits the histone deacetylase complex Rpd3C(S) (Carrozza et al., 2005; Keogh et al., 2005), which, we propose, contributes to the repression of the IME1 promoter. This is, to our knowledge, the first example of Set2 and Set3C working together to silence a promoter through lncRNA transcription. This novel mechanism of gene regulation could be wide-spread. A recent genome-wide study suggests that the majority of Set3 regulated genes have overlapping ncRNA transcript in yeast (S. Buratowski, personal communication). It may also occur in other species. In fission yeast, transcription of long mRNAs has recently been shown to establish heterochromatin islands to silence meiotic genes during vegetative growth (Zofall et al., 2012). This raises the interesting possibility that transcription of all kinds of RNAs serves to establish a silent chromatin state to inhibit the expression of neighboring genes. Transcription of lncRNAs has also been implicated in transcriptional activation (Hirota et al., 2008; Houseley et al., 2008; Pinskaya et al., 2009; Uhler et al., 2007). It will be interesting to determine the relative importance of lncRNA mediated transcriptional activation and repression in gene regulation and whether gene silencing mediated by long ncRNA transcription, as described here, also exists in higher eukaryotes.
The mechanism of IME1 repression by IRT1 has some parallels with what is observed at the SER3 locus. Like IRT1, SRG1, the non-coding RNA controlling SER3 expression, regulates its target in cis, increases nucleosome occupancy at the SER3 promoter, and prevents transcription factors from binding the SER3 promoter (Hainer et al., 2011; Martens et al., 2004). Nucleosome remodeling proteins, such as Spt2, Spt6 and Spt16, are important for transcription-dependent repression of SER3 by SRG1 (Hainer et al., 2011; Thebault et al., 2011). Whether these remodeling factors are needed for IME1 repression is not yet known. However, Set2 and Set3, important for IME1 repression, do not play a role in SER3 repression (Hainer et al., 2011). This is perhaps not surprising, given that repression of intragenic transcription by Set2 predominantly occurs at longer genes (Li et al., 2007) and SRG1 is a relative short ncRNA (~500 bp).
How Rme1 represses IME1 has been the subject of investigation for decades (Blumental-Perry et al., 2002; Covitz and Mitchell, 1993; Kassir et al., 1988; Mitchell and Herskowitz, 1986). Genetically, RME1 was shown to function as a repressor of IME1 expression but was found to activate transcription of CLN2 (Toone et al., 1995). Transcription reporter assays further showed that Rme1 functions as an activator or repressor depending on the position of the RME1 binding site within the promoter. A more distal binding site caused repression; location near the transcription start site brought about transcriptional activation (Covitz and Mitchell, 1993). Our findings provide a simple explanation for these results. Rme1 is an activator of transcription, which, when located at a distance from a transcriptional start site, can repress a target gene by inducing transcription through the promoter where it is located.
The single cell analysis of IME1 and IRT1 transcripts sheds light onto how IRT1 transcription through the IME1 promoter represses IME1 transcription in cells expressing the MATa or MATα haploid mating type. Both IRT1 and IME1 expression is under nutritional control. Both transcripts are repressed during vegetative growth. IRT1 transcription continues to be repressed in pre-sporulation medium and is activated only upon transfer into sporulation medium, which coincides with the recruitment of Rme1 to the IRT1 promoter. In contrast, IME1 transcription is already activated during growth in pre-sporulation medium. Remarkably, this pre-sporulation activation not only occurs in MATa/α diploid cells but also in cells expressing the MATa or MATα haploid mating type. Thus, IME1 is initially expressed in cells of all mating types in response to nutrient deprivation, but Rme1-mediated expression of IRT1 then down-regulates IME1 expression in haploid cells. Interestingly, the maximal number of IRT1 molecules per cell in MATa haploids is 10-fold lower compared to IME1 in MATa/α diploid cells. This finding that low level of IRT1 transcription is sufficient to repress IME1 expression is consistent with the idea that co-transcriptional silencing of the IME1 promoter by histone deacetylases is the major mechanism of IME1 repression. The observation that IRT1 is induced only after IME1 expression has been initiated, despite both promoters being under similar nutrient regulation, furthermore raises the interesting possibility that IME1 expression may be a prerequisite for IRT1 expression. Further studies will be needed to determine whether IME1 is required for its own downregulation in cells expressing the haploid mating types.
Transcription of IRT1 and IME4-AS is essential to prevent MATa or MATα haploid cells from entering a lethal meiosis. Interfering with their expression is sufficient to induce mating type independent sporulation that is indistinguishable from that of MATa/α diploid cells in both efficiency and kinetics. Conversely, deleting three a1–α2 binding sites, two at the RME1 promoter and one in the IME4-AS promoter, abolished the ability for MATa/α diploid cells to sporulate. Thus, transcription of two long non-coding RNAs is all that mediates mating type control of sporulation. Why did budding yeast evolve the use of lncRNA transcription to govern this key cell fate decision? Perhaps repression of transcription by lncRNA transcription is a more effective way of silencing complex promoters than by classic transcription repressors. The IME1 promoter is unusually long for a S. cerevisiae promoter (2.2 kb) and subject to complex regulation. Full repression of such a promoter would likely require the binding of repressors to multiple sites throughout the promoter. Repression by transcription of a lncRNA is simpler. It only requires two RME1 binding sites located upstream of the IME1 promoter. A similar rationale could apply to the use of antisense transcription to control the expression of genes with complex promoters. Antisense transcripts only require a single transcription initiation site at the 3’end. Another advantage of gene repression by lncRNA transcription is that repression is the default. Repression is alleviated only in MATa/α diploid cells, through the repression of IRT1 and IME4-AS.
lncRNAs are wide-spread both in vegetatively growing and sporulating budding yeast cells (Granovskaia et al., 2010; Lardenois et al., 2011). Many genes important for progression through sporulation have been shown to harbor antisense transcripts that are expressed during vegetative growth (Zhang et al. 2011). Regulation of gene-expression by lncRNAs also appears important for other developmental processes such as pseudohyphal growth or adaptation to changes in growth conditions (Bumgarner et al., 2009; van Dijk et al., 2011). The use of lncRNA transcription as a regulatory tool may impact biological processes beyond transcription. In fission yeast meiosis the sme2+ lncRNA has recently been shown to be required for pairing at this locus (Ding et al., 2012). Perhaps sme2+ transcription establishes a heterochromatic state at this locus that facilitates pairing of homologous chromosomes. Long non-coding RNAs are also frequently found in mammalian promoters (Guttman et al., 2009). The regulation of mammalian promoters is often complex, and integration of multiple inputs is the norm rather than the exception. Perhaps long non-coding RNAs in these systems too serve to inhibit transcription. The principles of cell fate control by lncRNAs in budding yeast may thus also shed light onto complex developmental decisions in higher eukaryotes.
All strains used in this study are derivatives of SK1 and are listed in Table S1, plasmids in Table S2. Gene or promoter deletions, tagging of genes, and plasmid constructions are described in the Extended Experimental Procedures.
Synchronous meioses were performed as described in (Falk et al., 2010). To examine viability (Figure 3C, ,4C,4C, S6G), cells were incubated for 14 days in sporulation medium at room temperature, before spotting five-fold serial dilutions on YPD plates.
Northern blot analysis was performed as described (Hochwagen et al., 2005) with minor modifications (Extended Experimental Procedures). Chromatin immunoprecipitation assays are as described in (van Werven and Timmers, 2006), RNA FISH analyses were performed as described in (Bumgarner et al., 2012) with minor modifications (Extended Experimental Procedures). β-galactosidase assays are described in (Jambhekar and Amon, 2008). Meiotic nuclear divisions were examined in cells fixed with 80% ethanol overnight and stained with DAPI. For each time point 100 cells were counted. Meiosis I or meiosis II cells were defined as cells with two or four distinct DAPI masses, respectively.
We are grateful to Sudeep Agarwala, Gerald Fink and Vincent Guacci for reagents, Stacie Bumgardner for suggestions, and Stephen Bell, Frank Solomon, Gerald Fink and members of the Amon lab for their critical reading of this manuscript. This work was supported by a grant GM62207 to A. A., a Rubicon-grant (825.09.004) from the Netherlands Organization for Scientific Research to F.W., by the National Science Foundation (ECCS-0835623) and NIH/NCI Physical Sciences Oncology Center at MIT (U54CA143874) to A.v.O. and G.N., and by the grants Inserm Avenir (R07216NS) and CREATE (NR11016NN) to M. P. A.A. is also an Investigator of the Howard Hughes Medical Institute.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.