Gcn4p is a well‐characterized bZIP transcription factor that activates more than 500 genes encoding amino acids and purine biosynthesis enzymes, and many stress–response genes under various stress conditions. Under these stresses, it had been shown that transcriptions of ribosomal protein (RP) genes were decreased. However, the detailed mechanism of this downregulation has not been elucidated. In this study, we present a novel mechanistic model for a repressive role of Gcn4p on RP transcription, especially under amino‐acid starvation. It was found that Gcn4p bound directly to Rap1p, which in turn inhibited Esa1p–Rap1p binding. The inhibition of Esa1p recruitment to RP promoters ultimately reduced the level of histone H4 acetylation and RP transcription. These data revealed that Gcn4p has simultaneous dual roles as a repressor for RP genes as well as an activator for amino‐acid biosynthesis genes. Moreover, our results showed evidence of a novel link between general control of amino‐acid biosynthesis and ribosome biogenesis mediated by Gcn4p at an early stage of adaptation to amino‐acid starvation.
Cells exist in various environmental conditions, some of which are beneficial for growth and proliferation and others that are harmful to the point of being life‐threatening. Such environmental stresses include abnormal oxygen concentrations, temperatures, osmotic pressures and pH in addition to nutrient limitations. Cell survival during exposure to stress limits energy availability, thereby increasing the degree of metabolic depression. Therefore, cells finely tune their use of energy and resources in order to adapt to extreme environmental changes. The energetic cost savings is accomplished through the depression of ion pumping, macromolecular synthesis and macromolecular turnover (Hand and Hardewig, 1996). These energy‐saving mechanisms are widespread, from animals during hibernation to single‐cell organisms such as yeasts.
The synthesis of ribosomes is the single most energy‐consuming process in living cells (Warner, 1999; Warner et al, 2001). In yeast, the ribosome is a massive RNA–protein complex that contains 79 ribosomal proteins (RPs), four different rRNAs and over 150 additional proteins involved in various steps of ribosome maturation. This implies that ribosome biogenesis requires the coordinated activities of all three RNA polymerases (Fatica and Tollervey, 2002). In fact, it has been revealed that 60% of total transcription activity is devoted to producing rRNA, and 50% of RNA polymerase II transcription is dedicated to RP genes (Warner, 1999). Thus, cells must adjust the synthesis of ribosomes to match the amount of energy and resources available, as growth conditions change. Many previous studies reported a strong and highly significant correlation between expression of RP genes and changes in growth conditions, such as growth dynamics, carbon sources, temperature and amino‐acid pools (Neuman‐Silberberg et al, 1995; Hand and Hardewig, 1996; Natarajan et al, 2001; Levy et al, 2007).
There are many signalling pathways known to influence in ribosome biogenesis. The target of rapamycin (TOR) kinase, ras–cAMP–protein kinase A (PKA), protein kinase C and membrane secretory signalling pathways are all known to be involved in RP transcription and ribosome biogenesis (Klein and Struhl, 1994; Neuman‐Silberberg et al, 1995; Nierras and Warner, 1999; Powers and Walter, 1999). Recent research shows that TOR pathway mediates RP expression through the RAS/cAMP pathway, independent of TAP42/SIT4 in yeast (Martin et al, 2004; Schmelzle et al, 2004; Chen and Powers, 2006). Nitrogen is a nutrient important for this signalling pathway (Crespo et al, 2002). In yeasts, TORC1 uses the AGC kinase Sch9p as a substrate to activate RP transcription (Urban et al, 2007). Otherwise, carbon is an important nutrient for activating the PKA signalling pathway (Schmelzle et al, 2004), which helps determine the availability of carbon and nitrogen (Neuman‐Silberberg et al, 1995) and also affects RP expression. Other stresses such as heat shock and amino‐acid starvation are also known to influence RP transcription (Li et al, 1999; Gasch et al, 2000; Causton et al, 2001; Natarajan et al, 2001). It is also important to note that the significant reduction in RP expressions is consistent with decreases in translation that occur under stressful conditions, and is explained by the redistribution of cellular resources towards the synthesis of stress–response proteins (Miller et al, 1979; Causton et al, 2001).
In yeasts, the level of RP expression is regulated almost entirely at the transcriptional level. Although RP genes are scattered over the entire genome and have different promoter sequences, common architectures have been found in the promoter regions (Lascaris et al, 1999). It has been revealed that most RP genes contain one or more Rap1p and/or Abf1p binding sites (UASrpg: upstream activating sequence of RP gene) in their promoters. This element is commonly found about 250–400 bp upstream of the transcription start site. More extensive chromatin immunoprecipitation (ChIP)‐chip analysis showed that RP promoters are the major functional targets of Rap1p, also known as ‘repressor and activator protein’ (Lieb et al, 2001). Rap1p is an abundant, sequence‐specific DNA binding protein located in the nucleus, and functions as a context‐dependent repressor or activator of transcription (Shore, 1994; Pina et al, 2003). It promotes the transcription of many heavily transcribed genes, including those encoding glycolytic enzymes, RPs and several components of the translational machinery (Goncalves et al, 1995; Lascaris et al, 1999; Joo et al, 2009b). As a transcriptional activator, Rap1p binds to a specific DNA sequence (ACACCCRYACAYM; Lieb et al, 2001). During transcription activation, it has a general role as a ‘chromatin opener’ to facilitate the binding of other transcription activators to binding sites adjacent to the T‐rich region (Yu and Morse, 1999).
Several factors also have been implicated in the transcription of RP genes in yeasts. For example, Fhl1p, Ifh1p and Crf1p are among the most well‐characterized transcription factors involved in the initiation of RP transcription (Martin et al, 2004; Schawalder et al, 2004; Wade et al, 2004; Rudra et al, 2005). TOR controls Fhl1p via the RAS–PKA–YAK1 effector pathway along with two Fhl1p cofactors, Ifh1p and Crf1p, the coactivator and corepressor, respectively. Sfp1p, known as a negative regulator of TORC1, specifically inhibits TOR kinase activity that acts towards Sch9p kinase (Lempiainen et al, 2009). Recently, the mechanisms by which the general amino‐acid control (GAAC) and TOR pathways are integrated for optimal growth adaptation were reported (Staschke et al, 2010). Recruitment of all these factors appears to depend upon the state of histone acetylation of RP promoters, which is regulated by Esa1p and Rad3p (Reid et al, 2000; Rohde and Cardenas, 2003). The catalytic subunit of the NuA4 complex, Esa1p, is recruited to RP promoters for transcriptional activation under normal physiological conditions (Reid et al, 2000), and Rpd3p is recruited for transcriptional repression under stressful conditions such as cold shock (Robert et al, 2004). This correlation between factor recruitment and transcriptional regulation does not appear to be global, but rather specific for the RP promoter regions. Moreover, all the mechanisms appear to occur in a Rap1p‐ and UASrpg‐dependent manner (Moehle and Hinnebusch, 1991; Reid et al, 2000; Zhao et al, 2006; Layer et al, 2010). It is also worth noting that recent studies have revealed the functional relevance of Rap1p in transcription activation through protein–protein interactions (Garbett et al, 2007; Bendjennat and Weil, 2008; Papai et al, 2010).
Gcn4p is a well‐characterized bZIP‐containing transcription factor that binds to a consensus sequence ATGA(C/G)TCAT (Hope and Struhl, 1985; Hill et al, 1986). A bZIP dimer of Gcn4p containing two parallel coiled‐coils allows binding to the consensus sequence (Ellenberger et al, 1992). This is essential for the induction of amino‐acid and purine biosynthetic genes upon environmental signals, such as adenine limitation and amino‐acid starvation induced by 3‐amino‐1,2,4‐triazole (3‐AT), a competitive inhibitor of His3p (Joo et al, 2009a; Natarajan et al, 2001). In such cases, the level of Gcn4p is upregulated by an intricate translational control mechanism involving Gcn2p kinase and the phosphorylation of its substrate, eIF2α (Hinnebusch, 2005). This regulatory response is known as GAAC, in which uncharged tRNA is an upstream activator (Dong et al, 2000). It was also previously reported that RP genes were repressed in a Gcn4p‐dependent manner along with the expression of other translational proteins (Natarajan et al, 2001). However, the bona fide mechanism of the repression has not been known.
In this study, we investigated the detailed mechanism of RP transcription mediated by Gcn4p. First, we confirmed the previously reported Gcn4p‐dependent regulation of RP transcription. Next, we revealed that Gcn4p physically interacts with Rap1p, resulting in a Gcn4p recruitment on the promoter region of RP genes. We found that Gcn4p inhibited Rap1p–Esa1p interaction, which is responsible for full transcription of RP genes through histone H4 acetylation on the promoter. From these results, we hypothesized that derepressed Gcn4p under amino‐acid starvation inhibits RP transcription by interacting with Rap1p, establishing an optimal and effective means for redirecting resources and apparatuses for preferential transcription of stress–response genes, thus reducing unnecessary energy consumption.
Transcription of RP genes are downregulated in a Gcn4p‐dependent manner under amino‐acid starvation
To examine the regulation of RP transcripts under various amino‐acid starvation conditions, the levels of RP transcripts were examined. Northern blotting results clearly showed that TRP3 mRNA significantly increased under all amino‐acid starvation conditions (Figure 1A). Simultaneously, transcription levels of some sets of RP genes (RPS3, RPS11B and RPL19B in Figure 1A; RPS31 in Supplementary Figure S1) and PGK1, which contain UASrpg(s) in their promoters, were inversely proportional to that of TRP3. However, the transcription of two RP genes including RPL3 and RPL18B, which do not possess UASrpg, did not show significant decrease under amino‐acid starvation. These results showed mutually exclusive transcription of Gcn4p target genes and RP genes only containing Rap1p site on the promoter region, implying that Rap1p on the RP promoter appears to be responsible for the downregulation of RP genes under amino‐acid starvation conditions.
To further examine Gcn4p‐dependent regulation of RP transcription, previously studied using microarray (Natarajan et al, 2001), northern blot analysis was performed with GCN4 wild‐type and null strains after 3‐AT treatment (Figure 1B). The level of RP mRNAs was decreased by 3‐AT treatment only in the wild type. Although RP mRNA was also slightly decreased after 2 h in the GCN4 deletion strain, the degree of downregulation was much lower than that in wild type. To exclude any off‐target effects produced by 3‐AT, we performed northern blot analysis with the same strains under amino‐acid depletion (Figure 1C). We found that this regulation does not occur in a strain‐specific manner using other GCN4 wild‐type and deletion strains possessing different genetic backgrounds (Supplementary Figure S2).
We next explored how Gcn4p regulates RP transcription. There are several plausible explanations for this regulation. Gcn4p could affect the steady‐state level of RP mRNA. Through northern blot analysis, this hypothesis could be excluded based on the results showing that basal levels of RP transcripts were almost the same in both wild‐type and null strains (Supplementary Figure S3). Gcn4p might also regulate the stabilities of RP mRNAs. We checked the half‐lives of RP mRNAs using thiolutin, which inhibits all three RNA polymerases in yeast. Under both amino‐acid depletion and normal physiological conditions, the stabilities of RP mRNAs were not different from each other in both isogenic stains (Supplementary Figures S4 and S5), implying that Gcn4p reduces transcription initiation of RP genes. These observations were consistent with our previous finding that transcription initiation was diminished by 3‐AT treatment (Joo et al, 2009b). From these results, it was clearly confirmed that RP genes are negatively regulated under amino‐acid starvation in a Gcn4p‐dependent manner. Moreover, it was suggested that Gcn4p may exert an effect on the initiation process of RP gene transcription.
Gcn4p directly interacts with Rap1p
To understand how Gcn4p represses initiation of RP transcription, we searched for Gcn4p sites in the promoter of RP genes. However, most of the RPL and RPS genes lack Gcn4p sites. Moreover, our observations of PGK1 mRNA led us to believe that Rap1p may have a pivotal role for the downregulation of RP genes under amino‐acid starvation (Supplementary Figure S1). Thus, we preferentially focused on the transcription factor Rap1p as a target of Gcn4p, as many Rap1p target genes have been shown to be repressed by 3‐AT in a Gcn4p‐dependent manner (Natarajan et al, 2001). In addition, many recent studies had revealed that several transcription factors bind to Rap1p directly, and this interaction was important for initiation mechanism for full transcription of RP genes (Garbett et al, 2007; Rudra et al, 2007; Bendjennat and Weil, 2008; Papai et al, 2010). Therefore, physical interaction between Rap1p and Gcn4p was strongly expected. To investigate Rap1p–Gcn4p interaction, we generated a strain bearing N‐terminal TAP (tandem affinity purification)‐tagged RAP1. After standard TAP purification, Gcn4p was identified as a co‐binding protein of Rap1p by mass spectrometry (Supplementary Figure S6). To confirm a direct interaction and identify the binding domain, in vitro GST pull‐down assays were performed with His6–Rap1 and GST–Gcn4 variants or with GST–Gcn4 and His6–Rap1 variants. Detailed mapping of the Gcn4 interaction site revealed that amino‐acid residues 221–247 of Gcn4p mediate the binding with Rap1p (Figure 2A). This region contains a DNA binding domain and a NLS2 signal. For Rap1p, the domain responsible for the interaction with Gcn4p is located between amino‐acid residues of 361 and 596 (Figure 2B), which also contains the DNA binding domain of Rap1p.
Next, both proteins were tagged at the C terminus with FLAG or seven c‐myc epitopes for overexpression of exogenous protein using a galactose‐inducible promoter. Subsequent co‐immunoprecipitation (Co‐IP) with an anti‐FLAG antibody clearly showed that Rap1–FLAG was associated with Gcn4–myc7 in vivo (Figure 2C). Finally, we investigated this interaction using native levels of Rap1p and Gcn4p. As shown in Figure 2D, we confirmed Rap1p–Gcn4p binding repeatedly in vivo by Co‐IP with a strain containing Gcn4–HA3, which was regulated by an endogenous promoter. In addition, alkaline phosphatase treatment led us to believe that binding occurred in a phosphorylation‐dependent manner. Although this binding was confirmed both in vitro and in vivo, we were unable to exclude the possibility that chromosomal DNA mediates the binding, as both have DNA binding activities. Therefore, Co‐IP was performed in the presence of DNaseI (Figure 2C) or EtBr (Figure 2D), confirming that this interaction was chromatin DNA independent. In summary, we found that Gcn4p binds directly to Rap1p through the DNA binding domains of each protein.
The promoters of RP genes are selectively occupied by Gcn4p under amino‐acid starvation
Given that Gcn4p interacts with Rap1p and this binding is responsible for transcriptional regulation of RP genes, Gcn4p may be recruited to RP promoter regions with respect to the Rap1p binding element. Other transcription factors such as Fhl1p and Ifh1p have been shown to be recruited to the RP promoter through direct binding to Rap1p (Rudra et al, 2007). ChIP analysis of a strain expressing tri‐HA‐tagged Gcn4p revealed that the promoter of RP genes was occupied by Gcn4p under amino‐acid starvation (Figure 3A). As starvation time increased, the promoter occupancy of the RP genes by Gcn4–HA3 was increased, as well as that of HIS3 and ADE3, which are well known as Gcn4p target genes. Weaker binding to the RP promoters compared with HIS3 and ADE3 promoter binding might be due to indirect association through protein–protein interaction. Moreover, it is worth noting that promoter occupancy of RPL18B, which lacks a Rap1p site (UASrpg) in its promoter region, was not detected. However, that of PGK1, non‐RP genes containing Rap1p site in the promoter, was clearly observed. These results implied promoter occupancy of RP genes by Gcn4p in a Rap1p site‐dependent manner. In order to confirm Gcn4p recruitment to the RP promoter, ChIP was performed using primer pairs spanning the promoter region of one of the RP gene, RPS3 (Figure 3B). Promoter occupancy by Gcn4p was highest in the UASrpg region, but relatively lower in the upstream and downstream ORF regions, thus validating our hypothesis that Gcn4p was recruited to RP promoter region in a Rap1p‐ or Rap1p site‐dependent manner.
Binding activity to Rap1p, not trans‐activity, is essential for the repressive role of Gcn4p in RP gene transcription
Although Gcn4p‐dependent transcriptional regulation of RP genes and Rap1p‐dependent Gcn4p recruitment to RP promoter were demonstrated, this does not convincingly suggest that Rap1p recruits Gcn4p through direct binding for repression of transcription. There is another possibility that Gcn4p indirectly affects RP transcription through trans‐target genes. This is plausible as Gcn4p is the best‐characterized transcription factor in yeasts and is known to have more than 500 trans‐target genes (Hinnebusch, 2005). To investigate whether trans‐activity has an indirect effect on RP transcription, Ser242, a highly conserved amino‐acid residue in the DNA binding domain of Gcn4p was mutated to leucine (Ellenberger et al, 1992; Kim and Struhl, 1995; Seong et al, 2007). Consequently, the transcriptional activity of this mutant was abolished (Figure 4B and E). On the other hand, both the point mutant and wild‐type proteins appear to have similar binding affinity to Rap1p, as equal amount of His6–Rap1 recombinant protein was precipitated by GST pull‐down (Figure 4A). To understand whether Gcn4p directly participates in RP transcription, wild‐type and mutant Gcn4p were overexpressed and the levels of RP mRNA were determined. Both exogenously expressed forms of Gcn4p repressed transcription of RP genes, although only wild‐type Gcn4p induced transcription of trans‐target genes, such as HIS4 and ADE3 (Figure 4B). Next, we executed ChIP experiment to investigate whether wild‐type and mutant Gcn4p associate with RP promoters. Statistically significant promoter occupancies of RP genes (RPS3 and RPL19B) were detected in cells expressing both wild‐type and mutant Gcn4p (Figure 4C). These results are consistent with the northern blot result shown in Figure 4B, confirming that Gcn4p associates with the RP promoter adjacent to UASrpg through direct binding to Rap1p. Furthermore, it was revealed that the trans‐activity of Gcn4p and its downstream trans‐target genes were not involved in the reduction of RP transcription under amino‐acid starvation.
To investigate the in vivo relevance of Rap1p–Gcn4p binding, strains were generated in which either Gcn4S242L–HA3 or Gcn4ΔC–HA3 allele was expressed at normal physiological levels as the only form of Gcn4p. The Gcn4ΔC–HA3 mutant is the same truncation mutant used in Figure 2A (GST–Gcn4Δ5). As shown in Figure 4D, both the wild‐type and point mutant of Gcn4p could bind to Rap1p in vivo, but as expected, the Gcn4ΔC–HA3 mutant could not (Figures 2A and 4A). Next, downregulation of RP transcription was investigated in four GCN4 isogenic strains using northern blotting analysis (Figure 4E). Induction of TRP3 by 3‐AT was detected only in the wild‐type strain. On the other hand, RP transcription was repressed in both wild‐type and Gcn4S242L–HA3 mutant strains, but not in the Gcn4ΔC–HA3 or gcn4‐null strains. Moreover, it was revealed by ChIP analysis that both wild‐type and S242L mutant Gcn4p were recruited to RP promoters, but not Gcn4ΔC (Figure 4F). Given the fact that expression levels of these Gcn4p variants were almost all identical (Supplementary Figure S7), these data led us to hypothesize that Gcn4p was recruited to the RP promoter for transcription repression through direct binding to Rap1p under amino‐acid starvation.
Promoter occupancy of RP genes by Esa1p was decreased in a Gcn4p‐dependent manner under amino‐acid starvation condition
We next wanted to understand how Gcn4p exerts a negative effect on RP transcription through direct binding to Rap1p. Previous reports found that transcriptional regulators, such as Fhl1p, Ifh1p, Sfp1p and Esa1p, were recruited to RP promoter in a Rap1p‐dependent manner (Shore and Nasmyth, 1987; Reid et al, 2000; Martin et al, 2004; Kasahara et al, 2007; Lempiainen et al, 2009). It is possible that Gcn4p influences these recruitments. Therefore, we generated strains expressing Ifh1‐HA3, Fhl1‐HA3, Sfp1‐HA3, Esa1‐HA3 and TAP‐Rap1 in each JS143‐7D (GCN4) and YJK101 (Δgcn4) genetic background, respectively. First, the expression levels of these factors were investigated by immunoblotting (Figure 5A). As expected, repressed basal expression and significant induction of Gcn4p were clearly observed following 3‐AT treatment. There were no changes in Rap1p, Esa1p or Pgk1p expression. However, the expression of other factors was significantly reduced by 3‐AT treatment. Although the reduction of Sfp1p was less than that of other factors, the patterns of reduction for three proteins (Ifh1p, Fhl1p and Sfp1p) were observed. To investigate Gcn4p‐dependent factor recruitment to RP promoters, ChIP was performed in the same strains (Figure 5B–F). The results shown in Figure 5F revealed that promoter occupancy by Rap1p was unchanged following 3‐AT treatment. The recruitment of Fhl1p was decreased under amino‐acid starvation, but does not seem to be related to Gcn4p (Figure 5B). It is important to note that the recruitment of Esa1p was diminished in a Gcn4p‐dependent manner (Figure 5D). However, Ifh1p and Sfp1p bindings were not detected in our ChIP experimental scale (Figure 5C and E). These results strongly suggested that Esa1p binding to RP promoters is suppressed in a Gcn4p‐dependent manner under amino‐acid starvation, indicating that Gcn4p may directly influence Esa1p recruitment.
Gcn4p inhibits the physical interaction of Esa1p–Rap1p
Direct involvement of Gcn4p in RP transcription was observed in our studies. In addition, previous studies reported that recruitment of Esa1p to RP promoters is mapped to Rap1p binding sites, which implies that Esa1p is recruited by Rap1p (Reid et al, 2000; Rohde and Cardenas, 2003). Therefore, physical interaction between Esa1p and Rap1p is strongly suggested. To validate this prediction, an in vitro GST pull‐down assay was performed, resulting in confirmation of Esa1p–Rap1p binding (Figure 6A). This binding was also revealed in vivo by Co‐IP analysis in cells with overexpressed (Figure 6B) and endogenous levels of Rap1p and Esa1p (Supplementary Figure S8).
To verify Gcn4p‐dependent regulation of Rap1p–Esa1p binding, another Co‐IP analysis was performed with isogenic strains, BY4741 and BY4741Δgcn4. As shown in Figure 6C and D, the association between ectopically expressed HA–Esa1 and Rap1–FLAG was repressed only in when Gcn4p was overexpressed. This was verified in cells expressing native levels of Rap1p and Esa1p (Figure 6E, Supplementary Figure S9). Interestingly, the decreased association between the two factors in the absence of sulfometuron methyl or 3‐AT was observed in the gcn4‐null mutant. This suggested that a tight interaction between Esa1p and Rap1p is not required under nonstarvation conditions in cells lacking GCN4 for efficient recruitment of Esa1p to RP promoters. The observations of Gcn4p–Rap1p and Esa1p–Rap1p binding led us to predict an inhibitory role of Gcn4p for Esa1p–Rap1p binding. Derepressed Gcn4p expression under amino‐acid starvation may be able to hinder the Esa1p–Rap1p interaction as a consequence of diminished Esa1p recruitment to RP promoter by Rap1p. To validate this model, an in vitro binding competition assay was performed. Co‐precipitation of His6–Rap1 was decreased as the amount of Gcn4–His6 in the binding mixture was increased (Figure 6F). Likewise, when Gcn4–His6 was mixed with the His6–Rap1–GST–Esa1 binding complex, His6–Rap1 was eluted from the pre‐existing binding complex by Gcn4–His6, but not by BSA (Figure 6G). The amount of His6–Rap1 eluted was increased depending on the concentration of Gcn4–His6. We also determined the molar ratio of the three proteins under amino‐acid starvation (Supplementary Figure S10). Although the level of Rap1p was about four‐ to five‐fold higher than others, the amount of Gcn4p was sufficient for our mechanistic model. Taken all together, our data clearly showed that Gcn4p physically bound to Rap1p, which in turn inhibited Rap1p–Esa1p interaction.
It has been known that the recruitment of Esa1p to RP promoters along with the acetylation of histone H4 by NuA4 coordinately regulate transcription (Reid et al, 2000). Therefore, disturbance of Rap1p–Esa1p binding by Gcn4p may lead to a decrease in histone H4 acetylation specifically in the RP promoter. ChIP analysis with an anti‐acetyl‐histone H4 antibody revealed that histone H4 acetylation of RP promoters was decreased only in the GCN4 wild‐type strain under amino‐acid starvation (Figure 7A), consistent with the Esa1p ChIP data in Figure 5D. In summary, we concluded that derepressed Gcn4p binds to Rap1p under amino‐acid starvation, thereby inhibiting Rap1p–Esa1p interaction and resulting in decreased histone H4 acetylation of the promoter. Taken together, our results clearly concluded that transcription of RP genes was repressed by Gcn4p through direct binding to Rap1p. This regulation appears to give yeast cells a means to re‐direct resources from major energy‐consuming RP transcription process for energy conservation under amino‐acid starvation.
In this study, we uncovered a novel role for Gcn4p as a negative regulator of transcription, and proposed a novel link between GAAC and ribosome biogenesis. Gcn4p–Rap1p binding was confirmed both in vitro and in vivo, and its biological relevance was investigated by ChIP and northern blot analyses. This interaction promotes promoter occupancy of RP genes by Gcn4p during amino‐acid‐limiting conditions. We also found target proteins influenced by Gcn4p–Rap1p binding based on the hypothesis that this interaction is responsible for the reduction of RP transcription under amino‐acid starvation. ChIP analysis using five putative target proteins confirmed that the recruitment of Esa1p was inhibited in a Gcn4p‐dependent manner. Taken together, we generated a simple mechanistic model defining the roles of Gcn4p under amino‐acid starvation (Figure 7B). When yeast cells are grown in an amino‐acid‐rich condition, GAAC signalling is turned off and RP expression becomes activated by Rap1p and Esa1p, as well as by many other transcription factors such as Ifh1p, Fhl1p and Hmo1p. On the contrary, Gcn4p is induced through GAAC signalling when amino acids are limited, initiated by uncharged tRNA and propagated by activation of eIF2α kinase, Gcn2p (Hinnebusch, 2005). Subsequently, derepressed Gcn4p activates the transcription of many target genes encoding enzymes responsible for amino acid and purine biosyntheses, or that of amino‐acid permeases in order to compensate for limitations in amino‐acid availability (Joo et al, 2009a; Hinnebusch and Natarajan, 2002). Upon activation, transcription of RP genes is suppressed by Gcn4p through the direct binding of Gcn4p to Rap1p.
Cross‐talk between GAAC and ribosome biogenesis
Several benefits can be provided by a present Gcn4p‐dependent mechanism. In yeasts, regulation of gene expression occurs predominantly at the transcriptional level, and reprogramming of transcriptions is pivotal for encountering drastic changes in environment. Among all transcription processes in yeasts, the mechanism of RP transcription is the best characterized with respect to growth rate and various environmental cues (Jorgensen et al, 2004), and also known to be the most energy‐consuming process. Therefore, the transcription of RP genes must be finely regulated to avoid wasting resources. Under amino‐acid starvation, cells must stop the synthesis of general proteins. It is noteworthy that inhibition of general protein expression and induction of amino‐acid biosynthetic enzymes are thought to be executed by trans‐acting proteins that control GCN4 translation. Moreover, the molecular events that stimulate GCN4 translation also inhibit the rate of general protein synthesis (Hinnebusch, 2005).
Reduction of ribosome biogenesis can be executed by degradation of protein and protein complexes. Adaptation to severe environmental stresses is related to metabolic depression, such as the ability to reduce ion pumping activities, macromolecular synthesis and macromolecular turnover (Hand and Hardewig, 1996). Furthermore, it was reported that turnover of mature ribosomes proceeds nonselective and selective autophagy, called ‘ribophagy’, under severe and long‐term amino‐acid starvation (Beau et al, 2008; Kraft et al, 2008). Therefore, these previous data along with our observations suggest that the reduction of RP transcription and activation of amino‐acid biosynthetic enzymes by Gcn4p is responsible for rapid and short‐term stress adaptation. Otherwise, when the starvation period is extended, cells might induce ribophagy resulting in the breakdown of excess mature ribosomes into materials for the synthesis of other essential molecules. The repression of RP transcription by Gcn4p discovered here has therefore been assigned a new role in adaptations, as it may lead yeast cells to save and redirect available energy, resources and transcriptional machineries in a quick response to amino‐acid depletion. Accordingly, it is important to note that Gcn4p has more than 500 target genes that require lots of transcriptional machinery for activation under stress conditions. Therefore, plenty of transcriptional apparatuses as well as resources and energy are needed for their transcriptional induction. However, as previously mentioned, 50% of RNA polymerase II transcription as well as 90% of mRNA splicing are devoted to producing RPs (Warner, 1999; Warner et al, 2001). Reduction of this massive transcriptional process leads to the efficient induction of its target genes, as well as other genes related with adaptation. This cross‐talk between GAAC and ribosome biogenesis facilitated by Gcn4p can occur under conditions other than amino‐acid starvation. Gcn4p can be stimulated by ultraviolet light and methyl methanesulfonate (Hill et al, 1986; Natarajan et al, 2001), and expression of ribosome biosynthetic genes is highly responsive to environmental changes (Levy et al, 2007). For a full understanding of regulation between massive biological processes, Gcn4p‐dependent regulation of RP genes provides a sophisticated explanation for how regulation between major biological processes may be accomplished.
Complicated regulation of RP transcription through multiple pathways
All ribosomal components exist in equimolar levels in yeasts as a consequence of a common regulatory mechanism. The rate of ribosome synthesis responds to not only nutritional cues but also signals associated with other macromolecular synthesis pathways. Recent results have revealed that many factors, such as Rap1p, Abf1p, Fhl1p, Ifh1p, Crf1p, Hmo1p, Sfp1p and Rpd3p, participate directly in the transcription of RP genes, but their detailed mechanisms are not fully described. Such complications could be due to the complexity and heterogeneity that are common in RP promoter architectures. Computational research revealed that all RP promoters are not identical, although duplicated Rap1p sites are common (Lascaris et al, 1999). RP genes can be divided into several subclasses based on the abundance of Hmo1p and the dependence of Fhl1p and Rap1p promoter binding to Hmo1p (Kasahara et al, 2007). However, there are some questions about the involvement of Hmo1p. For instance, a previous study found that loss of Hmo1p abolishes Fhl1p binding but does not significantly affect RP transcription (Hall et al, 2006), an unexpected result that challenges previous observations (Martin et al, 2004; Schawalder et al, 2004; Wade et al, 2004; Rudra et al, 2007). Moreover, the response of RP promoters to TOR inhibition is independent of the Ifh1p‐related protein Crf1p, indicating that the role of this protein is strain‐specific (Zhao et al, 2006). Other groups revealed that histone acetylation is selectively increased in the promoter regions of RP genes (Reid et al, 2000; Rohde and Cardenas, 2003). The histone acetyl transferases Gcn5p and Esa1p are recruited to the promoters of actively transcribing genes, including those encoding RP (Robert et al, 2004). However, other studies have found that histones are essentially absent from the Rap1p sites of RP promoters (Zhao et al, 2006).
Here, we showed several lines of evidence demonstrating parallel mechanisms of RP gene downregulation under amino‐acid starvation. Western blot analysis in Figure 5A clearly revealed that the levels of Fhl1p and Ifh1p were decreased by 3‐AT in both wild‐type and null strains of GCN4. Moreover, expression levels of Fhl1p seemed to be regulated in a Gcn4p‐dependent manner. It was possible that Gcn4p might have an effect on RP gene regulation indirectly through the modulation of Fhl1p expression level. Fhl1p and Ifh1p reduction by 3‐AT could explain why RP mRNA (Figure 1B) and promoter occupancy by Fhl1p (Figure 5B) were both decreased in the gcn4‐null strain, although to a lesser degree compared with wild type. These results showed that yeast cells have many parallel pathways that regulate RP transcription, and that the novel regulatory mechanism shown here appears to be one of these. However, it is worth noting that all these phenomena occurred in a Rap1p‐dependent manner, although no specific regulator controls RP transcription.
Our observations also showed several lines of evidence that other factor(s) can be involved in the repressive mechanism of Gcn4p. First of all, RP transcriptions were decreased in the absence of GCN4 (Figure 1B), and transcripts of RP genes that contain no Rap1p sites in their promoters also were slightly reduced (Figure 1A). Interestingly, the degree of reduction in both cases was almost the same (80% of the control). These results indicated that 20% of RP mRNA appears to be decreased by other mechanisms not yet elucidated. Moreover, the binding domain of Rap1p, which is responsible for Gcn4p binding, is exactly identical to that for TBP binding (Bendjennat and Weil, 2008). These led us to hypothesize that Gcn4p could hinder the association of Rap1p with general transcription machineries, such as TFIID. Furthermore, we also found that the phosphorylation status of Gcn4p had an effect on Rap1p binding (Figure 2D). This implies that an unknown upstream kinase also participates in this process. Moreover, in the case of the S242L mutant, the repressive effect on RP transcription and recruitment to RP promoters were slightly diminished, although binding affinity to Rap1p was equal in vitro (Figure 4D and E). These results imply that other factors should be involved in Gcn4p–Rap1p binding to RP promoters in vivo. Therefore, further research into Rap1p binding partners to the RP promoter region will be performed in order to fully understand how yeast cells regulate RP expression.
Materials and methods
Yeast strains and culture conditions
Characteristics of the Saccharomyces cerevisiae strains used in this study are summarized in Supplementary Table S1. All strains containing tri‐HA‐tagged and N‐terminal TAP‐tagged proteins were generated using PCR‐based gene modification methods as previously described (Longtine et al, 1998). For N‐terminal TAP tagging, cassettes were amplified by PCR with pBS1761–RAP1pro or pBS1761–ESA1pro generated from pBS1761 (EUROSCRAF). To generate Gcn4S242L–HA3 containing strain (YJK125), a general two‐step PCR‐based targeted mutagenesis method was performed. For generation of a tagging cassette containing point mutation, genomic DNA from a strain harbouring Gcn4–HA3 (YJK102) was used as a PCR template. Strain generation was confirmed by genomic DNA sequencing. Yeast cells were grown in proper SC media (synthetic complement; contained 2% glucose, 0.67% yeast nitrogen base without amino acids (BD) and 15 or less proper amino acids). For galactose induction, cells were grown in SRG media (SC with raffinose and galactose; contains 2% raffinose, 2% galactose, 0.67% yeast nitrogen base without amino acids (BD) and 15 or less proper amino acids), and detailed procedures are described in the Supplementary Experimental Procedures.
Primers and plasmids
Primers used in this research are summarized in Supplementary Table S2. A pESC yeast epitope tagging system (Stratagene) was used for the exogenous expression of c‐myc‐, HA‐ and FLAG‐tagged proteins using a galactose‐inducible promoter. For the in vitro pull‐down assay, a pGEX‐5X‐1 plasmid (GE Healthcare) was used for generating GST fusion proteins. In addition, pET21a and pET15b plasmids (Novagen) were used for hexa‐histidine tagging. Detailed vector constructions are described in Supplementary Experimental Procedures.
Total RNA preparation and northern blotting
Total yeast RNA was obtained using a phenol/freeze RNA preparation methods as previously described (Schmitt et al, 1990). A total amount of 5–10 μg RNA was subjected to general northern blot analysis using digoxigenin (DIG; Boehringer Mannheim; DIG system User's Guide for Filter Hybridization) as previously described (Joo et al, 2009a).
In vitro pull‐down assay with GST fusion proteins
All recombinant proteins used in this study were overexpressed in E. coli (BL21‐CodonPlus (DE3)‐RIL; Stratagene). Bacterial lysates containing GST or GST fusion proteins were applied to 30 μl of glutathione Sepharose 4B resin (Amersham Pharmacia) and incubated for 30 min at room temperature (RT). After beads were washed twice with 1 ml of PBS, other bacterial lysates containing overexpressed His6–Rap1p or its variants were added, followed by incubation with rotation for 2 h at RT. Finally, the binding mixtures were washed three times with PBS. Proteins bound to the GST resins were recovered by boiling for 5 min and subjected to 10% SDS–PAGE, followed by Coomassie brilliant blue (CBB) staining or western blotting.
Preparation of protein lysates, western blotting and Co‐IP
Harvested cells along with PBS washing were resuspended in FA lysis buffer as previously described (Kuras and Struhl, 1999), complemented with protease inhibitors (1 mM PMSF, 1 μg/ml pepstatin A, 5 μg/ml leupeptin and 1 μg/ml aprotinin). All procedures were performed at 4°C. An equal volume of glass beads were added. The cells were lysed by vortexing with a bead beater (MP) for 5 min on ice. Whole‐cell lysates were obtained by centrifugation at 10 000 g for 10 min. An equal amount of total lysate protein was analysed by western blot with specific antibodies against HA (Santa Cruz), c‐myc (Santa Cruz), His6 (Santa Cruz), Pgk1p (Molecular Probes), FLAG (Sigma), Rap1p (Santa Cruz), Esa1p (Santa Cruz) and TAP (OPEN Biosystems). For Co‐IP, 6 mg of protein lysate was incubated with antibodies against the indicated epitopes for 2 h. The immuno‐complexes were immobilized by the addition of 35 μl of protein‐A agarose (Roche). Following incubation for 2 h on a rotator, precipitants were recovered and subjected to 10% SDS–PAGE for further western blotting.
Chromatin DNA preparation, immunoprecipitation and semi‐quantitative PCR (semi‐qPCR) were performed as previously described with some modifications (Joo et al, 2009a). Cells were harvested and lysed by glass beads, as previously described (Hecht and Grunstein, 1999). Chromatin DNA was sheared by sonication, and input lysates containing fragmented chromatin DNA were fractionated by centrifugation at 10 000 g for 10 min. A total protein amount of 3 mg for each input sample was subjected to immunoprecipitation with the indicated antibodies, and 500 μg of total input samples were saved for input control PCR. Semi‐qPCR was performed with 1/40 of the immunoprecipitated DNA and 1/1200 of the total input chromatin DNA as a template. Also included were primers specific for the promoter regions of RPS3, RPS11B, RPL19B, RPL18B, HIS3, ADE3, HIS4, ACT1 and the coding sequence of POL1 (sequences are summarized in Supplementary Table S2).
Supplementary data are available at The EMBO Journal Online (http://www.embojournal.org).
Conflict of Interest
The authors declare that they have no conflict of interest.
This work was supported by FPR05C2‐390 grants.
- Copyright © 2011 European Molecular Biology Organization