Genome-wide expression analysis reveals TORC1-dependent and -independent functions of Sch9

Abstract

The protein kinase Sch9 is proposed to be a downstream effector of TORC1 that is required for activation of ribosome biogenesis and repression of entry into G₀. However, Sch9 apparently functions antagonistically to TORC1, when considering the induction of several stress defence genes that are normally repressed by TORC1. To further investigate the relationship between Sch9 and TORC1, we compared the rapamycin-induced transcriptional responses in an sch9Δ mutant and the isogenic wild type. The data indicate that Sch9 is necessary for proper integration of the rapamycin-induced stress signal, i.e. in sch9Δ cells, typical effects of rapamycin-like repression of ribosomal protein genes and induction of stress response genes are diminished or abolished. Moreover, they reveal for the first time a direct link between Sch9 and nitrogen metabolism. A sch9Δ mutant has an increased basal activation of targets of the general amino acid control pathway and of the nitrogen discrimination pathway, including the ammonium permease MEP2 and the amino acid permease GAP1. The mutant also shows enhanced expression of the transcription factor Gcn4 required for amino acid biosynthesis. Our data favour a model in which (1) the role of Sch9 in the general stress response switches depending on TORC1 activity and (2) Sch9 and TORC1 have independent and additive effects on genes induced upon nitrogen and amino acid starvation.

TOR, Sch9, rapamycin, yeast, stress response, translation, nitrogen sensing

Introduction

Saccharomyces cerevisiae has two homologues of the well-conserved Target of Rapamycin kinases, Tor1 and Tor2, which can both be found in a multiprotein complex TORC1 together with Lst8, Kog1 and Tco89 (Loewith et al., 2002; Wedaman et al., 2003; Reinke et al., 2004; De Virgilio & Loewith, 2006a, b). It is generally assumed that TORC1 couples carbon and nitrogen source availability and quality to cell growth, although a direct link has not yet been established. The immunosuppressant rapamycin specifically inhibits TORC1 and induces a phenotypic response characteristic for starved cells, such as inhibition of ribosome biogenesis and translation and specific changes in the activity of nutrient permeases and in the accumulation of storage carbohydrates (De Virgilio & Loewith, 2006a, b). Many of these responses are, at least in part, regulated by TORC1 at the level of gene transcription. TORC1 was shown to inhibit the activity of Msn2/4, Gln3/Gat1, Gcn4 and Rtg1/3, which are transcriptional activators of genes, respectively, involved in the general stress response, the use of less favourable nitrogen sources, amino acid biosynthesis and the retrograde response (Beck & Hall et al., 1999; Komeili et al., 2000; Valenzuela et al., 2001; Kubota et al., 2003). In addition, TORC1 is known to promote the expression of ribosomal protein (RP) genes through regulation of the transcription factors Sfp1 and Fhl1 and its cofactors Ifh1 and Crf1 (Jorgensen et al., 2004; Marion et al., 2004; Martin et al., 2004).

Like TORC1, the protein kinase Sch9 has been implicated in nitrogen and carbon source-dependent regulation of RP and stress defence genes (Crauwels et al., 1997). Several studies have linked Sch9 to the glucose-sensitive protein kinase A (PKA) pathway. Initially, Sch9 was found to suppress the growth defect resulting from deficient PKA signalling and vice versa (Toda et al., 1988) and it was proposed that Sch9 could act as an upstream modulator of PKA signalling (Crauwels et al., 1997; Trott et al., 2005). However, it has recently become clear that Sch9 and PKA act in parallel, as they were shown to have additive or opposite effects on the given gene targets and that they independently regulate Rim15, a protein kinase that is important for proper entry into G₀ (Reinders et al., 1998; Pedruzzi et al., 2003; Roosen et al., 2005; Swinnen et al., 2006). Sch9 is believed to be the yeast orthologue of the highly conserved PKB/Akt kinase (Geyskens et al., 2000; Sobko et al., 2006), which acts upstream of mTORC1 (Corradetti & Guan et al., 2006). The relationship between TORC1 and Sch9 in yeast, however, remains a matter of debate. It was shown that Sch9 is phosphorylated in vitro by TORC1 and replacement of the phosphorylated residues with Asp/Glu blocks rapamycin-induced inhibition of translation and nuclear import of Rim15 (Reinders et al., 1998; Urban et al., 2007). This result suggests that Sch9 is a major component of TORC1 signalling. Sch9 was also reported to be essential for proper induction of several stress defence genes upon glucose exhaustion or addition of rapamycin (Pedruzzi et al., 2003; Roosen et al., 2005), and it was recently reported that under osmostress conditions, Sch9 is recruited to the chromatin to activate transcription (Pascual-Ahuir & Proft, 2007a, b). These data suggest that Sch9 is essential for proper adaptation to stressful conditions when TORC1 is believed to be inactive, implying that Sch9 acts independent of TORC1. This idea is further supported by a recent finding showing that Sch9 inhibits autophagy independent of TORC1 (Yorimitsu et al., 2007).

In order to obtain new insights into the relationship between TORC1 and Sch9, we used a genome-wide expression analysis. Our data demonstrate that depending on the targets under study, Sch9 acts downstream or in parallel with TORC1, thereby exerting similar or opposite effects when compared with TORC1. Moreover, in particular, we found that sch9Δ cells are characterized by a diminished repression of RP genes and an abolished or diminished induction of typical stress response genes after rapamycin treatment, as compared with the isogenic wild type. In addition, an increased basal expression of several nitrogen discrimination pathway (NDP) and general amino acid control (GAAC) target genes was observed in an sch9Δ strain, independent of TORC1. These data suggest a model where Sch9 has a TORC1-dependent and -independent role in the general stress response, whereas both kinases act in parallel and trigger additive effects on the expression of nitrogen starvation genes and genes required for amino acid biosynthesis.

Materials and methods

Yeast strains, plasmids and growth media

Saccharomyces cerevisiae strains used in this study are listed in Table 1. Deletion strains were prepared using PCR-derived deletion cassettes as described previously (Brachmann et al., 1998). The plasmids p316Mep2-GFP (Marini et al., 2006), pLG670-CYC1∷lacZ and pLG670-UAS_GATA-CYC1∷lacZ (Andre et al., 1995) were kindly provided by Bruno André and the plasmid pME1112 (GCRE6∷lacZ) (Albrecht et al., 1998) was a kind gift from Gerhard Braus. The plasmid used to monitor posttranscriptional and translational control of Gcn4, i.e. p180 (Hinnebusch et al., 1985), was a kind gift from Alan G. Hinnebusch. In the assays described below, yeast strains were grown at 30 °C in rich YP (1% yeast extract; 2% Bacto peptone) or in SC-URA (0.5% ammonium sulphate; 0.17% yeast nitrogen base w/o amino acids and w/o ammonium sulphate; complete amino acid supplement mixture w/o uracil, BIO 101 systems) medium supplemented with 2% glucose.

1

Strains used in this study

Strain	Genotype	References
W303-1A (wild type)	MATa ade2-1 can1-100 his3-11,15 leu2-3/112 trp1-1 ura3-1	Thomas & Rothstein (1989)
TVH301 (JW 00 035)	W303-1A with sch9∷TRP1	Roosen (2005)
JW 01 418	W303-1A with sch9∷NATMX4	This study
JW 01 131	W303-1A with gcn4∷KANMX4	This study
BY4741 (wild type)	MATa his3Δ1 leu2Δ0 met15Δ0 ura3Δ0	YKO Collection
JW 01 306	BY4741 with sch9∷HIS3	Swinnen (2005)

Genome-wide expression analysis

Yeast strains W303-1A and sch9Δ (TVH301) were grown in YPD (1% yeast extract, 2% peptone, 2% glucose) medium supplemented with 50 mg L⁻¹ adenine. Cultures were maintained in the exponential growth phase (OD_{600 nm}<1.5) for a period of at least 24 h by repeated dilution in fresh YPD medium to ensure complete depletion of stationary phase-specific transcripts. Cells were then grown to an OD_{600 nm} of c. 1.5 and a first sample was taken (0 min). Next, rapamycin was added to a final concentration of 200 nM and samples were taken after 15, 30, 60 and 120 min. Total RNA was extracted using RNApure^™ (GeneHunter^® Corporation, catalogue no. P501) according to the protocol supplied by the company. Fluorescent-labelled cDNA was prepared using the CyScribe First-strand cDNA labelling kit (Amersham Biosciences) and used to hybridize the S. cerevisiae DNA microarrays of Eurogentec (AR-SCGS-01, Eurogentec), containing 5803 yeast ORFs. A loop design was applied, with one loop for the W303-1A samples and one for the sch9Δ samples. After overnight hybridization at 37 °C, the microarray was washed, dried and scanned with a GenePix 4100A 01 scanner using genepix pro 5.0.1.24 software. All the arrays in this experiment were outfitted with the Lucidea^™ Universal Scorecard^™ (Amersham Biosciences), generating a series of external controls including 10 calibration spikes (added to the labelling reaction in a ratio of 1 : 1 and spanning up to 4.5 orders of magnitude), eight ratio spikes provided at both low and high concentrations and two negative controls. The entire set of controls was spotted once per pin (32 pins in total). A calibration model was fitted from the measurement data for these external control spikes and used to estimate expression levels for all genes, under all conditions being surveyed (Engelen et al., 2006; Zhao et al., 2007). Probability distributions for the estimated expression levels were obtained by a parametric bootstrap analysis and used to select genes showing significant differential expression between the time profiles for the W303-1A and the sch9Δ strain (α=0.02). These genes were clustered using the AQBC clustering algorithm (significance parameter was set to 0.9) as described by De Smet (2002). To look for significantly overrepresented gene ontologies (GOs) in the different clusters, the online program gostat (http://gostat.wehi.edu.au/; Beissbarth & Speed et al., 2004) was used with the Saccharomyces Genome Database (SGD; http://www.yeastgenome.org/) as the GO-gene association database and the false discovery rate (Benjamini) test to correct for multiple testing. The YEASTRACT database (version August 2007; http://www.yeastract.com/; Teixeira et al., 2006) was used to identify enriched regulatory associations between transcription factors and the genes found in a cluster. If the percentage of genes that was regulated by a certain transcription factor was higher for a certain cluster (group I) compared with the percentage retrieved for all the other genes present on the microarray that are not part of this cluster (group II), Fisher's exact test was used to assess the statistical significance of this overrepresentation of genes regulated by a certain transcription factor in group I.

Northern blot analysis

Yeast strains W303-1A and sch9Δ (TVH301) were prepared for sample taking in the same way as described above for the genome-wide expression analysis. Samples were taken before and after the addition of 200 nM rapamycin at the indicated timepoints. RNA extraction, probe preparation and Northern blotting were performed as described previously (Swinnen et al., 2005). The blots were hybridized with the ³²P-labelled probes consisting of fragments of the desired coding region. The hybridized blots were exposed to a CL-X Posure film (Pierce Biotechnology) for visualization.

Growth assay

Cells were grown overnight in YPD medium. The next morning cells were diluted in fresh YPD medium to an OD_{600 nm} of 0.5 and incubated at 30 °C for 4–6 h. Next, OD_{600 nm} was measured and 5 μL of a 10-fold 1/10 000 dilution series, starting with an OD_{600 nm} of 0.5, was plated out on different media and incubated for 3 days at 30 °C. For the UV-sensitivity assay, cells were spotted on YPD and exposed to 10 mJ cm⁻² UV light (UVP Laboratory products HL-2000 Hybrilinker), before incubation at 30 °C.

Green fluorescence protein (GFP) fluorescence microscopy

W303-1A and sch9Δ (TVH301) cells growing exponentially in YPD medium and expressing GFP-tagged Mep2 were used directly without fixation. Cells were viewed with a Leica DM 4000B fluorescence microscope before and after a 2-h treatment with 200 nM rapamycin. Images were taken with a Leica DFC 420c camera using Leica application suite software.

Galactosidase assay

For the experiment with the Gcn4 response element (GCRE)-driven reporter gene, W303-1A and sch9Δ (TVH301) cells containing the integrative plasmid pME1112 (GCRE6∷lacZ) were grown till the exponential phase (OD_{600 nm}≤1.5) on YPD medium. At OD_{600 nm}c. 1, the first sample was taken and 200 nM of rapamycin was added; after 2 h of rapamycin treatment, the second sample was taken. For the experiment that monitors translational control of a Gcn4–LacZ fusion, wild-type W303-1A cells and isogenic sch9Δ cells (TVH301) were transformed with the plasmid p180. After pregrowth on selective SD-ura medium, the cells were transferred to YPD medium (OD_{600 nm}=0.2) and grown further till the early exponential phase (OD_{600 nm}≤1.5). For the experiment with the GATA-element-driven reporter gene, W303-1A and sch9Δ (TVH301) cells containing pLG670-CYC1∷lacZ or pLG670-UAS_GATA-CYC1∷lacZ were kept in the exponential phase (OD_{600 nm}≤1.5) for a period of at least 24 h in SC-URA. Next, cells were diluted to an OD_{600 nm} of 0.5 in YPD medium and after 2 h of incubation the first sample was taken, after which 200 nM of rapamycin was added. After 2 h of rapamycin treatment the second sample was taken.

Samples were collected, centrifuged and the pelleted cells were dissolved in ice-cold breaking buffer containing proteinase inhibitor cocktail tablets (Complete Mini, EDTA free, Roche Diagnostics) and 1 mM of benzamidine. Cells were broken using glass beads. The clarified extract was diluted in breaking buffer (total volume of 0.5 mL) and incubated at 30 °C for 5 min and the reaction was initiated by adding 0.1 mL of 4 mg mL⁻¹o-nitrophenylgalactoside. At the precise time, the reaction was terminated by addition of 0.25 mL of 1 M Na₂CO₃. The OD_{405 nm} was measured (Thermo Labsystems Multiskan Ascent) and protein concentration was determined according to the Bradford method. Specific activity was calculated by the formula (OD_{405 nm}× 1.7)/(0.0045 × protein concentration × extract volume × time). The specific activity of the empty vector pLG670-CYC1∷lacZ in the W303-1A and the sch9Δ (TVH301) strain or of pME1112 (GCRE6∷lacZ) in the gcn4Δ strain was, respectively, subtracted from the specific activity of pLG670-UAS_GATA-CYC1∷lacZ or pME1112 (GCRE6∷lacZ) in the W303-1A and the sch9Δ (TVH301) strain, giving the actual specific activity (nmol min⁻¹ mg⁻¹ protein). Data presented are the mean of two to three independent experiments. Error bars represent the SD and a t-test was used to test for statistically significant differences in activity.

Western blot analysis

Total protein extracts were prepared from wild-type cells and isogenic sch9Δ cells (TVH301 and JW 01 418) grown to early exponential phase (OD_{600 nm}≤1.5) on rich YPD medium and treated with or without 200 nM rapamycin for 5 h. Extraction procedures, protein quantification and immunodetection were as described previously (see above and Albrecht et al., 1998). The polyclonal antibody directed against Gcn4, i.e. FL-281 raised against full-length Gcn4, was purchased from Santa Cruz Biotechnology Inc. (Santa Cruz, CA).

Results

Genome-wide expression analysis of the rapamycin response in a sch9Δ strain

To gain further insight into the processes controlled by Sch9 and to look for the convergence of Sch9- and TORC1-dependent signalling, we used cDNA microarrays to analyse the expression profiles of a sch9Δ mutant and its isogenic wild-type strain before and after treatment with rapamycin. We thereby focused on genes showing a significant differential expression between the time profiles for the wild type and the sch9Δ mutant. The 1751 genes that met this criterion were subjected to cluster analysis using the AQBC clustering algorithm (De Smet et al., 2002) (Fig. 1). The online tool gostat (Beissbarth & Speed et al., 2004) was used to search for significantly enriched GOs in the clusters generated (see Supporting Information Table S1). To select for major effects, we decided to focus on clusters consisting of at least 50 genes, a property that applied to three clusters (clusters 1, 2a and 3) (Fig. 1, Table S3). However, we added a fourth cluster consisting of only 32 genes (cluster 2b) to our selection, because both the expression profile and the function of the genes in this cluster are similar to those of the genes present in cluster 2a. Furthermore, we used the YEASTRACT database to identify enriched regulatory associations between transcription factors and the genes that were present in the different clusters (Table S2) (Teixeira et al., 2006).

1

Cluster analysis using the AQBC clustering algorithm. (a) Genes were clustered based on their normalized expression levels in the wild type (W303-1A) and the sch9Δ (TVH301) strain at different timepoints (0, 15, 30, 60 and 120 min) after the addition of 200 nM rapamycin. For each cluster, the number of genes in the cluster (NOG) and the mean expression profile of these genes in the wild-type strain (WT) and the strain lacking Sch9 (sch9Δ) are shown. (b) Schematic representation for each cluster, depicting the effect on expression of Sch9 and TORC1. For each cluster, the most important enriched gene ontologies are shown. A more complete list with further specifications can be found in Table S1.

In the following sections, the most important Sch9-dependent effects on the transcription of rapamycin-sensitive genes will be described.

Sch9 is required for the expression of stress defence genes

We previously showed that Sch9 is necessary for proper rapamycin-induced upregulation of several typical stress defence genes, such as HSP12 and HSP26 (Pedruzzi et al., 2003). Furthermore, Sch9 is needed for the activation of several osmostress-responsive genes (Pascual-Ahuir & Proft, 2007a, b). These genes are normally induced after the addition of rapamycin and this is, at least in part, due to the fact that TORC1 no longer inhibits Rim15 and the transcription factors Msn2/4 by keeping them in the cytoplasm (Beck & Hall et al., 1999; Pascual-Ahuir & Proft, 2007a, b). Our analysis revealed many other stress-responsive genes that show a transcription profile similar to HSP12 and HSP26 and together they constitute cluster 1 (Fig. 1a). Many of these genes are involved in DNA damage repair and osmostress defence, including HSP42, DDR2, SSA4 and HOR2 (see Tables S1 and S3 for a complete list). As compared with the wild type, induction of these genes upon rapamycin treatment is slightly reduced in the sch9Δ mutant, but more strikingly, the basal expression level of these genes is reduced. This suggests that Sch9 controls expression of these genes via a mechanism independent of TORC1, as shown in Fig. 1b. Northern blot analysis of DDR2 and SSA3, two typical stress defence genes, further confirmed these profiles (Fig. 2a). In line with these data, we observed that the sch9Δ strain is more sensitive to UV radiation (10 mJ cm⁻²), a known inducer of DNA damage, and to hyperosmotic stress, induced by the addition of KCl to the medium (Fig. 3a).

2

Sch9 effects on gene expression in the presence and absence of active TORC1. (a–e) Northern blot analysis of the rapamycin (200 nM)-induced transcriptional response in the wild type (W303-1A) and the sch9Δ (TVH301) strain of (a) two typical stress defence genes DDR2 and SSA3, (b) glucose-responsive genes HXK1, GLK1 and HXK2, (c) ribosomal protein genes RPS10a and RPL30, (d) two targets of the nitrogen catabolite repression pathway, MEP2 and GAP1, and one target of the general amino acid control (GAAC) pathway SNZ1 and of (e) the central activator of the GAAC GCN4. ACT1 or 18S is used as a loading control. Samples were taken at the indicated time points. (f) Analysis of the localization and expression of Mep2-Gfp, expressed from its endogenous promoter in exponentially growing wild-type (W303-1A) and sch9Δ (TVH301) cells before and after a 2-h treatment with 200 nM rapamycin.

3

Growth of sch9Δ mutants on different media and conditions. Serial dilutions of exponentially growing WT (W303-1A and BY4741) and sch9Δ (TVH301, JW 01 418 and JW 01 306) cells were spotted on (a) YPD, YPD followed by exposure to 10 mJ cm⁻² UV or YPD containing the indicated KCl concentrations, (b) YPD, YPGlycerol or YPAcetate. Results are shown for two different genetic backgrounds: W303-1A and BY4741. For the W303-1A background, two independent sch9Δ mutants, sch9Δ 1 (TVH301) and sch9Δ 2 (JW 01 418), were tested.

Sch9 affects the transition from fermentation to respiration

Previously, we reported that the function of Sch9 function appears to switch depending on the available carbon source. During fermentative growth, Sch9 is needed to maintain phenotypes known to be induced by high PKA activity, while on a nonfermentable carbon source, Sch9 appears to counteract PKA activity (Crauwels et al., 1997). This suggested that Sch9 has an important role in glucose signalling and in the shift from fermentation to respiration. Accordingly, we found that during the diauxic shift and respiratory growth, Sch9 exerts a positive control on target genes of Gis1, a transcriptional activator required during diauxic shift for the induction of stress-responsive genes (e.g. SSA3, SSA4, HSP26, GRE1, …), several of which are found in cluster 1 (Pedruzzi et al., 2000; Roosen et al., 2005).

Apart from typical stress-responsive genes, cluster 1 also contains many genes that are involved in mitochondrial function, such as the tricarboxylic acid (TCA)-cycle genes (e.g. KGD1, KGD2, LSC1, LSC2, etc.), genes involved in fatty-acid metabolism (e.g. POX1, POT1, FOX2, etc.) or mitochondrial RP genes (e.g. MRPL20, MRPL25, MRP20, etc.) (Fig. 1, Tables S1 and S3). These genes encode proteins required during respiration and are known to be upregulated at the diauxic shift (DeRisi et al., 1997). Our finding that their expression is also sensitive to rapamycin treatment is consistent with reported data that revealed high levels of overlap between the carbon stress- and the rapamycin-induced expression profiles (Hardwick et al., 1999; Shamji et al., 2000). Consistent with the involvement of Sch9 in the expression of these genes is the observation that the slow growth phenotype previously ascribed to exponentially growing sch9Δ cells on glucose-containing medium is further aggravated when these cells are transferred to media with nonfermentable carbon sources (Fig. 3b).

Two genes whose expression is tightly linked to the presence of glucose are HXK1 and GLK1. They encode for two of the three enzymes in yeast that can phosphorylate glucose, although expression of HXK1 and GLK1 is repressed in the presence of glucose and derepressed when cells are shifted to a nonfermentable carbon source (De Winde et al., 1996; Rodriguez et al., 2001). HXK1 was present in cluster 1, whereas GLK1 was not withheld for cluster analysis due to the low signal on the microarray. Nonetheless, Northern blot analysis showed that induction of both genes upon rapamycin treatment is delayed and that their basal expression is strongly dependent on Sch9 (Fig. 2b). HXK2, encoding for the third enzyme that can phosphorylate glucose in yeast, is typically induced in the presence of glucose and repressed after glucose exhaustion, i.e. the opposite transcriptional regulation of HXK1 and GLK1 (De Winde et al., 1996). Interestingly, HXK2 can be found in cluster 2A (Fig. 1a; Table S3), which contains genes that are repressed after rapamycin addition, but this repression is diminished and decreased in the sch9Δ strain, in comparison with the wild type. For HXK2, Northern blot analysis only confirmed the presence of a slight delay in the rapamycin-induced repression in the sch9Δ strain, but not a difference in the total level of repression (Fig. 2b). This diminished repression of HXK2, however, might be coupled to the delayed and diminished induction of HXK1 and GLK1, as it was shown that Hxk2 is involved in glucose-induced repression of HXK1 and GLK1 (Rodriguez et al., 2001).

Sch9 controls the translational machinery

Transcription of the RP genes and, to a lesser extent, the Ribi genes, which are non-RP genes required for proper functioning of the translational machinery, have been reported to be under the positive control of Sch9 (Jorgensen et al., 2004; Roosen et al., 2005; Urban et al., 2007). The major part of the RP genes present in the yeast genome and also several Ribi genes were contained in clusters 2a and 2b. Cluster 2b comprised RP genes that, in line with previous reports, show a strong reduction in their basal expression level in the sch9Δ strain as compared with the wild-type strain, while cluster 2a contains RP genes with a similar basal expression level in the sch9Δ mutant and the wild-type strain. Interestingly, gostat analysis revealed an enrichment of RP genes encoding proteins involved in ribosomal subunit assembly (GO:0042257) in cluster 2b compared with cluster 2a, indicating that transcription of RP genes involved in ribosomal subunit assembly is possibly more dependent on Sch9. In order to find differences in regulatory associations between the two groups of RP genes, we used the YEASTRACT database (Teixeira et al., 2006). However, this only revealed that both the genes in cluster 2a as well as cluster 2b show a strong regulatory association with the transcription factors Sfp1, Fhl1, Ifh1 and Rap1, which are all well-known regulators of RP genes (Table S2).

Upon addition of rapamycin, expression of the RP and Ribi genes in cluster 2a and 2b rapidly declines in the wild-type strain. In the sch9Δ strain, however, this rapamycin-induced decline was less pronounced as compared with the wild-type strain. The difference in response between the wild-type strain and the sch9Δ mutant was further confirmed by Northern blot analysis (Fig. 2c).

In line with previous reports (Jorgensen et al., 2004; Urban et al., 2007), these data demonstrate that TORC1-dependent control of RP and Ribi genes involves not only an Sch9 signalling branch but also an Sch9-independent regulatory circuit. Most probably, other regulators, which act in parallel to Sch9 and some possibly also independent of TORC1, compensate for the loss of Sch9. This would then counteract the rapamycin-induced downregulation of RP and Ribi genes in the sch9Δ strain and, furthermore, explain the absence of a significant difference in the basal expression level of many RP genes in the mutant in comparison with the wild type. In this context, it is important to note that a previous report described the recovery of RP gene expression, after an initial decline, only 90 min after the inactivation of Sch9 (Jorgensen et al., 2004), and that our data also reveal a similar recovery of RP gene expression after 2 h of rapamycin treatment in the wild type as well as the sch9Δ mutant.

Genes activated during nitrogen limitation are upregulated in a sch9Δ strain

Yeast cells have a highly versatile nitrogen metabolism that allows the usage of a wide range of nitrogen sources and is tightly regulated by the nitrogen content of the medium (Magasanik & Kaiser et al., 2002; Hinnebusch et al., 2005; Godard et al., 2007). When preferred rich nitrogen sources become limiting, which is mimicked by the addition of rapamycin, target genes of the NDP will be derepressed to allow optimal scavenging and usage of alternative, poorer nitrogen sources. Furthermore, the GAAC pathway will be activated to induce the expression of a large number of genes involved in amino acid biosynthesis.

Cluster 3 is enriched with genes involved in nitrogen metabolism (Tables S1 and S3). Most of these genes are well-known target genes of the NDP (e.g. MEP2, GLT1, …) and/or the GAAC pathway (e.g. SNZ1, LYS21, …) (Natarajan et al., 2001; Scherens et al., 2006). Accordingly, we found enriched transcription factor–target gene associations between the genes in cluster 3 and Gln3 and Gcn4, which are central transcriptional activators of, respectively, the NDP and the GAAC pathway (Table S2). The basal expression levels of these NDP and GAAC target genes found in cluster 3 are increased in the sch9Δ strain, whereas upregulation of these genes after the addition of rapamycin is independent of Sch9, as was reported before (Urban et al., 2007).

We confirmed the observed expression profiles of MEP2 and SNZ1, which are both members of cluster 3, by Northern blot analysis (Fig. 2d). MEP2 is an NDP target gene coding for a high-affinity ammonium permease (Marini et al., 1997), while SNZ1 is a GAAC target gene coding for an enzyme involved in the synthesis of vitamin B6, an important cofactor for many enzymes involved in amino acid biosynthesis and also has antioxidative properties (Braun et al., 1996; Stolz & Vielreicher et al., 2003). Upregulation of MEP2 was also observed when the sch9Δ strain was transformed with a plasmid that contained MEP2-GFP behind the endogenous MEP2 promoter (Fig. 2f). Furthermore, this experiment also shows that Mep2 is properly targeted to the plasma membrane in the sch9Δ strain. In order to test whether only ammonium transport or nitrogen transport in general is affected in the sch9Δ strain, we also examined the expression of GAP1, another well-known NDP target gene. Although this gene was not present in cluster 3, both microarray and Northern blot analyses showed that GAP1 was upregulated in the sch9Δ strain (Fig. 2d, Table S3).

Next, because NDP target genes are known to be regulated by transcription factors that bind to the GATA-elements in their promoter, we examined the role of Sch9 in transcription driven by the upstream activation site UAS_GATA, a cluster of 5′-GAT(A/T)AG-3′ previously identified upstream of the GABA-inducible UGA4 gene. Regulation of this element was shown to be complex with several activators (Gln3, Gat1) and repressors (Uga43, Gzf3), and a shift from a rich to a poor nitrogen-source-containing medium only slightly induces expression driven by this element (Andre et al., 1995; Soussi-Boudekou et al., 1997). Accordingly, the addition of rapamycin, which is believed to mimic nitrogen stress, only moderately activates expression of this element in our wild-type strain background (Fig. 4a). In contrast, we found that loss of Sch9 function dramatically increased the UAS_GATA-supported transcription (Fig. 4a), which clearly indicates that Sch9 affects GATA-driven expression independent of TORC1. We also tested whether Sch9 affects the expression of a lacZ-reporter gene that contained six GCREs in its promoter (Albrecht et al., 1998). As shown, the expression of the LacZ reporter was more than doubled in the sch9Δ mutant cells as compared with wild-type cells. Note that the LacZ activities were very low, probably explaining why the difference between the wild type and the sch9Δ mutant was only marginally significant (Fig. 4b). Upon rapamycin treatment, both strains displayed high LacZ activities and once more, the lacZ levels were on average slightly higher in the sch9Δ mutant as compared with the wild-type strain. However, in this case the difference appeared not to be statistically relevant.

4

Sch9 controls GATA- and GCRE-driven gene expression. (a) Expression of the UAS_GATA-CYC1-lacZ reporter gene carried on a low-copy-number plasmid was monitored in wild-type (W303-1A; open bars) and sch9Δ (TVH301; closed bars) cells before and after treatment with 200 nM rapamycin for 2 h. The UAS_GATA is derived from the UGA4 gene and exists out of cluster of 5′-GAT(A/T)A-3′ motifs (Andre et al., 1995). Error bars represent the SD of the measurement of two independent transformants and activities were corrected for the average activity in the different strains of the CYC1-lacZ gene without the upstream UAS. (b) Wild-type (W303-1A; open bars) and sch9Δ (TVH301; closed bars) cells were transformed with pME1112, an integrative plasmid containing a Gcn4-dependent GCRE-lacZ reporter gene (Albrecht et al., 1998). Expression of this reporter gene was determined for exponentially growing cells before and after treatment with 200 nM rapamycin for 2 h. Error bars represent the SD of three independent experiments and activities were corrected for the average activity of the reporter gene in the gcn4Δ (JW 01 131) strain. (c) Wild-type (W303-1A; open bars) and sch9Δ (TVH301; closed bars) cells were transformed with p180, a plasmid containing the GCN4 leader sequence and GCN4–LacZ fusion allowing one to monitor translational control (Hinnebusch et al., 1985). Expression of this reporter gene was determined for exponentially growing cells before and after treatment with 200 nM rapamycin for 6 h. Error bars represent the SD of three independent experiments. (d) Expression of Gcn4 in wild-type and gcn4Δ cells or sch9Δ cells (JW 01 418) with or without treatment with rapamycin as determined by Western blot analysis. For the β-galactosidase assays, the significant differences between wild-type and sch9Δ cells are denoted with asterisks (^*P<0.05; ^**P<0.01; ^***P<0.001).

Because the GCRE is bound by Gcn4, the central activator of GAAC, we examined in more detail the expression of this protein. The gene encoding Gcn4 is found in cluster 3 and Northern blot analysis confirmed that this gene is indeed upregulated in a sch9Δ strain (Fig. 2e), which is in line with a potential role for Sch9 in GCRE-driven expression. We next monitored whether the deletion of SCH9 would also lead to increased translation of GCN4 (Hinnebusch et al., 2005). To this end, we first used the p180 plasmid described previously by Hinnebusch (1985) where the four small ORFs, which are present in the GCN4 leader and are involved in translational control, precede a GCN4–LacZ fusion. Upon measurement of the β-galactosidase activities, it became clear that the deletion of SCH9 led to a partial derepression and about twice as much GCN4–LacZ enzyme levels when compared with the wild-type cells (Fig. 4c). Interestingly, higher β-galactosidase activities were also observed in the sch9Δ mutant after rapamycin treatment, indicating that Sch9 mediates expression regulation of Gcn4 independent of TORC1. Consistently, increased Gcn4 levels could also be detected in the sch9Δ mutant cells (TVH301 and JW 01 418) by Western blot analysis using immunodetection with polyclonal antibodies (Fig. 4d and data not shown).

Discussion

In this study, we investigated the relationship between Sch9 and TORC1 using genome-wide expression analysis. In general, our data suggest that (1) Sch9 function switches depending on TORC1 activity, which is needed for proper fine tuning of RP and Ribi gene expression as well as stress response genes to changes in the growth conditions (Fig. 5a), and (2) Sch9 controls important components of yeast nitrogen metabolism independent of TORC1, demonstrating that Sch9 has a broader role in this process than previously thought (Fig. 5b).

5

TORC1-dependent and -independent functions of Sch9. (a) Under favourable conditions, TORC1-phosphorylated Sch9 activates the expression of genes involved in translation, while under unfavourable conditions, when TORC1 is inactive, Sch9 is necessary for the expression of stress defence genes and genes involved in respiration. (b) Sch9 and TORC1 exert, at least partly, independent and additive control on genes typically induced upon nitrogen starvation. Active and inactive interactions are, respectively, shown by bold black and grey lines. Arrows and bars represent, respectively, positive and negative interactions. Dashed lines represent putative interactions. See text for details.

TORC1-dependent switch in Sch9 function?

In this study, we found that a major part of the RP genes in yeast are downregulated in a sch9Δ strain. A role for Sch9 as a transcriptional activator of RP genes was reported before (Jorgensen et al., 2004; Roosen et al., 2005) and the mechanistic and regulatory aspects hereof are starting to become elucidated. Under favourable conditions, TORC1 phosphorylates Sch9, thereby directing the kinase to stimulate RP gene expression (Urban et al., 2007). The fact that for many RP genes no significant difference in the basal expression level between wild-type and sch9Δ cells was observed is probably due to a compensatory effect of other Sch9-independent regulators of RP genes, which might also explain the diminished rapamycin-induced repression of these genes in the sch9Δ strain. Apart from controlling genes involved in translation, the TORC1-Sch9 effector branch was also proposed to inhibit entry into G₀ by preventing nuclear localization of Rim15 (Urban et al., 2007). Accordingly, one would expect increased expression of Rim15 targets, such as the stress defence genes, in sch9Δ cells. Nevertheless, we observed a diminished induction after rapamycin addition of these genes and, more strikingly, a reduced basal expression level, which indicates that Sch9 is also a transcriptional activator of these genes. Hence, the data indicate that Sch9 influences transcription independent of Rim15 or TORC1. This is in line with our previous findings showing a requirement of Sch9 independent of Rim15 for induction of several target Msn2/4 and Gis1 target genes upon glucose exhaustion or rapamycin addition (Pedruzzi et al., 2003; Roosen et al., 2005). In line with this, it was recently reported that under conditions of osmotic stress, Sch9 itself acts as a chromatin-associated activator of osmostress-responsive genes (Pascual-Ahuir & Proft, 2007a, b).

Interestingly, the role of Sch9 in nutrient and stress-induced signalling was reported to switch depending on the environmental conditions (Crauwels et al., 1997; Piper et al., 2006). For instance, when grown on a nonfermentable carbon source, sch9Δ cells have phenotypic characteristics that point to a higher activity of the glucose-responsive PKA signalling pathway as compared with the wild type, while on a fermentable carbon source, the situation was reversed (Crauwels et al., 1997). Additionally, the increased chronological longevity that is usually associated with loss of Sch9 function (Fabrizio et al., 2001) is also dependent on the carbon source, because sch9Δ cells that are pregrown on respiratory media have a decreased longevity (Piper et al., 2006). In line with this, we found that many genes whose expression levels are linked to the presence of glucose, such as HXK1, GLK1 and HXK2 (De Winde et al., 1996; Rodriguez et al., 2001), are aberrantly regulated after the loss of Sch9 function. Apparently, the fermentable carbon source starvation signal, which is mimicked by the addition of rapamycin (Hardwick et al., 1999; Shamji et al., 2000), is not properly transduced in sch9Δ cells. This is further reflected by the diminished expression levels in the sch9Δ strain of genes involved in respiration, such as TCA cycle genes, genes involved in fatty acid metabolism and mitochondrial RP genes. These genes are normally induced upon glucose exhaustion when cells shift to respiratory metabolism.

Taken together, the above-mentioned results suggest that Sch9 has a dual role in the general stress response, which is consistent with previous reports (Pedruzzi et al., 2003; Roosen et al., 2005; Pascual-Ahuir & Proft, 2007a, b). Our data suggest that Sch9 function apparently switches depending on TORC1 activity. Under favourable conditions, the TORC1-dependent function of Sch9 will be predominant and the TORC1-phosphorylated Sch9 then stimulates translation while preventing the expression of Rim15-targets such as stress defence genes. When conditions become less favourable, like on exhaustion of the fermentable carbon source, TORC1 will no longer be active and under these conditions Sch9 is necessary for proper activation of stress defence genes and genes needed for respiration.

Sch9 as a regulator of nitrogen metabolism

It was recently reported that the slow growth phenotype and the genetic instability of an sch9Δ strain could partially be rescued by additional deletion of GLN3 and GAT1, the central activators of the NDP (Urban et al., 2007). The latter was observed on rich YPD medium, i.e. conditions under which the Gln3 and Gat1 are normally inactive. Therefore, this result suggested that NDP target genes are aberrantly regulated in the sch9Δ strain. Accordingly, our data now demonstrate that the loss of Sch9 function indeed results in genome-wide expression profiles characteristic of nitrogen-starved cells, including the upregulation of many targets of the NDP and of the GAAC pathway. Importantly and consistent with previously reported data (Urban et al., 2007), we observed no difference in the rapamycin-mediated upregulation of NDP or GAAC genes between the wild type and the sch9Δ strain, suggesting that TORC1 and Sch9 operate in parallel and have additive effects when it comes to regulation of these genes.

The NDP target genes are controlled by several GATA transcription factors that bind to 5′-GATA-3′ core sequences and activate (Gln3 and Gat1) or repress (Uga43 and Gzf3) gene expression (Magasanik & Kaiser et al., 2002). We found that expression driven by UAS_GATA (Andre et al., 1995), consisting of a cluster of 5′-GAT(A/T)AG-3′ sequences, was strongly increased in the sch9Δ strain independent of TORC1. This suggests that Sch9 regulates one of the GATA transcription factors and it would be worthwhile to study this in more detail.

Recently, it was shown that GCN4, the central activator of GAAC, is also a target of the NDP (Godard et al., 2007). Accordingly, we found that in sch9Δ cells GCN4 expression was increased, as compared with wild-type cells. This suggested that upregulation of the GAAC target genes in sch9Δ cells might be mediated via the Sch9-dependent regulation of Gcn4. Consistently, we observed a modest but significant increase in the sch9Δ strain as compared with the wild type for GCRE-driven gene expression, which depends on Gcn4 (Albrecht et al., 1998). We also showed that the upregulation of GCN4 at the transcriptional level is paralleled by an increased Gcn4 activity in the sch9Δ strain. Most interestingly, the difference in expression of Gcn4 between the sch9Δ mutant and the wild-type strain was still observed when both strains were treated with rapamycin, suggesting that the mechanisms for Gcn4 expression control elicited by Sch9 are, at least in part, independent from those regulated through TORC1. Rapamycin and TORC1 were found to affect Gcn4 expression by translational control via the protein kinase Gcn2 but several lines of evidence pointed towards additional regulatory mechanisms affecting Gcn4 expression (Hinnebusch et al., 2005). One such mechanism involves the Ras-cAMP pathway, which appears to trigger Gcn4 activation in response to glucose stimulation and UV irradiation (Engelberg et al., 1994; Marbach et al., 2001). We postulate that Sch9 would act in concert with the Ras-cAMP pathway, because the Ras-cAMP and Sch9 signalling routes are closely interconnected as described previously (Winderickx et al., 2003; Roosen et al., 2005).

Included among the upregulated NDP targets were also the genes encoding for the high-affinity ammonium permease Mep2 and the general amino acid permease Gap1. These permeases are necessary for the rapid activation of trehalase, the repression of stress defence genes and the activation of RP genes after the addition of ammonium (Mep2) or a high-quality amino acid (Gap1) to nitrogen-starved cells (Donaton et al., 2003; Van Nuland et al., 2006). Interestingly, Sch9 was also shown to be essential during this process (Crauwels et al., 1997). These results suggest that a negative feed-back loop exists in which nitrogen source-activated Sch9 downregulates the expression of MEP2 and GAP1. Expression of these genes is no longer required when nitrogen is abundant, because ammonium is possibly transported by diffusion when present in high concentrations (Marini et al., 1997), and more specific amino acid permeases are expressed for uptake of the amino acids present in the medium (Beck & Hall et al., 1999).

In conclusion, our data clearly demonstrate that Sch9 affects key components in nitrogen metabolism. Loss of Sch9 function seems to result in a nitrogen-starved phenotype, to which cells respond by the induction of genes that allow maximal usage of the (scarce) nitrogen sources present, such as Mep2 and Gap1. Moreover, also, this function of Sch9 seems to be independent of TORC1.

Conclusions

Here we demonstrated that although Sch9 function is probably partly controlled by TORC1, Sch9 is more than just a downstream effector of TORC1 signalling and that Sch9 function switches depending on the environmental conditions. Because Sch9 has been localized at the vacuolar membrane (Jorgensen et al., 2004; Urban et al., 2007), but is also recruited to the nucleus (Pascual-Ahuir & Proft, 2007a, b), it might be possible that these different Sch9 functions are exerted by different pools of Sch9. Furthermore, the presence of TORC1-independent phosphorylation sites in Sch9 indicates that other modulators of Sch9 exist (Urban et al., 2007), designating Sch9 as an integration point of different environmental stimuli. For example, the phytosphingosine-activated PDK1 orthologues, Pkh1 and Pkh2, also phosphorylate Sch9 in vitro (Liu et al., 2005). These proteins might modulate Sch9 function in response to heat stress, as the intracellular concentration of sphingolipid precursors, such as phytosphingosine, is known to be responsive to stress (Dickson et al., 2006). Therefore, it would be of interest to monitor the effects of Sch9 inactivation on different media and under different conditions, as this might reveal other unknown functions of Sch9-dependent signalling.

Acknowledgements

This work was supported by funds of K.U. Leuven and FWO-Vlaanderen.

Supporting Information

Table S1. Enriched GOs in the different clusters identified with the AQBC clustering algorithm.

Table S2. Enriched regulatory associations between transcription factors and the genes in the different clusters.

Table S3. The differences in the genome-wide transcriptional response to rapamycin of wild type and sch9D cells.

Please note: Blackwell Publishing is not responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.

Author notes

Editor: Terrance Cooper