Nitrogen Catabolite Repression-Sensitive Transcription as a Readout of Tor Pathway Regulation: The Genetic Background, Reporter Gene and GATA Factor Assayed Determine the Outcomes
Isabelle Georis, André Feller, Jennifer J. Tate, Terrance G. Cooper, Evelyne Dubois

This article has a correction. Please see:

Abstract

Nitrogen catabolite repression (NCR)-sensitive genes, whose expression is highly repressed when provided with excess nitrogen and derepressed when nitrogen is limited or cells are treated with rapamycin, are routinely used as reporters in mechanistic studies of the Tor signal transduction pathway in Saccharomyces cerevisiae. Two GATA factors, Gln3 and Gat1, are responsible for NCR-sensitive transcription, but recent evidence demonstrates that Tor pathway regulation of NCR-sensitive transcription bifurcates at the level of GATA factor localization. Gln3 requires Sit4 phosphatase for nuclear localization and NCR-sensitive transcription while Gat1 does not. In this article, we demonstrate that the extent to which Sit4 plays a role in NCR-sensitive transcription depends upon whether or not (i) Gzf3, a GATA repressor homologous to Dal80, is active in the genetic background assayed; (ii) Gat1 is able to activate transcription of the assayed gene in the absence of Gln3 in that genetic background; and (iii) the gene chosen as a reporter is able to be transcribed by Gln3 or Gat1 in the absence of the other GATA factor. Together, the data indicate that in the absence of these three pieces of information, overall NCR-sensitive gene transcription data are unreliable as Tor pathway readouts.

THE central position of second and third generation rapamycin derivatives in clinical settings has greatly stimulated investigation of its target, Tor, and the mechanisms through which Tor participates in the regulation of cellular processes. One of the most formative discoveries of these investigations was the finding that rapamycin induced nuclear localization of Gln3 and activation of nitrogen catabolite repression (NCR)-sensitive transcription under repressive growth conditions where this activator would normally be restricted to the cytoplasm and the transcription of NCR-sensitive genes (e.g., DAL5, MEP2, GAP1, etc.) would be quiescent (Figure 1A) (Beck and Hall 1999; Cardenas et al. 1999; Hardwick et al. 1999; Bertram et al. 2000; Cox et al. 2000; Shamji et al. 2000). By inference, Gln3-mediated activation itself has become a prominent reporter of Tor pathway function. In the wild, Saccharomyces cerevisiae uses NCR to selectively utilize repressive nitrogen sources (e.g., glutamine or ammonia as nitrogen source) in preference to derepressive sources that support less robust growth (reviewed in Hofman-Bang 1999; Cooper 2002, 2004; Magasanik and Kaiser 2002). In nitrogen excess, transcription of genes required to transport and degrade poor nitrogen sources is repressed. When preferred sources are limited or absent, transcription of these NCR-sensitive genes is derepressed so that the cell can scavenge alternative nitrogen sources that might be available in its environment.

A highly abridged version of the regulatory model reported for the Tor pathway and regulation of GATA-factor-activated, NCR-sensitive gene expression is shown in Figure 1A. Although central components of this model have not stood up to detailed investigation (Cox et al. 2002, 2004; Wang et al. 2003; Tate et al. 2005, 2006,a,b, 2009; Feller et al. 2006; Kulkarni et al. 2006; Yan et al. 2006; Georis et al. 2008), it will adequately frame the experiments developed in this work. When nitrogen is in excess, Sit4 protein phosphatase is in a complex with Tor-associated protein (Tap42) and Tor complex 1 (TorC1) (Figure 1A) (Di Como and Arndt 1996; Beck and Hall 1999; Jiang and Broach 1999; Bertram et al. 2000; Loewith et al. 2002; Düvel et al. 2003; Wang et al. 2003; Reinke et al. 2004; Giannattasio et al. 2005; Di Como and Jiang 2006; Yan et al. 2006; Adami et al. 2007; Aronova et al. 2007). In this form, it is inactive and hence incapable of dephosphorylating Gln3. Under these growth conditions, Gln3 remains cytoplasmic and NCR-sensitive gene expression is repressed. If nitrogen becomes limiting or glutamine-grown cells are treated with the Tor-specific inhibitor rapamycin, the Sit4-Tap42 complex is released from TorC1 and thus becomes active. Gln3 can now be dephosphorylated, enter the nucleus, and activate NCR-sensitive catabolic gene expression. Gat1, the second NCR-sensitive transcription activator, responds to nitrogen levels and rapamycin similarly to Gln3 (Coffman et al. 1995, 1996a,b; Stanbrough et al. 1995). However, it has been recently shown that Tor pathway regulation of Gln3 and Gat1 bifurcates at the level of their localization and DNA binding (Figure 1A): (i) Ure2 plays a much more important role in restricting Gln3 to the cytoplasm during nitrogen excess than it does for Gat1, and (ii) Sit4 is required for nuclear localization of Gln3 but not of Gat1 following rapamycin treatment (Georis et al. 2008). Once in the nucleus, Gln3 and Gat1 bind to their target GATA sequences in the promoters of NCR-sensitive genes and activate their transcription. For DAL5, Gat1 binding is Gln3 independent, whereas Gln3 binding is Gat1 dependent (Figure 1A) (Georis et al. 2008).

NCR-sensitive gene expression is further regulated by the finely tuned and integrated action of the two transcription activators discussed above and two transcription repressors (Dal80/Uga43 and Gzf3/Deh1/Nil2) (Figure 1B) (Coffman et al. 1995, 1996a,b, 1997; Stanbrough et al. 1995; Rowen et al. 1997; Soussi-Boudekou et al. 1997; Scherens et al. 2006). While Gln3 (Cunningham et al. 1996), and presumably Gat1, binds to single GATA sequences, the GATA repressors contain leucine-zipper motifs near their C termini and form homo and heterodimers in vivo (Svetlov and Cooper 1998). Consistent with the formation of dimers, Dal80 binding to DNA requires two GATA sequences that are properly spaced and oriented (Cunningham and Cooper 1993; Cunningham et al. 1994). Gzf3 also binds, in a GATA sequence-dependent manner, to DNA fragments (UGA4, GAP1) containing two or more GATAs (Coffman et al. 1997). With these DNA fragments, multiple protein-DNA species were observed with the one possessing the highest electrophoretic mobility being the most prominent. Gzf3 has also been suggested as regulating transcription from a single GATA element (Rowen et al. 1997). Integration of the actions of GATA activators and repressors is achieved by their cross and autogenous regulation (Figure 1B) (Daugherty et al. 1993; Coffman et al. 1996a,b, 1997; Rowen et al. 1997; Soussi-Boudekou et al. 1997; Cunningham et al. 2000a,b). Three of the four GATA factors (Gat1, Dal80, Gzf3/Deh1) are encoded by genes containing multiple GATA sequences in their promoters and their expression is NCR sensitive (Coffman et al. 1997; Soussi-Boudekou et al. 1997). DAL80 and GAT1 expression is strongly Gln3 dependent as well as Dal80 and Gat1 regulated. On the other hand, regulation by Gzf3 is very weak (Figure 1B). GZF3 expression is NCR sensitive and Dal80 regulated and requires at least one of the two GATA activators (Figure 1B ). The most striking characteristic of Gzf3 is that its regulation of NCR-sensitive genes occurs in a repressive medium (Coffman et al. 1996a,b, 1997; Rowen et al. 1997; Soussi-Boudekou et al. 1997).

Figure 1.—

Current models describing Tor pathway regulation of Gln3 dephosphorylation, localization, and activation of NCR-sensitive transcription (A) and interactions of the positively and negatively-acting GATA factors that fine-tune the levels of NCR-sensitive transcription (B). The model incorporates conclusions drawn from many laboratories (see text for references). Green lines and arrows indicate positive regulation, whereas red lines, arrows, and bars indicate negative regulation. In A, red and green letters indicate inactive and active forms of the proteins, respectively. Dashed lines indicate weak regulatory relationships. The model in A is not exhaustive—for example, the Tip41 protein (Jacinto et al. 2001) is not described—and does not present all contending views of some steps, for example, those reported by the Beck and Hall (1999) vs. Carvalho et al. (2001) or Jacinto et al. (2001) vs. Jiang and Broach (1999). A is modified from Tate et al. (2009) and B is modified from Coffman et al. (1997).

During investigation of the bifurcated regulation of Gln3 and Gat1 by the Tor pathway discussed above, a surprising observation was made. The Gat1 requirement for rapamycin-induced DAL5 gene expression was much greater than that for Gln3 (Georis et al. 2008). In fact, DAL5 transcription decreased only about threefold in a gln3Δ and was not affected in a sit4Δ, in which Gln3 is cytoplasmic. This contradicted a large body of earlier work demonstrating just the opposite, i.e., that the Gln3 requirement of NCR-sensitive DAL5 expression was much greater than that for Gat1 (Daugherty et al. 1993; Coffman et al. 1994, 1996a,b, 1997; Tate et al. 2002; Tate and Cooper 2003). Therefore, experiments described in this work were initiated to resolve the contradiction.

We were able to effectively explain the apparent paradox surrounding the GATA factor requirements of DAL5 expression. In doing so, however, we found that comparing conclusions about Tor pathway function derived from analyses employing NCR-sensitive reporter gene or transcriptome assays may be complicated by the genetic background of the strains analyzed and the specific NCR-sensitive gene assayed. For example, whether or not Sit4 is required for NCR-sensitive transcription of a reporter gene depends on whether GATA repressor Gzf3 effectively functions in the strain assayed and whether or not Gat1 or Gln3, in the absence of the other GATA factor, is able to activate transcription of the particular NCR-sensitive gene assayed. We also discovered (i) a new regulatory function for GATA repressor Gzf3, (ii) evidence pointing toward the existence of one or more Tor pathway components acting downstream of rapamycin-induced nuclear GATA factor localization, and (iii) a potential explanation for variations in the sets of genes identified as NCR-sensitive by transcriptome analyses performed in different laboratories.

MATERIALS AND METHODS

Strains and culture conditions:

S. cerevisiae strains used in this work are listed in Table 1 along with the DNA primers used in their construction. Growth conditions were identical to those described in Tate et al. (2006a,b), Scherens et al. (2006), and Georis et al. (2008). Rapamycin (Sigma and LC Laboratories) was used at the concentrations and times indicated in the figure legends.

View this table:
TABLE 1

Strains used in this work

Strain construction:

Deletion strains, involving insertion of kanMX or natMX cassettes, were constructed as described earlier (Tate et al. 2006a,b; Georis et al. 2008). Chromosomal GLN3 or GAT1 were tagged at their C termini with 13 copies of the Myc epitope (Myc13) as described in Georis et al. (2008).

Steady-state mRNA and DNA-binding analyses:

Quantitative RT–PCR (qRT–PCR) and chromatin immunoprecipitation (ChIP) analyses were performed as described by Georis et al. (2008). Data were analyzed with LightCycler software, version 5.32. The immunoprecipitated DNA/input DNA (IP/IN) ratio corresponds to the concentration of target DNA in the IP sample relative to that in the corresponding IN sample, multiplied by 10. IP/IN values obtained for the unbound control (DAL5U) were substracted from IP/IN values obtained for the DAL5 promoter (DAL5P). To counterbalance variation generated by the immunoprecipitation step, we treated all of our data as follows. The wild-type-induced value was set as 1 and the IP/IN value of every simultaneously immunoprecipitated sample was normalized accordingly. For every independent culture, the mean of the IP/IN ratios for one to three replicate immunoprecipitations was calculated. Values in Figures 4, 5, 9, and 11 correspond to the mean IP/IN value of at least two independent cultures. DAL5U, DAL5O, DAL5P, and TBP1O primers have been described in Georis et al. (2008). ChIP primers amplified a 126-bp region in the promoter of DAL7 (DAL7P1: 5′-AAATCTCCGCTGAAAGTTGC-3′; DAL5P2: 5′-TTTCACGATGTACCTTATCCAAGA-3′) or a 141-bp region in the promoter of GAP1 (GAP1P1: 5′-GACCTCATGCAGCAAAGTCA-3′; GAP1P2: 5′-CCGGTTGCTCCAGAAGATAA-3′). qRT–PCR primers amplified a 122-bp region in DAL7 (DAL7O1: 5′-CACGGAAACAGCTTTTAGCC-3′; DAL7O2:5′-AGCACCTTGCCATGTAGGAT-5′), a 147-bp region in GAP1 (GAP1O1: 5′-AAATGGCTCCGCTGTTTCTA-3′; GAP1O2: 5′-GGAGTTTGGGCAGTGATGAT-3′), and a 167-bp region in GZF3 (GZF3O1: 5′-TTATGGCATCGCAGGCTACA-3′; GZF3O2: 5′-TTGCTCGGCAGATACTGCTT-3′).

Indirect immunofluorescence microscopy:

Indirect immunofluorescence microscopy was performed as described in Georis et al. (2008). Cells were imaged at room temperature using a Zeiss Axioplan 2 imaging microscope with a ×100 Plan-Apochromat 1.40 oil objective at room temperature. Images were acquired using a Zeiss Axio camera and AxioVision (Zeiss) software, processed with Adobe Photoshop and Illustrator programs. Gamma settings were altered where necessary to avoid any change or loss in cellular detail during processing; changes were applied uniformly to the image presented.

Determination of intracellular Gln3-Myc13 and Gat1-Myc13 distribution:

To provide more representative and complete descriptions of Gln3-Myc13 and Gat1-Myc13 localization than can be obtained from an isolated image of a few cells, we manually scored Gln3-Myc13 or Gat1-Myc13 localization in ≥200 cells in multiple, randomly chosen microscopic fields from which each image presented was taken. Cells were classified into one of three categories: cytoplasmic (cytoplasmic fluorescence only), nuclear cytoplasmic (fluorescence appeared in the cytoplasm as well as colocalizing with DAPI-positive material), and nuclear (colocalizing only with DAPI-positive material). Limitations placed on interpretations of these descriptions have been described in detail in earlier publications (Tate et al. 2006a,b, 2009; Georis et al. 2008).

RESULTS

The requirements of Gln3 and Sit4 for rapamycin-induced DAL5 transcription depends on the genetic background:

Recent studies, analyzing the effects of deleting the SIT4 gene on GATA factor localization, DNA-binding and transcriptional activation revealed a quite surprising and unsettling result. Rapamycin-induced DAL5 transcription was independent of Sit4 and even partially independent of Gln3; i.e., in a gln3Δ mutant, expression was still one-third of the wild-type level (Figure 2A). The latter was a much weaker requirement than observed for the gat1Δ in which DAL5 expression was ∼25-fold lower than wild type (Figure 2A). These results differed from multiple previous studies demonstrating that DAL5 transcription possessed a much greater requirement for Gln3 than Gat1; i.e., transcription had been abolished in a gln3Δ while being only moderately decreased in a gat1Δ (e.g., Daugherty et al. 1993; Coffman et al. 1994, 1996a,b, 1997; Tate et al. 2002; Tate and Cooper 2003).

Figure 2.—

Genetic background influences the relative contributions of Gat1 and Gln3 to rapamycin-induced DAL5 expression. Wild-type and mutant cells were grown in yeast nitrogen base (YNB)–glutamine medium. Split cultures were left untreated (Gln) or treated with rapamycin (0.2 μg/ml) for 30 min (+Rap) and processed for qRT–PCR analysis as described in materials and methods. (A) Total RNA was isolated from wild-type (wt; TB50), gln3Δ (FV005), gat1Δ (FV006), and sit4Δ (FV029) cells of the TB genetic background. Data in A appeared in Georis et al. (2008). (B) Total RNA was isolated from wild-type (25T0b), gln3Δ (FV022), gat1Δ (FV023), and sit4Δ (FV027) cells of the Sigma genetic background. (C) Total RNA was isolated from wild-type (BY4709), gln3Δ (4709ΔGLN3), gat1Δ (4709ΔGAT1), and sit4Δ (FV001) cells of the BY genetic background. (D) Total RNA was isolated from wild-type (25T0b), ure2Δ (FV025), sit4Δ (FV027), and ure2Δsit4Δ (FV076) cells of the Sigma genetic background. (A–D) DAL5 mRNA levels were quantified by qRT–PCR, as described in materials and methods. DAL5 values were normalized with TBP1. Values represent the average of at least three experiments from independent cultures, and error bars indicate standard errors.

The extensive use of DAL5 as a reporter in mechanistic studies of NCR-sensitive transcription and Tor pathway signaling necessitated that we rectify these apparently conflicting results. Since significant differences could not be detected in growth or assay conditions, we shifted our attention to the S. cerevisiae strains themselves. We noted that Σ1278b (Sigma)-related strains had been used in most earlier studies of NCR (Coffman et al. 1995, 1996a,b, 1997; Daugherty et al. 1993; Tate et al. 2002; Tate and Cooper 2003), while those used in more recent Tor studies derived from TB123 (TB)- and occasionally BY4709 (BY)-related strains (Beck and Hall 1999; Tate et al. 2006a,b, 2009; Georis et al. 2008). This correlation suggested that the conflicting results might derive from the strain backgrounds assayed, a possibility strengthened by reports demonstrating a surprisingly high impact of genetic background on NCR-sensitive and retrograde transcriptional regulation (Tate et al. 2002; Dilova and Powers 2006).

Therefore, we compared the requirements of rapamycin-induced DAL5 expression in the three genetic backgrounds. In contrast with data from TB-related strains (Figure 2A), deleting GLN3 (FV022), GAT1 (FV023), or SIT4 (FV027) in the Sigma background decreased rapamycin-induced DAL5 mRNA to almost background levels (Figure 2B). In other words, DAL5 expression exhibited strong requirements for all three proteins rather than for just Gat1. The parallel absence of DAL5 expression in a BY sit4Δ (FV001) strain suggested that this genetic background perhaps possessed requirements that were more similar in this respect to those of Sigma than to those of TB strains (Figure 2C). Together, these data indicate that it was genetic background differences that accounted for the inconsistent GATA factor and Sit4 requirements of DAL5 transcription.

Genetic background specificities rely on mechanisms downstream of GATA factor binding:

Our next objective was to identify the genetic background-specific difference(s) in the strains. The stronger deleterious effect of a sit4Δ on DAL5 expression in Sigma vs. TB strains (compare A and B in Figure 2) enabled us to test the epistatic relationship between sit4 and ure2 mutations (Figure 2D). Rapamycin induced DAL5 transcription to the same high levels in wild type, ure2Δ and ure2Δsit4Δ cells, whereas DAL5 transcription was minimal in the sit4Δ. This demonstrated that ure2 mutations were epistatic to those at SIT4. There was also a modest (twofold) positive synergistic effect of additionally deleting SIT4 in an untreated, glutamine-grown ure2Δ mutant. These data were similar to those obtained in the TB-related strains (Georis et al. 2008) and argued that the regulatory relationships between Ure2 and Sit4 functions were similar in the two backgrounds and thus did not account for the differences in transcription.

We next turned to the sequential component events leading up to transcription, first assaying nuclear GATA factor localization in Sigma-derived wild-type and mutant strains. Rapamycin induced Gln3-Myc13 localization in the wild type, but not in a sit4Δ mutant. Deleting URE2 resulted in constitutive nuclear Gln3-Myc13 localization in the absence of rapamycin whether or not Sit4 was present; i.e., a ure2Δ mutation was clearly epistatic to a SIT4 deletion. The behavior of Gat1-Myc13 was distinct from that of Gln3-Myc13 in two respects: there was a much more modest negative effect of Ure2 on nuclear Gat1-Myc13 localization in untreated cells than for Gln3-Myc13, and the Sit4 requirement for nuclear Gat1-Myc13 localization was minimal. These data indicated that Sigma and TB strains behaved similarly (compare Figure 3 in this work with Figure 6 in Georis et al. 2008). Therefore, requirements for nuclear GATA factor localization also did not account for the differences in DAL5 transcription.

Figure 3.—

Genetic background does not affect rapamycin-induced intracellular localization of Gln3-Myc13 and Gat1-Myc13 in wild-type, ure2Δ, sit4Δ, and ure2Δsit4Δ mutant strains. Wild-type and mutant cells were grown to mid-log phase (A600nm = 0.50–0.52) in YNB–glutamine medium. At this cell density, they were left untreated (Gln) or treated with rapamycin (0.2 μg/ml) for 20 min (+Rap). Samples were then taken, fixed, prepared for indirect immunofluorescence microscopy, and analyzed as described in materials and methods. (A and C) The images are presented in pairs with Gln3-Myc13 (red) visualized in the top image and DAPI-positive material (blue) in the bottom one. Histograms providing a fuller description of the images depicted are shown in B and D. Gln3-Myc13 localization was categorized as cytoplasmic (red), nuclear and cytoplasmic (yellow), and nuclear (green). Strain numbers and pertinent genotypes appear above the images and below the histograms.

The next known downstream event in the pathway was GATA factor binding to its target sequences within NCR-sensitive promoters. Here we found that, with one major exception, GATA factor binding to the DAL5 promoters of Sigma and TB strains was similar (compare Figure 4, A and B, in this work with Figure 7, A and B, in Georis et al. 2008). The exception occurred in the sit4Δure2Δ double mutant. Rapamycin induced much greater Gln3-Myc13 binding and much weaker Gat1-Myc13 binding to the Sigma DAL5 promoter than it did to the TB promoter where there was no demonstrable effect. Additionally, as in the TB background, rapamycin-induced Gat1-Myc13 binding to the Sigma DAL5 promoter was independent of Gln3 (compare Figure 5A in this work with Figure 4A in Georis et al. 2008), whereas Gln3-Myc13 binding was Gat1 dependent (compare Figure 5B in this work with Figure 4B in Georis et al. 2008).

Figure 4.—

Genetic background does not affect rapamycin-induced in vivo binding of Gln3-Myc13 and Gat1-Myc13 to the DAL5 promoter in wild type, ure2Δ, sit4Δ, and ure2Δsit4Δ mutant strains. Wild-type and mutant cells were grown in YNB–glutamine medium. Split cultures were left untreated (Gln) or treated with rapamycin (0.2 μg/ml) for 30 min (+Rap), sampled, and processed for ChIP analyses as described in materials and methods. Wild-type untagged cells (25T0b) were used as a negative control. (A) In vivo binding of Gln3-Myc13 to the DAL5 promoter in wild type (FV036), ure2Δ (FV090), sit4Δ (FV069), and ure2Δsit4Δ (FV076) strains of the Sigma genetic background. (B) In vivo binding of Gat1-Myc13 to the DAL5 promoter in wild-type (FV034), ure2Δ (FV086), sit4Δ (03803c), and ure2Δsit4Δ (FV087) strains of the Sigma genetic background. (A and B) ChIP was performed using antibodies against c-myc as described in materials and methods. qPCR of IP/IN fractions was performed with primers specific for the DAL5 promoter [DAL5P] and for a region 2.5 kb upstream of the DAL5 ORF as a control [DAL5U]. For each immunoprecipitation, IP/IN values were calculated as follows: ([DAL5P]IP/[DAL5P]IN − [DAL5U]IP/[DAL5U]IN). Histograms represent the average of two IPs performed on at least two experiments from independent cultures. Error bars indicate standard errors.

Figure 5.—

Genetic background does not affect rapamycin-induced in vivo binding of Gln3-Myc13 and Gat1-Myc13 to the DAL5 promoter in gat1Δ and gln3Δ mutant strains. The experimental format and ChIP analyses were as described in Figure 4 using the wild-type and mutant strains listed below. Wild-type untagged cells (25T0b) were used as negative control. (A) In vivo binding of Gat1-Myc13 to the DAL5 promoter in wild-type (FV034) and gln3Δ (03740c) strains of the Sigma genetic background. (B) In vivo binding of Gln3-Myc13 to the DAL5 promoter in wild type (FV036) and gat1Δ (FV041) strains of the Sigma genetic background.

The similarity of rapamycin-induced nuclear Gln3-Myc13 and Gat1-Myc13 localization and binding to the DAL5 promoters of TB and Sigma strains led us to conclude, by exclusion, that the genetic background-dependent difference occurred after GATA factor binding to DAL5. In the TB genetic background, Gln3-independent Gat1-Myc13 binding alone was sufficient to activate rapamycin-induced DAL5 expression, albeit at a somewhat diminished level (Georis et al. 2008). In contrast, the above data showed that in the Sigma background Gat1-Myc13 could bind to the DAL5 promoter independently of Gln3, but was unable to activate transcription unless Gln3 was also present. Thus genetic background-specific differences in DAL5 transcription likely derived from differences in the requirements for Gat1-dependent transcriptional activation.

Differing DAL5 expression requirements in the TB and Sigma genetic backgrounds are not monogenic:

The genetic background-specific differences in Gat1's potential to activate rapamycin-induced DAL5 transcription prompted us to determine whether one or more genes were involved. To this end, we crossed Sigma FV027 (MATα, sit4Δ) with TB FV029 (MATa, sit4Δ) and assayed rapamycin-induced DAL5 expression in two independently isolated diploids and segregants of four tetrads (03847, 03848, 03849, and 03850) derived from them. Low DAL5 expression in heterozygous diploid cells, equivalent to levels observed in the Sigma-derived strains, indicated that Sigma strains possessed the dominant trait, i.e., the inability of Gat1 to mediate Gln3-independent transcription (Figure 6). However, consistent 2:2 segregation of the parental phenotypes was not observed, suggesting that, overall, two or more genes were involved (Figure 6).

Figure 6.—

Differing DAL5 expression in TB and Sigma genetic backgrounds are not monogenic. sit4Δ cells of the Sigma (FV027) and TB (FV029) genetic backgrounds were crossed. Two diploids (diploids 1 and 2) from this cross (FV027 × FV029) were assayed along with the segregants of four tetrads issued from the sporulation of diploid 1 (03847, 03848, 03849, and 03850). All of the strains were grown in YNB–glutamine medium, treated with rapamycin (0.2 μg/ml) for 30 min, and processed for qRT–PCR analysis as described in Figure 2.

Genetic background-specific requirements for rapamycin-induced DAL5 transcription derive from a negatively-acting GATA factor:

The finding that a dominant trait was responsible for the inability of Gat1 alone to support rapamycin-induced DAL5 transcription in Sigma strains suggested that a negatively-acting gene product might be responsible for the background-specific difference. Since two negatively-acting proteins are well known to regulate NCR-sensitive transcription, the GATA repressors Dal80 and Gzf3/Deh1/Nil2 were our first candidates. Deleting DAL80 has no effect on DAL5 transcription in glutamine-grown Sigma cells, whereas deleting DEH1/GZF3 weakly increases DAL5 expression under these culture conditions (Daugherty et al. 1993; Coffman et al. 1995, 1997). The effects of these mutations in rapamycin-treated cells, on the other hand, have never been reported. Therefore, we obtained these data and found that loss of either Dal80 or Gzf3 modestly increased DAL5 expression (less than twofold) compared to the wild-type level (Figure 7A). However, the two deletion phenotypes differed greatly if additionally accompanied by the deletion of GLN3. Rapamycin-induced DAL5 expression in the dal80Δ was completely Gln3 dependent, whereas in the gzf3Δ it was largely Gln3 independent and highly Gat1 dependent (Figure 7A). These data were those expected if, in contrast to the situation in Sigma strains, Gzf3 activity in TB strains had little effect on Gat1-mediated transcription. They were also in keeping with the observation that Sigma strains possessed the dominant trait.

Figure 7.—

Genetic background-specific requirements for rapamycin-induced DAL5 transcription derive from the negative GATA factor Gzf3. The experimental format and qRT–PCR analyses were as described in Figure 2 using wild-type and the mutant strains listed below. (A) Total RNA was extracted from wild-type (25T0b), gln3Δ (FV022), dal80Δ (FV080), dal80Δgln3Δ (FV110), gzf3Δ (FV083), gzf3Δgln3Δ (FV113), gat1Δ (FV023), and gzf3Δgat1Δ (FV114) cells of the Sigma genetic background. DAL5 expression was assayed as described in Figure 2. (B) Total RNA was extracted from wild-type TB (TB50) or Sigma (25T0b) and gzf3Δ TB (FV211) or Sigma (FV083) cells. DAL5 expression was assayed as described in Figure 2. (C) Total RNA was extracted from wild-type TB (TB50) or Sigma (25T0b) cells. GZF3 mRNA levels were quantified by qRT–PCR, as described in materials and methods. GZF3 values were normalized with TBP1. Values represent the average of at least three experiments from independent cultures, and error bars indicate standard errors.

To test this conclusion further, we compared the effects of deleting GZF3 in the TB- and Sigma-derived strains (Figure 7B). Deleting GZF3 in a TB background had little, if any, negative effect on rapamycin-induced DAL5 transcription, whereas in the Sigma background deleting GZF3 modestly increased transcription. Finally, we determined the effects of rapamycin treatment on expression of the GZF3 gene itself and found that it had no effect in either genetic background (Figure 7C). However, there were approximately fourfold greater amounts of GZF3 mRNA in a Sigma wild type compared to that in the TB background (Figure 7C). In agreement with this observation it was possible to detect Gzf3 protein by Western blot analysis in extracts derived from Sigma but not TB strains (data not shown). From these data, we concluded that Gzf3 inhibited the ability of Gat1, in the absence of Gln3, to activate transcription only in the Sigma genetic background.

Genetic background-dependent ability of Gat1 to activate transcription without Gln3 is gene specific:

Since it was conceivable that the difference in Gat1's ability to activate rapamycin-induced DAL5 transcription independently of Gln3 might derive from some peculiarity in the DAL5 promoter, we analyzed the expression of two other NCR-sensitive genes, DAL7 and GAP1. As occurred with DAL5, deleting SIT4 did not adversely affect rapamycin-induced DAL7 expression in the TB genetic background (Figure 8A), but significantly decreased DAL7 mRNA in the corresponding Sigma strain (Figure 8B). Yet, rapamycin-induced Gat1-Myc13 binding to the DAL7 promoter in sit4Δ cells was comparable in both genetic backgrounds (Figure 9, A and B). Thus, in the Sigma background, Gat1-Myc13 binding was again not sufficient to activate rapamycin-induced DAL7 expression, showing that genetic background-dependent differences could be demonstrated in more than one gene and hence were unlikely to be caused by some pecularity in a single promoter.

Figure 8.—

Genetic background-specific requirements for rapamycin-induced transcription are not restricted to DAL5. The experimental format and qRT–PCR analyses were performed as described in Figure 2 using wild-type and sit4Δ cells. (A) Total RNA was extracted from wild-type (TB50) and sit4Δ (FV029) cells of the TB genetic background. (B) Total RNA was extracted from wild-type (25T0b) and sit4Δ (FV027) cells of the Sigma genetic background. (A and B) DAL7 mRNA levels were quantified by qRT–PCR, as described in materials and methods. DAL7 values were normalized with TBP1.

Figure 9.—

Genetic background does not affect rapamycin-induced in vivo binding of Gat1-Myc13 to the DAL7 promoter in a sit4Δ mutant strain. The experimental format and ChIP analyses were as described in Figure 4 using the wild-type and mutant strains listed below. (A) In vivo binding of Gat1-Myc13 to the DAL7 promoter in wild-type (FV063) and sit4Δ (FV066) strains of the TB genetic background. Wild-type untagged cells (TB50) were used as negative control. (B) In vivo binding of Gat1-Myc13 to the DAL7 promoter in wild-type (FV034) and sit4Δ (03803c) strains of the Sigma genetic background. Wild-type untagged cells (25T0b) were used as negative control. qPCR of IP/IN fractions was performed with primers specific for the DAL7 promoter [DAL7P] and for a region 2.5 kb upstream of the DAL5 ORF as a control [DAL5U]. IP/IN values were calculated as follows: ([DAL7P]IP/[DAL7P]IN − [DAL5U]IP/[DAL5U]IN).

On the other hand, the responses and requirements of DAL5 and DAL7 transcription to rapamycin treatment were not general. As shown in Figure 10, A and B, GAP1 transcription occurred in gln3Δ and gat1Δ single mutants, but not in a gln3Δgat1Δ double mutant. In other words, a GATA activator was required, but either Gln3 or Gat1 would do. Consistent with the absence of a Gln3 requirement for rapamycin-induced GAP1 expression, Sit4 was also dispensable in both genetic backgrounds (Figure 10, A and B).

Figure 10.—

Genetic background-specific requirements for rapamycin-induced transcription are not identical for all NCR-sensitive genes. The experimental format and qRT–PCR analyses were as described in Figure 2 using wild-type and the mutant strains listed below. (A) Total RNA was extracted from wild-type (TB50), gln3Δ (FV005), gat1Δ (FV006), gat1Δgln3Δ (FV007), and sit4Δ (FV029) cells of the TB genetic background. (B) Total RNA was extracted from wild type (25T0b), gln3Δ (FV022), gat1Δ (FV023), gat1Δgln3Δ (FV024), and sit4Δ (FV027) cells of the Sigma genetic background.

These results made a clear prediction that rapamycin-induced Gln3-Myc13 and Gat1-Myc13 binding to the GAP1 promoter should be independent of the other GATA factor irrespective of the genetic background assayed. As shown in Figure 11, these predictions were substantiated experimentally. Gln3-Myc13 binding to the GAP1 promoter occurred at nearly wild-type levels in a gat1Δ of either background (Figure 11, A and B). Similarly, Gat1-Myc13 binding occurred at nearly wild-type levels in both TB gln3Δ and Sigma gln3Δ strains (Figure 11, C and 1D).

Figure 11.—

In vivo binding profiles of Gat1-Myc13 and Gln3-Myc13 to the GAP1 promoter are similar in TB and Sigma genetic backgrounds and are only weakly affected by the deletion of one or the other GATA activator. The experimental format and ChIP analyses were as described in Figure 4, using the wild-type and mutant strains listed below. qPCR of IP/IN fractions was performed with primers specific for the GAP1 promoter [GAP1P] and for a region 2.5 kb upstream of the DAL5 ORF as a control [DAL5U]. IP/IN values were calculated as follows: ([GAP1P]IP/[GAP1P]IN − [DAL5U]IP/[DAL5U]IN). (A) In vivo binding of Gln3-Myc13 to the GAP1 promoter in wild-type (TB123) and gat1Δ (FV018) strains of the TB genetic background. Wild-type untagged cells (TB50) were used as negative control. (B) In vivo binding of Gln3-Myc13 to the GAP1 promoter in wild-type (FV036) and gat1Δ (FV041) strains of the Sigma genetic background. Wild-type untagged cells (25T0b) were used as negative control. (C) In vivo binding of Gat1-Myc13 to the GAP1 promoter in wild-type (FV063) and gln3Δ (FV064) strains of the TB genetic background. Wild-type untagged cells (TB50) were used as negative control. (D) In vivo binding of Gat1-Myc13 to the GAP1 promoter in wild-type (FV034) and gln3Δ (03740c) strains of the Sigma genetic background. Wild-type untagged cells (25T0b) were used as negative control.

Finally, we asked how rapamycin-induced GAP1 expression in TB and Sigma strains might be affected by deletion of GZF3. As shown in Figure 12, there was little to perhaps a slightly negative effect of a gzf3Δ on rapamycin-induced GAP1 transcription in the TB genetic background, thereby supporting and extending observations obtained with DAL5. In the Sigma background, there was a slight increase in rapamycin-induced GAP1 expression, but it was more modest than observed with DAL5. In marked contrast, however, GAP1 expression in an untreated gzf3Δ was greatly increased relative to wild type only in the Sigma genetic background, suggesting that Gzf3 had no repressive effect in glutamine-grown cells of the TB genetic background (Figure 12).

Figure 12.—

Gzf3 does not control GAP1 expression in the TB genetic background. The experimental format and qRT–PCR analyses were as described in Figure 2 using wild-type and the mutant strains listed below. Total RNA was extracted from wild-type (25T0b) and gzf3Δ (FV083) cells of the Sigma genetic background and from wild-type (TB50) and gzf3Δ (FV211) cells of the TB genetic background.

DISCUSSION

Dissecting the detailed molecular events through which the Tor signal transduction pathway influences NCR-sensitive transcription has been and continues to be a complex task. However, this work will likely contribute to rectifying some of the apparently conflicting conclusions, such as the differences in Gln3-/Gat1-dependent transcription in Sigma and TB strains, and avoid future inconsistencies between predicted and observed results. Our experiments have identified and explained several unexpected variables that potentially compromise interpretation of many reporter gene assays and transcriptome analyses employed in the past to assess Tor pathway function, including (i) the genetic background of the strains analyzed, (ii) the regulatory characteristics of the reporter genes being assayed, (iii) the nature of their GATA factor requirements, and (iv) the interactions between GATA and other transcription factors that are cumulatively responsible for the transcription of NCR-sensitive genes.

The conceptual problem encountered in interpreting reporter gene assays is the fact that they measure the gene's overall expression, which is the cumulative outcome of multiple levels of regulation (Smart et al. 1996). While this has long been known, what has not been appreciated until recently is that Gln3 and Gat1 are regulated quite differently (Kulkarni et al. 2006; Georis et al. 2008). Gln3 is more sensitive than Gat1 to negative regulation by Ure2. Furthermore, unlike Gln3, nuclear Gat1 accumulation is much less Sit4 dependent (Georis et al. 2008). There are additional variables that alone or in combination can influence overall NCR-sensitive gene expression and hence conclusions derived from their use in investigations of the Tor pathway: (i) the ability of one GATA activator to activate transcription of the NCR-sensitive gene assayed in the absence of the other GATA activator; (ii) the ability of the negatively acting GATA factors (Dal80, Gzf3) to control the NCR-sensitive gene assayed; (iii) the ability of non-GATA transcription factors to function with one GATA factor, thereby relieving the necessity of the second one's participation in supporting transcription when both GATA factors are required; and (iv) the influence of the genetic background of the strain being investigated.

The above variables and potential problems of interpretation that derive from them are exemplified by the two strains (TB and Sigma) and three genes (DAL5, DAL7, and GAP1) assayed in this work. Gat1, in the absence of Gln3, can activate DAL5 and DAL7 transcription in TB- but not Sigma-derived strains. The latter genetic background requires both Gln3 and Gat1 to activate DAL5 and DAL7 transcription. Since Gln3 but not Gat1 requires Sit4 for nuclear localization, DAL5 and DAL7 transcription in TB strains exhibits no Sit4 requirement, whereas in Sigma strains the Sit4 requirement is nearly absolute. Therefore, apparent participation of Tor regulation in this scenario depends upon which strain is assayed. Consider a second example, i.e., assaying two genes in the Sigma genetic background. DAL5 and DAL7 transcription requires both Gln3 and Gat1 and hence there is a strong Sit4 requirement for DAL5 and DAL7 transcription. On the other hand, either Gln3 or Gat1, in the absence of the other GATA activator, is capable of activating GAP1 transcription in both TB and Sigma strains. Therefore, if DAL5 and DAL7 expression is assayed in Sigma strains, a strong Sit4 requirement is observed, but if GAP1 is assayed in the same genetic background, no such requirement is observed. For results employing NCR-sensitive gene transcription as a reporter of Tor pathway activity to be interpretable, measurements of the reporter gene's expression in gln3Δ and gat1Δ single mutants as well as in the gln3Δgat1Δ double mutant is necessary. Data with the double mutant alone are insufficient.

The fact that one can obtain conflicting results is clear from these examples. However, the disparate conclusions may be rectified by considering the characteristics of the strains and genes assayed: (i) Sit4 was required for Gln3, but not Gat1 nuclear localization in both the TB and Sigma genetic backgrounds; (ii) Gzf3 prevented Gat1 from activating GATA-factor-dependent gene transcription in Sigma but not in the TB strain; (iii) without Gat1, Gln3 was able to bind to the promoters of some NCR-sensitive genes but not others and activate their transcription even in the Sigma genetic background; and (iv) Gat1 did not require Gln3 to bind to NCR-sensitive promoters but was not always able to activate transcription.

The apparently paradoxical nature of the last two characteristics may derive from the participation of both GATA and non-GATA factors in the activation of genes whose transcription exhibits these characteristics. Several examples of such functional cooperation in GATA-factor-mediated transcription have been previously documented:

  1. The CAR1 promoter consists of at least 14 functional upstream activation sequences working together to produce the overall CAR1 expression (Smart et al. 1996). Two UASNTR GATA elements are responsible for its NCR sensitivity, while three UASI elements are responsible for inducibility by arginine. Under some culture conditions, all of these elements function, while with others only a subset of them are operating (Smart et al. 1996; Dubois and Messenguy 1997).

  2. In the case of the PUT1 promoter, a synthetic fragment from the promoter, containing a single GATA element and a Put3 binding site, supports heterologous reporter gene transcription. In this case, only a single GATA factor can be functioning because there is only one binding site present (Rai et al. 1995).

  3. In the case of DAL5, there are required upstream activation sequences immediately downstream (two and one turns, 21 and 11 bp, respectively) of each of the two GATA elements that account for the majority (∼90%) of DAL5 transcription (Rai et al. 2004).

  4. Finally, placing a wild-type Dal82-binding site (cis-acting element responsible for allophanate-inducible gene expression) adjacent to a mutated GATA element (Gln3/Gat1-binding site) suppressed the mutant phenotype (Van Vuuren et al. 1991).

This work makes three additional contributions that further our understanding of nitrogen regulation in S. cerevisiae. The first is identification of a new regulatory function for the negatively acting GATA factor, Gzf3. Although the existence of Gzf3 and the regulation of its production have been known for some time, little was known about why two homologous, negatively acting GATA factors (Dal80 and Gzf3/Deh1) existed. This work provides one such function. Why, however, Gzf3 is present and functional in the Sigma but not in the TB genetic background remains to be elucidated. Second is the remarkably robust ability of rapamycin to increase DAL5 transcription and GATA factor binding to the promoters of these genes in ure2Δ cells as well as the genetic background-specific effects brought about by additionally deleting SIT4 in both the presence and absence of rapamycin. These observations forecast that one or more yet-to-be-discovered components of the Tor1,2 regulatory pathway function downstream of intracellular transcription factor localization. Third, the above observations may help to explain why lists of NCR-sensitive genes that appear in the results of various transcriptome analyses vary so much and also frequently fail to identify multiple genes whose transcription is previously known to be NCR sensitive.

Acknowledgments

We thank Michael Hall for strains, Tim Higgins for preparing the artwork, Fabienne Vierendeels for excellent technical assistance, and the University of Tennessee Yeast Group for suggestions for improving the manuscript. Work by E.D. and I.G. was supported by the Commission Communautaire Française. Work by J.J.T. and T.G.C. was supported by National Institutes of Health (NIH) grant GM-35642 and NIH/National Science Foundation grant DMS-0443901.

Footnotes

  • Communicating editor: A. P. Mitchell

  • Received December 2, 2008.
  • Accepted December 18, 2008.

References

View Abstract