All strains constructed in this study are isogenic to the FY1679 wild-type strain (listed in Table ​Table1).1). Deletion strains, involving insertion of KanMX and NatMX cassettes, were constructed using the long flanking homology strategy of Wach (41) as described in reference 39 using primers listed in Table ​Table2.2. Chromosomal DAL80, GAT1, GLN3, and GZF3 gene products were tagged at their C termini with 13 copies of the Myc epitope (Myc¹³) as described by Longtine et al. (25), using primers described in Table ​Table22 or in reference 20. FV222 was constructed by replacing the wild-type GAT1 allele by GAT1-TAP, which was obtained by PCR amplification of a GAT1-TAP-tagged strain using GAT1L1 and GAT1L4 primers (20). Functionality of the tagged constructs was validated using qRT-PCR to compare NCR regulation in tagged versus untagged strains (data not shown).

Total protein extracts were obtained from 10-ml cultures using trichloroacetic acid precipitation as described in reference 40. One-twentieth of the total extracts was loaded on a 4 to 12% NuPage gel (Invitrogen), run for 75 min according to the manufacturer's instructions, and transferred to a Hybond-ECL nitrocellulose membrane (GE Healthcare) for 1 h in an XCell II blot module (Invitrogen) according to the manufacturer's protocol. Membranes were incubated with mouse monoclonal anti-Myc antibodies (9E10; Santa Cruz) (1:800) using an ECL Advance Western blotting detection kit, and Myc-tagged proteins were visualized using a chemiluminescence camera (Chemi-Smart from Vilbert-Lourmat). Band intensities were quantified using the Bio-1D algorithm and normalized with the intensity of two different bands in the corresponding Ponceau-stained lane.

Cultures (100 ml) were harvested, washed once in 50 mM Tris, pH 8, and resuspended in 2 ml of buffer (50 mM Tris, pH 8, 150 mM NaCl, 5 mM EDTA, 0.05% NP-40, 1 mM phenylmetyhlsulfonyl fluoride, and complete protease inhibitor cocktail tablets [Roche]). Lysis was performed by shaking with 425- to 600-μm acid-washed glass beads (Sigma) on an IKA Vibrax VXR orbital shaker at maximum speed for 30 min at 4°C. Cell debris and glass beads were removed by centrifugation. Immunoprecipitation of TAP-tagged proteins was performed by incubating 500 μl of total cell extracts with 50 μl of prewashed (three times in 0.1% phosphate-buffered saline-bovine serum albumin) Dynabeads PAN mouse immunoglobulin G (Invitrogen) and 50 μl of 1% phosphate-buffered saline-bovine serum albumin during 2 h of orbital shaking (800 rpm) at 30°C. Immune complexes were washed three times in lysis buffer, eluted by boiling in sodium dodecyl sulfate (SDS) sample buffer, and loaded on SDS-polyacrylamide gel for anti-Myc Western blotting.

Gat1-Myc¹³, Gln3-Myc¹³, and Gzf3-Myc¹³ subcellular localization was performed similarly to the Gln3-Myc¹³ localization described in reference 18.

Total RNA was extracted from 4-ml cultures as described previously (32). cDNA was generated from 100 to 500 ng of total RNA using a RevertAid H Minus first-strand cDNA synthesis kit with oligo(dT)₁₈ primers from Fermentas using the manufacturer's recommended protocol. cDNAs were subsequently quantified by RT-PCR using the Maxima SYBR green qPCR master mix from Fermentas as described in reference 20. Sequences of the primers for qPCR are provided in Table ​Table22.

The presence of Gat1 at NCR promoters obviously generates transcriptional activation of NCR genes, since Gat1 is one of the transcriptional activators of NCR genes, but it may also favor the recruitment of other GATA factors or stabilize activating complexes. Indeed, in our previous study, we provided evidence in favor of Gat1 participating in the promotion of Gln3 binding upon rapamycin treatment (20). In order to study the influence of Gat1 on Gln3 binding to NCR-sensitive promoters when cells are grown on proline, we performed ChIP analysis of Gln3-Myc¹³ binding to P_GAP1 in wild-type (FV036), gat1Δ (FV041), dal80Δ (FV082), and gat1Δ dal80Δ (FV098) strains grown on glutamine- and proline-containing media. As was the case upon rapamycin treatment (19, 20), in vivo binding of Gln3-Myc¹³ to P_GAP1 was significantly enhanced in wild-type cells grown on proline versus that with glutamine (Fig. ​(Fig.2A).2A). Deleting GAT1 drastically decreased the amount of Gln3-Myc¹³ bound to P_GAP1 on proline (Fig. ​(Fig.2A),2A), although the absence of Gat1 does not influence GLN3 expression (28; also data not shown) and had no effect on Gln3-Myc¹³ subcellular localization on proline (Fig. ​(Fig.2B).2B). Deleting DAL80 moderately increased the binding of Gln3-Myc¹³ to P_GAP1 on proline (Fig. ​(Fig.2A).2A). Given the low Gln3-Myc¹³ binding observed in a proline-grown strain lacking both DAL80 and GAT1 (Fig. ​(Fig.2A),2A), we assume that the increase of Gln3-Myc¹³ binding to P_GAP1 in proline-grown dal80Δ cells is caused by the increase of Gat1 binding in these cells. Parallelism between the binding efficiency of Gat1-Myc¹³ (Fig. ​(Fig.1A)1A) and that of Gln3-Myc¹³ (Fig. ​(Fig.2A)2A) strongly suggests interdependence between the two GATA activators for DNA binding.

Our data and previous observations that Gat1 expression is tightly regulated and that this regulation influences its in vivo DNA binding, affecting gene expression directly (as a transcriptional activator) and indirectly (as a modulator of Gln3 binding), led us to hypothesize that Gat1 could occupy a central position in NCR-sensitive gene activation. In this context, it was of high interest to define how Dal80 could interfere with Gat1-mediated transactivation. We therefore analyzed, by qRT-PCR, the expression of a gene that is particularly sensitive to Dal80-mediated repression: UGA4 (38). UGA4 expression was analyzed in a wild-type strain and in cells lacking GAT1, DAL80, or both (FV111). In proline-grown, wild-type cells, UGA4 activation was very low (Fig. ​(Fig.3A).3A). Deleting DAL80 increased UGA4 expression 40-fold (Fig. ​(Fig.3A),3A), and gat1Δ was epistatic to dal80Δ, since the expression level in a gat1Δ dal80Δ strain was similar to that in a gat1Δ strain. To determine whether Dal80 interfered with Gat1 DNA binding, we performed ChIP experiments in wild-type and dal80Δ Gat1-Myc¹³ cells (Fig. ​(Fig.3B).3B). Gat1-Myc¹³ binding to P_UGA4 was induced in proline-grown, wild-type cells, compared to levels in glutamine-grown cells, even if UGA4 was only weakly expressed in the wild-type strain (Fig. 3A and B). Moreover, deleting DAL80 led to an increase in Gat1-Myc¹³ binding to P_UGA4 on proline compared to the level in the wild type. The rise in Gat1-Myc¹³ DNA binding could be at least partially associated with an increase in GAT1 expression levels, since the GAT1 transcript level is higher in proline-grown dal80Δ cells than in wild-type cells (Fig. ​(Fig.1B).1B). Consistently, Dal80-Myc¹³ binding to P_GAT1 was induced in proline-grown, wild-type cells (Fig. ​(Fig.3C).3C). In order to evaluate the direct action of Dal80 at NCR-sensitive promoters without interference caused by the NCR-sensitive, Dal80-regulated expression of GAT1, we placed GAT1 under the control of the GAL1 promoter, thereby rendering GAT1 expression no longer NCR sensitive but glucose repressible. We quantified the expression of UGA4 in the P_GAL1-GAT1-MYC¹³ strain carrying either the wild-type DAL80 or a dal80Δ (FV190) allele and cultivated it in the presence of galactose and proline with increasing amounts of glucose. UGA4 transcript levels remained low in wild-type P_GAL1-GAT1-MYC¹³ cells grown in the presence of proline and galactose, even when no glucose was added to the culture medium (Fig. ​(Fig.4A).4A). Deletion of DAL80 generated higher UGA4 transcript levels for every concentration of glucose tested (Fig. ​(Fig.4A),4A), suggesting that DAL80 can act directly at P_UGA4. We then checked if, in these two P_GAL1-GAT1-MYC¹³ strains, Gat1-Myc¹³ binding to P_UGA4 was affected by Dal80. Binding of Gat1-Myc¹³ to P_UGA4 followed its expression level (Fig. 4B and C) and was moderately increased in dal80Δ cells compared to the level in the wild type (Fig. ​(Fig.4C),4C), indicating that Dal80 interferes with Gat1 binding to DNA, but only very weakly compared to the expression increase caused by the absence of Dal80. In other words, at equal binding levels of Gat1-Myc¹³ (compare 1% Gal for the wild type and 1% Gal plus 0.1% Glu for the dal80Δ strain [Fig. ​[Fig.4C]),4C]), we noticed a huge difference in UGA4 transcript levels (compare 1% Gal for the wild type and 1% Gal plus 0.1% Glu for the dal80Δ strain [Fig. ​[Fig.4A]),4A]), suggesting that Dal80 has an additional role in UGA4 repression, interfering with transcriptional activation. The last step in demonstrating in vivo competition between Dal80 and Gat1 for binding to NCR-sensitive promoters was to analyze the profile of Dal80 binding to P_UGA4. Therefore, we performed ChIP analysis of wild-type (FV078), gln3Δ (FV108), and gat1Δ (FV109) strains expressing Dal80-Myc¹³ to determine its binding at the promoter of UGA4 (Fig. ​(Fig.5A).5A). In proline-grown, wild-type cells, Dal80-Myc¹³ binding to P_UGA4 was induced. Reducing Dal80 expression, by growing cells on glutamine or in GATA activator mutants (Fig. ​(Fig.5B),5B), resulted in a corresponding decrease of Dal80-Myc¹³ binding (Fig. ​(Fig.5A),5A), suggesting that Dal80-Myc¹³ binding to promoters in vivo is solely controlled by its expression level.

Aiming at the elucidation of the mechanisms regulating Gat1 activity on glutamine, we quantified the expression of GAP1—a well-characterized target of Gzf3-mediated repression (34)—using qRT-PCR in a wild-type strain and in cells lacking GAT1, GLN3, or GZF3, either in single units (FV023, FV022, and FV083) or in pairwise combinations (gat1Δ gln3Δ, FV024; gat1Δ gzf3Δ, FV114; and gln3Δ gzf3Δ, FV113). Detailed analysis of Fig. ​Fig.66 generates the following observations: (i) expression of GAP1 in proline-grown cells was abolished only when the two activators were lacking (gat1Δ gln3Δ), suggesting that both Gln3 and Gat1 are involved in GAP1 transactivation; (ii) maximal expression of GAP1 in proline-grown cells required the presence of both Gln3 and Gat1; and (iii) GAP1 expression in glutamine-grown cells lacking GZF3 reached the wild-type derepressed level regardless of the presence of Gln3 (gln3Δ gzf3Δ) but depending on Gat1 (gat1Δ gzf3Δ). These results are consistent with earlier reports suggesting that the negative effect of GZF3 on glutamine occurs through interference with Gat1-mediated activation (30, 34).

We next investigated the impact of a GZF3 deletion on Gat1 DNA binding. Glutamine- and proline-grown GAT1-MYC¹³ wild-type and gzf3Δ (FV084) cells were subjected to ChIP analysis for the presence of Gat1-Myc¹³ at P_GAP1. Whereas deleting GZF3 had no impact on Gat1-Myc¹³ binding in proline-grown cells, it moderately increased in glutamine-grown gzf3Δ cells versus the level in the wild type (Fig. ​(Fig.7A).7A). Consistent with the increased Gat1-Myc¹³ binding to P_GAP1 in glutamine-grown gzf3Δ cells, the observed GAT1 expression levels were higher than those in wild-type cells grown in the same conditions (Fig. ​(Fig.7B).7B). As expected, Gat1-Myc¹³ binding remained proline inducible, given the fact that GAT1 expression (Fig. ​(Fig.7B)7B) and Gat1 localization (data not shown) were still regulated by the nitrogen source. The elevated GAP1 expression level reached in gzf3Δ cells (comparable to the wild-type, derepressed level; Fig. ​Fig.6),6), despite rather low Gat1 binding, is explained by Dal80-Myc¹³ binding in glutamine-grown gzf3Δ cells that is correspondingly lower than that in proline-grown, wild-type cells (data not shown). Next, we analyzed the binding of Gat1-Myc¹³ and Gzf3-Myc¹³ (FV079) at P_GAT1. In glutamine-grown, wild-type cells, Gat1-Myc¹³ bound its own promoter but, as expected, in a rather small amount compared to the level it reached on proline (Fig. ​(Fig.7C).7C). These results corroborate previous data indicating that GAT1 is autoregulated (30), though the opposite has also been reported (8). Deleting GZF3 moderately increased Gat1-Myc¹³ binding at P_GAT1. In addition, Gzf3-Myc¹³ significantly bound P_GAT1 on glutamine (Fig. ​(Fig.7D),7D), which is in agreement with previous hypotheses that Gzf3 downregulates GAT1 expression, possibly by preventing its autoactivation (30) through competition for binding to the same GATA sequences in P_GAT1, as it was presumed to occur at other NCR-sensitive promoters (34).

In order to determine whether Gzf3 controls NCR gene expression only through GAT1 expression, we quantified GAP1 expression by qRT-PCR in P_GAL1-GAT1-MYC¹³ strains carrying either a wild-type or a deleted allele of GZF3 (FV172) and cultivated in the presence of galactose and glutamine with increasing amounts of glucose (Fig. ​(Fig.8A).8A). In both wild-type and gzf3Δ strains, reducing GAT1 expression (Fig. ​(Fig.8B)8B) led to a decrease in GAP1 expression. Interestingly, at equal Gat1 concentrations in wild-type and gzf3Δ strains (Fig. ​(Fig.8B),8B), GAP1 expression was always higher in the absence of Gzf3 (Fig. ​(Fig.8A),8A), suggesting a negative role for Gzf3 in GAP1 expression in addition to its control of GAT1 expression. ChIP analysis was conducted to assess the binding of constitutively expressed Gat1-Myc¹³ in wild-type and gzf3Δ strains expressing P_GAL1-Gat1-Myc¹³ grown on galactose and glutamine. Deleting GZF3 increased the in vivo binding of Gat1-Myc¹³ to P_GAP1 (Fig. ​(Fig.8C),8C), explaining the raised GAP1 expression levels. All together, our results suggest that besides affecting GAT1 expression, Gzf3 affects GAP1 expression directly by reducing Gat1 binding at P_GAP1.

Dal80-negative action can take place in vivo at three different levels (Fig. 10B). First, our results confirmed earlier data showing that Gat1 and Dal80 regulators are linked by reciprocal transcriptional control (7, 12, 34). Indeed, GAT1 expression increased in dal80Δ proline-grown cells compared to that in the wild type and DAL80 expression was severely affected in proline-grown gat1Δ cells. Thus, in proline-grown cells, Dal80 downregulates NCR-sensitive genes by lowering the amount of Gat1 produced. Second, at the UGA4 promoter, which contains a canonical URS_GATA (1, 2, 15), Dal80 competes with Gat1 for binding to similar or overlapping sites: (i) Dal80 binds in vitro to P_UGA4 (14); (ii) the in vivo binding of Dal80 and Gat1 to P_UGA4 was elevated in nitrogen derepression conditions; and (iii) the binding of constitutively expressed Gat1-Myc¹³ to P_UGA4 was raised upon DAL80 deletion. Third, impressive changes in UGA4 expression have been observed following DAL80 deletion (about 20×), although constitutively expressed Gat1-Myc¹³ binding was only weakly affected (about 2×). These data suggest that Dal80 could have a third, yet-unreported effect that does not rely on GAT1 expression nor on its binding and rather corresponds to Dal80-mediated repression interfering with transcriptional activation.

Initial observations of Gzf3 function, using GAP1-lacZ fusions, have demonstrated an antagonistic role for Gzf3 on activation by Gat1 in cells grown under nitrogen repression conditions (30, 34). Our qRT-PCR results confirm that derepression of GAP1 expression in glutamine-grown gzf3Δ cells requires Gat1 and not Gln3. The absence of involvement of Gln3 is explained by its cytoplasmic sequestration by Ure2 in these conditions. Our study provides more details on how Gzf3 negatively influences NCR-regulated gene expression (Fig. 10A). First, several lines of evidence validate previous assumptions that Gzf3 competes with Gat1 at its own promoter (30), thereby reducing the expression of the activator in conditions of nitrogen abundance: (i) GAT1 expression was higher in glutamine-grown gzf3Δ cells than in wild-type cells, confirming earlier lacZ fusion data (30); (ii) rendering GAT1 expression constitutive diminished the difference between wild-type and gzf3Δ glutamine-grown strains; and (iii) in vivo binding of Gzf3-Myc¹³ to P_GAT1 was elevated on glutamine, whereas binding of Gat1-Myc¹³ to its own promoter was weak. Second, the fact that GAP1 expression still responded to Gzf3 in cells constitutively expressing GAT1 indicates that Gzf3 also acts downstream of GAT1 expression.

Our results indicate that proline-induced Gzf3 binding to NCR-sensitive promoters occurs via Dal80- and Gat1-mediated recruitment or is stabilized by them. Attempts to identify a function for Gzf3 during growth on poor nitrogen sources gave contradictory results: in proline-grown cells, deletion of GZF3 reduced DAL5 and GAP1 expression (Fig. ​(Fig.6;6; also data not shown), suggesting that Gzf3 may carry out an activating function or be required to stabilize activating complexes when bound on some NCR-sensitive promoters. On the other hand, we have also demonstrated that Gzf3 could perform an inhibitory task at the promoter of DAL5 in glutamine-grown cells treated with rapamycin (19). Hints of a repressive function for Gzf3 have previously been reported in derepressive growth conditions for DAL80 (34) and CIS2 (35) regulation.

Although we have observed a concomitant increase of binding of the four GATA factors in proline-grown cells, it does not necessarily imply that all are bound to the same promoter at the same time. However, hints of protein-protein interaction have been provided throughout our work: (i) the fact that Gln3-Myc¹³ binding to DNA required Gat1 suggests that the latter may help in recruiting the former or in stabilizing its binding; (ii) Dal80 interference with transcriptional activation, besides its competitive effect with Gat1-Myc¹³ binding to DNA, suggests that it might bind simultaneously and alter the activator potency; and (iii) Dal80- and Gat1-mediated recruitment of Gzf3-Myc¹³ to NCR-sensitive promoters has been demonstrated, with no clearly established function for the latter during transcriptional activation. It is, however, conceivable that Gzf3 may relay transduction signals to Gat1, due to its demonstrated physical interactions with Ure2 (33) and Tor1 (5). However, these interaction data, obtained using the two-hybrid technique, must be taken with care since Ure2 is currently thought to be cytoplasmic, whereas Gzf3 is nuclear.