NPR1 Kinase and RSP5-BUL1/2 Ubiquitin Ligase Control GLN3-dependent Transcription in Saccharomyces cerevisiae*

The GATA transcription factors GLN3 and GAT1 activate nitrogen-regulated genes in Saccharomyces cerevisiae. NPR1 is a protein kinase that controls post-Golgi sorting of amino acid permeases. In the presence of a good nitrogen source, TOR (target of rapamycin) maintains GLN3 and NPR1 phosphorylated and inactive by inhibiting the type 2A-related phosphatase SIT4. We identified NPR1 as a regulator of GLN3. Specifically, loss of NPR1 causes nuclear translocation and activation of GLN3, but not GAT1, in nitrogen-rich conditions. NPR1-mediated inhibition of GLN3 is independent of the phosphatase SIT4. We also demonstrate that the E3/E4 ubiquitin-protein ligase proteins RSP5 and BUL1/2 are required for GLN3 activation under poor nitrogen conditions. Thus, NPR1 and BUL1/2 antagonistically control GLN3-dependent transcription, suggesting a role for regulated ubiquitination in the control of nutrient-responsive transcription.

The GATA transcription factors GLN3 and GAT1 activate nitrogen-regulated genes in Saccharomyces cerevisiae. NPR1 is a protein kinase that controls post-Golgi sorting of amino acid permeases. In the presence of a good nitrogen source, TOR (target of rapamycin) maintains GLN3 and NPR1 phosphorylated and inactive by inhibiting the type 2A-related phosphatase SIT4. We identified NPR1 as a regulator of GLN3. Specifically, loss of NPR1 causes nuclear translocation and activation of GLN3, but not GAT1, in nitrogen-rich conditions. NPR1-mediated inhibition of GLN3 is independent of the phosphatase SIT4. We also demonstrate that the E3/E4 ubiquitin-protein ligase proteins RSP5 and BUL1/2 are required for GLN3 activation under poor nitrogen conditions. Thus, NPR1 and BUL1/2 antagonistically control GLN3-dependent transcription, suggesting a role for regulated ubiquitination in the control of nutrient-responsive transcription.
In yeast, the quality of the available nitrogen source controls the expression of genes encoding proteins required for the uptake and assimilation of nitrogenous compounds. This regulatory process is mediated by four GATA transcription factors, the activators GLN3 and GAT1/NIL1 and the repressors DAL80 and DEH1/NIL2/GZF3. GLN3 is the major activator of nitrogen-regulated genes (for recent reviews, see Refs. 1 and 2). URE2 is a GLN3-binding protein that retains GLN3 in the cytoplasm to prevent its nuclear translocation and activation during growth under good nitrogen conditions. The two homologous kinases TOR1 1 and TOR2 are central controllers of cell growth, and their inactivation with rapamycin causes physiological changes characteristic of nutrientstarved cells (3,4). A TOR signaling pathway regulates the expression of nitrogen-regulated genes by inhibiting GLN3 and GAT1 (5). In the presence of a preferred nitrogen source such as NH 4 ϩ or glutamine, TOR promotes complex formation between GLN3 and URE2 by maintaining GLN3 in the phosphorylated state necessary for URE2 binding (5). Upon nitrogen limitation or rapamycin treatment, the TOR-controlled phosphatase SIT4 causes GLN3 dephosphorylation, and GLN3 dissociates from URE2 and translocates into the nucleus to activate target genes (5).
NPR1 encodes a Ser/Thr kinase that controls the post-Golgi sorting and degradation of amino acid permeases. NPR1 inversely regulates GAP1, the general amino acid permease, and TAT2, a tryptophan permease (2). In cells grown in poor nitrogen sources, NPR1 is dephosphorylated by the TOR-modulated SIT4 phosphatase to become active (6). In turn, active NPR1 allows GAP1 to reach the plasma membrane and causes the rapid degradation of TAT2. In cells shifted to a good nitrogen source medium, plasma membrane GAP1 is internalized and pre-existing permease is no longer delivered to the plasma membrane, whereas TAT2 is stabilized at the plasma membrane (7)(8)(9). GAP1 down-regulation occurs via its ubiquitination by the E3 ubiquitin protein ligase RSP5 in conjunction with the RSP5-binding proteins BUL1 and BUL2 (8,10,11).
Here, we identify NPR1 as a negative regulator of GLN3-dependent transcription. Deletion of NPR1 causes nuclear localization and activation of GLN3 during growth on a good nitrogen source. Activation of GLN3 in npr1 mutant cells requires the presence of functional RSP5 and BUL1/2 proteins, and BUL1/2 proteins are necessary for the activation of GLN3 under nitrogen starvation conditions. Thus, a ubiquitin-dependent signaling pathway appears to control GLN3-dependent transcription.

MATERIALS AND METHODS
Strains, Media, and Reagents-The complete genotypes of the yeast strains used in this study are listed in Table I. Standard techniques and media were used (22). The SD medium contained NH 4 ϩ as nitrogen source and the amino acids required for auxotrophies. All cultures were incubated at 30°C. Rapamycin was used at a final concentration of 200 ng/ml.
Plasmids-pJC10 (YCplac111-based), used to screen for regulators of GLN3, was obtained by PCR amplification of the MEP2 promoter (from nucleotide Ϫ827 to Ϫ1) and the HIS3 gene (from nucleotide 1 to 913). pJC20 expressing functional single HA-tagged NPR1 under control of own promoter was obtained by cloning a 2.2-kb fragment containing the NPR1 promoter and a 2.6-kb fragment from pAS103 (6) containing HA-NPR1 into YCplac33. pJC21 expressing kinase-dead single HAtagged NPR1 was constructed by replacing the 2.6-kb fragment from pJC20 by a 2.6-kb fragment from pAS104 containing HA-NPR1 with the mutation K467R. YCpMEP2-lacZ was described previously (12).
Screen for Regulators of GLN3-TB50 cells containing pJC10 were transformed with DNA from an mTn-3ϫHA/GFP/URA3 library, grown in SD medium for 2 h, and plated on SD-Ura-Leu-His medium supplemented with 5 mM 3-amino-1,2,4-triazole. Colonies that formed after 4 days incubation at 30°C were tested for growth on plates of SD medium without His containing 10 mM 3-amino-1,2,4-triazole. The genomic site of transposon insertion was identified following the "vectorette PCR" method (23).
Microarray Analysis-Total RNA was prepared by hot phenol extraction (24) from duplicate samples grown in SD medium to mid-logarith-mic phase. For microarray analysis, RNA was reverse transcribed, and cDNA was prepared using Superscript II reverse transcriptase (Invitrogen). The cDNA was used as template for the production of biotinlabeled cRNA using a RNA transcript labeling kit (Enzo). Finally, the labeled cRNA was purified using RNeasy spin columns (Qiagen), fragmented by incubation with potassium/magnesium acetate and hybridized to Affymetrix S98 Yeast GeneChips, as recommended by the manufacturer. Raw expression signals for each transcript were computed using the algorithm implemented in MAS 5.0 (Affymetrix). The robust multichip average method, implemented as a part of the BioConductor (bioconductor.org) package affy, was employed for data normalization, background correction, and summarization (25).
Determination of Intracellular Amino Acid Pools-The intracellular levels of amino acids were determined as described (26).

Identification of the NPR1 as a Negative
Regulator of Nitrogen-controlled Genes-To find new regulators of GATA transcription factors, we performed a transposon insertion screen to identify mutations in genes whose products are required to repress GLN3 and/or GAT1 under rich nitrogen conditions. A strain expressing the HIS3 gene from the GLN3-/GAT1-dependent promoter of the NH 4 ϩ permease MEP2 (P MEP2 -HIS3) (12) cannot survive in NH 4 ϩ medium lacking histidine because GLN3 and GAT1 are inactive, whereas e.g. his3 ure2 cells can grow under the same conditions because of the constitutive activation of at least GLN3 (Fig. 1A). We identified a mutation in NPR1 that allows growth of his3 P MEP2 -HIS3 cells on NH 4 ϩ medium lacking histidine (Fig. 1A), suggesting that NPR1 is normally required to repress GLN3 and/or GAT1 during growth in a good nitrogen source.
To confirm that npr1 causes expression of MEP2, we assessed the expression of a second reporter, P MEP2 -lacZ, in wild type, npr1, and ure2 cells. Expression from P MEP2 -lacZ was ϳ12 times higher in npr1 cells and ϳ25 times higher in ure2 cells compared with that from the wild-type strain (Fig. 1B).
NPR1 is a protein kinase, and we transformed npr1 cells with plasmids carrying NPR1 (pJC20) or npr1 kinϪ that encodes kinase-dead NPR1 (pJC21). Only the kinase-dead NPR1 failed to reduce P MEP2 -lacZ expression, suggesting that the kinase activity of NPR1 is required for the repression of MEP2 expression (Fig. 1B).
NPR1-mediated Inhibition of GLN3 Is Independent of LST4, LST8, and the Intracellular Level of Amino Acids-NPR1 is a protein kinase required for post-Golgi sorting of the general amino acid permease GAP1 and the three known ammonia permeases, MEP1/2/3, to the plasma membrane. It is therefore possible that the loss of NPR1 may indirectly activate GLN3 by altering amino acid and/or ammonia metabolism. However, under the conditions in which we measured GLN3 activity, GAP1 activity is extremely low regardless of expression levels (13), and even if GAP1 is expressed significantly higher in npr1 cells as compared with wild type, GAP1 would still not be expected to reach the plasma membrane (14).
To confirm this hypothesis, we considered other mutations known to affect Gap1p sorting to the plasma membrane. lst4 and lst8 mutations, like npr1, prevent GAP1 arrival at the plasma membrane under poor nitrogen conditions (15). We asked whether an lst4 mutation, like an npr1 mutation, affects GLN3 activity. Loss of LST4 failed to induce P MEP2 -lacZ expression (Fig. 3A). We demonstrated previously that cells depleted of LST8 express GLN3-regulated genes at levels similar to those of wild-type cells (16). Thus, although loss of NPR1, LST4, or LST8 can affect GAP1 sorting, loss of only NPR1 results in the activation of GLN3 under our experimental conditions.
Compared with wild type, cells lacking all three ammonia permeases exhibit a severe growth defect in media containing 1 mM NH 4 ϩ but grow normally on media containing 40 mM or more NH 4 ϩ. Thus, at 76 mM NH 4 ϩ , the conditions our cells experience, there are additional mechanisms to take up NH 4 ϩ ions, so loss of MEP1/2/3 activity caused by the loss of NPR1 does not appear to affect the availability of ammonia for intracellular metabolism. Furthermore, the npr1 strain exhibited the same very mild slow growth defect at 1 mM as at 40 mM NH 4 ϩ as assessed by colony size following serial dilution spotting onto solid minimal media (data not shown). Thus, it is unlikely that GLN3 activation is occurring primarily because of reduced intracellular levels of NH 4 ϩ . It is possible that GLN3 is activated in npr1 cells because of reduced intracellular amino acid concentrations resulting from altered trafficking of permeases other than GAP1 or MEP1/2/3. To rule out this possibility, we measured the concentration of amino acids in wild type and npr1 strains grown in minimal media (SD) containing only those amino acids required to correct for auxotrophies (His, Leu, and Trp). In the npr1 strain the total amount of all measured amino acids remained similar (Fig. 3B). Importantly, glutamine and glutamate levels were only slightly altered in the npr1 strain as compared with wild type, with glutamine levels slightly increasing and glutamate levels slightly decreasing. We demonstrated previously that GLN3 is activated by a decrease in glutamine (26). Thus, these minor alterations in amino acid concentrations in an npr1 strain do not correlate with the greatly increased activity of GLN3 in the same strain. Thus, the activation of GLN3 in npr1 cells does not appear to be due to an indirect effect of altered amino acid concentrations that could occur by inappropriate permease sorting.
NPR1-mediated Inhibition of GLN3 Is Independent of SIT4 -NPR1 is regulated in part by SIT4 phosphatase, which, in turn, is regulated by TOR (17). The SIT4 protein phosphatase is also required for GLN3 activation following rapamycin treatment, so we asked whether GLN3 activation in npr1 cells also requires SIT4. The expression of P MEP2 -lacZ in wild type and sit4 strains was similarly low, and in npr1 and npr1 sit4 strains it was similarly high (Fig. 4A). This epistasis of npr1 to sit4 suggests that NPR1 acts downstream of SIT4 in GLN3 repression. We also examined the phosphorylation state of GLN3 in npr1 mutant cells. In npr1 cells GLN3 appears to be phosphorylated as in wild-type cells, because rapamycin treatment increases the mobility of GLN3 (Fig. 4B). The simplest model compatible with these observation is that SIT4 regulates GLN3 via NPR1, but NPR1 does not regulate GLN3 directly. It has been shown that the GLN3 nuclear importin SRP1 is capable of binding only non-phosphorylated GLN3 in vitro, and it has thus been proposed that dephosphorylation of GLN3 is necessary for its nuclear translocation (18). However, in npr1 and ure2 cells GLN3 is phosphorylated and predominantly nuclear, suggesting that the phosphorylation state of GLN3 does not directly regulate its translocation. An alternative model consistent with these data is that GLN3 phosphorylation controls its binding to URE2 (5).
Activation of GLN3 in npr1 Mutants Requires RSP5, BUL1, and BUL2-NPR1 and the E3/E4 ubiquitin ligase complex composed of RSP5, BUL1, and BUL2 control nitrogen-regulated trafficking of GAP1 antagonistically (2). To investigate whether RSP5 and BUL1/2 are required for GLN3 activation in npr1 cells, we analyzed the expression of P MEP2 -lacZ in wild type or npr1 cells lacking RSP5 or BUL1/2. The absence of RSP5 or BUL1/2 alone had no effect on lacZ activity. However, the mutation of either RSP5 or BUL1/2 in npr1 mutant cells completely suppressed MEP2 expression (Fig. 5A), indicating that RSP5 and BUL1/2 are necessary for GLN3 activation in npr1 cells. The finding that rsp5 and bul1/2 mutations are epistatic to npr1 suggests that the ubiquitin ligase is downstream of the NPR1 kinase.
BUL1/2 Proteins Are Required for Activation of GLN3 upon a Shift to Proline but Not upon Rapamycin Treatment-To better define the antagonistic role of NPR1 and BUL1/2 in the control of nitrogen-regulated genes, we assessed the relative changes in mRNA levels of a subset of genes known to be regulated by GLN3 in wild type, npr1, and bul1 bul2 cells grown under three different nitrogen related conditions (see "Materials and FIG. 5. Derepression of MEP2 in an npr1 mutant requires RSP5, BUL1, and BUL2. Wild-type (wt) (TB50), npr1 (JC19-1a), bul1 bul2 (JC60-4b), and npr1 bul1 bul2 (JC59-12a) and wild-type (23344c), npr1 (30788a), rsp5 (27038a), and npr1 rsp5 (30788d) strains in the ⌺1278b genetic background were transformed with the YCpMEP2-lacZ. All strains were grown in SD medium to an A 600 of 0.5 and processed for ␤-galactosidase assay. 100% of ␤-galactosidase activity corresponds to 1319 Miller units. Methods"). Table II shows the relative changes of a subset of nitrogen source-regulated genes in the three strains mentioned above grown under the following conditions: (i) grown in a good nitrogen source (NH 4 ϩ ); (ii) shifted from a good to a poor nitrogen source for 30 min (NH 4 ϩ to proline); and (iii) grown in a good nitrogen source (NH 4 ϩ ) and treated with rapamycin for 30 min. As expected, the known GLN3-regulated genes in Table II were derepressed in npr1 cells and unaltered or down-regulated in bul1 bul2 as compared with wild-type, confirming the antagonistic effect of npr1 and bul1 bul2 mutations. These genes are also all derepressed upon shifting a wild type strain to either proline or treating with rapamycin, events known to increase GLN3-dependent transcription. Although we are preparing a comprehensive genome-wide study of NPR1/BUL-1/2-dependent gene expression, 2 the subset of known GLN3 regulated genes that we have selected here clearly demonstrates that NPR1 and BUL1/2 are antagonistically involved in the control of GLN3 activity.
Proline failed to induce a significant transcriptional response in bul1 bul2 cells (Table II), indicating that these ubiquitination factors are an integral part of the proline-sensing transcription pathway. Curiously, rapamycin-induced transcription of the genes in Table II in bul1 bul2 cells was similar to that of wild type cells. To further investigate the differential effect of proline and rapamycin, we examined the nuclear translocation of GLN3 in wild-type and bul1 bul2 cells following a shift to proline or rapamycin treatment. Consistent with the above findings, both conditions induced nuclear localization of GLN3 in wild type cells, but only rapamycin induced nuclear localization of GLN3 in a bul1 bul2 mutant (Fig. 6). Thus, the BUL1/2 proteins are required for nuclear localization of GLN3 upon a shift to proline but not upon rapamycin treatment, supporting previous suggestions that proline is sensed differently and suggesting that the RSP5/BUL1/2 complex is specifically involved in GLN3 regulation. Indeed, growth in proline specifically activates the rapamycin-sensitive proline utilization pathway, including the transcription factor PUT3 (19).  We have shown previously that NPR1 is controlled by TOR and the phosphatase SIT4 (17). Here, we show that the kinase NPR1 and the E3/E4 ubiquitin ligase complex composed of RSP5 and BUL1/2 have antagonistic roles in the nuclear import and activation of the transcription factor GLN3. NPR1 inhibits and the ubiquitin ligase promotes GLN3 activity. Furthermore, our finding that rsp5 (npi1) and bul1 bul2 mutations are epistatic to an npr1 mutation suggests that the ubiquitin ligase acts downstream of NPR1 in activating GLN3 (Fig. 7). Thus, this may constitute a novel ubiquitin-dependent signaling pathway controlling transcription. The membrane-associated transcription factors SPT23 and MGA1 are ubiquitinated and activated in an RSP5-dependent manner (20). However, we have been unable to detect GLN3 ubiquitination. Alternative possibilities are that the GLN3 inhibitor URE2 or the GLN3 nuclear import machinery is regulated by ubiquitination.
In yeast cells grown in a good nitrogen source, GLN3 is repressed by the nitrogen sensing TOR-SIT4 pathway (5). As shown in Fig. 7, we suggest that NPR1 acts downstream of SIT4 to repress GLN3 directly or indirectly. It has been proposed that, under nitrogen rich conditions, NPR1 is phosphorylated and thus inactive toward amino acid permeases (6,14). The results presented here suggest that phosphorylated NPR1 is active with regard to GLN3 repression. Thus phosphorylation of NPR1 may regulate substrate specificity rather than activity per se.
Several lines of evidence argue against the idea that npr1 mutation affects GLN3 activity solely indirectly, via altered nitrogen source uptake, due to incorrect nitrogen source permease sorting (21). First, a reduction in the cytoplasmic nitrogen source would be expected to activate both GLN3 and GAT1, and the npr1 mutation causes constitutive nuclear localization and activation of only GLN3 and not GAT1 (Fig. 2). Second, under the conditions we are using for these experiments (76 mM NH 4 ϩ ), even a strain lacking all three known ammonia permeases (mep1/2/3) can grow perfectly well (12), and our npr1 strain forms colonies of similar size on solid media containing concentrations of NH 4 ϩ ranging from 1 to 80 mM, suggesting that the availability of intracellular ammonia is not significantly compromised. Third, the other permeases that NPR1 is known to positively regulate are GAP1 and PUT4, and neither of these permeases are expected to be active under our experimental conditions regardless of their transcription; also, mutations in genes that also regulate GAP1, LST4 and LST8 do not cause GLN3 activation. Fourth, total and specific intracellular amino acid concentrations do not alter significantly in a npr1 strain as compared with the wild type (Fig. 3B). Thus, although we cannot absolutely rule out that there is some effect due to altered permease sorting, it seems likely that a more direct pathway is involved.
Importantly rapamycin but not proline induces GLN3 activation in a bul1 bul2 mutant. Thus, rapamycin-mediated inhibition of the TOR pathway can bypass the requirement for RSP5/BUL1/2 in GLN3 activation. Loss of the GLN3 cytoplasmic binding protein URE2 also circumvents the requirement for the ubiquitination machinery in GLN3 activation (data not shown). Thus, there appears to be two TOR-dependent inputs into GLN3 activation, depending on the nature of the signal. The first would be an RSP5-and BUL1/2-dependent input in response to proline. The second would be an RSP5-and BUL1/ 2-independent input in response to some other nitrogen source. The qualitative and/or quantitative differences between the signals transduced by these two inputs remains to be elucidated, but either can be sufficient for GLN3 activation. Because a ure2 mutation has a similar effect in a bul1 bul2 mutant as rapamycin treatment, the RSP5-BUL1/2-independent input may be the one controlling the binding of GLN3 to URE2.