gln3 Mutations Dissociate Responses to Nitrogen Limitation (Nitrogen Catabolite Repression) and Rapamycin Inhibition of TorC1

Background: Gln3 localization is hypothesized to be co-regulated by TorC1 and nitrogen limitation.

Results: gln3 mutations abrogating Gln3-Tor1 interaction abolish the Gln3 response to rapamycin without adversely affecting its response to nitrogen limitation.

Conclusion: Different Gln3 regions mediate responses to rapamycin and nitrogen limitation.

Significance: Controlled Gln3 localization occurs via two separable regulatory pathways, both of which are required for overall WT Gln3 control.

Abstract

The GATA family transcription activator, Gln3 responds to the nitrogen requirements and environmental resources of the cell. When rapidly utilized, “good” nitrogen sources, e.g., glutamine, are plentiful, Gln3 is completely sequestered in the cytoplasm, and the transcription it mediates is minimal. In contrast, during nitrogen-limiting conditions, Gln3 quickly relocates to the nucleus and activates transcription of genes required to scavenge alternative, “poor” nitrogen sources, e.g., proline. This physiological response has been designated nitrogen catabolite repression (NCR). Because rapamycin treatment also elicits nuclear Gln3 localization, TorC1 has been thought to be responsible for NCR-sensitive Gln3 regulation. However, accumulating evidence now suggests that GATA factor regulation may occur by two separate pathways, one TorC1-dependent and the other NCR-sensitive. Therefore, the present experiments were initiated to identify Gln3 amino acid substitutions capable of dissecting the individual contributions of these pathways to overall Gln3 regulation. The rationale was that different regulatory pathways might be expected to operate through distinct Gln3 sensor residues. We found that C-terminal truncations or amino acid substitutions in a 17-amino acid Gln3 peptide with a predicted propensity to fold into an α-helix partially abolished the ability of the cell to sequester Gln3 in the cytoplasm of glutamine-grown cells and eliminated the rapamycin response of Gln3 localization, but did not adversely affect its response to limiting nitrogen. However, overall wild type control of intracellular Gln3 localization requires the contributions of both individual regulatory systems. We also found that Gln3 possesses at least one Tor1-interacting site in addition to the one previously reported.

Introduction

The nitrogen catabolite repression (NCR)-sensitive² GATA transcription activator Gln3 is one of the most often used reporters of TorC1 signaling activity in Saccharomyces cerevisiae. As such, its regulation has been investigated in substantial detail. The intracellular localization of Gln3 and another GATA family transcription activator, Gat1, respond to both nitrogen limitation (NCR) and rapamycin treatment (TorC1-mediated regulation). The central question addressed in this work is whether NCR- and rapamycin-inhibitable, TorC1-mediated activity represent sequential steps of a single regulatory pathway or two independent regulatory mechanisms that work in concert to control the GATA family transcription factors.

When cells are cultured under nitrogen-rich conditions, Gln3 and Gat1 are sequestered in the cytoplasm, and transcription of the genes required to scavenge poor nitrogen sources is highly repressed (1–5). Ure2, a pre-prion protein, is required for this cytoplasmic sequestration and repressed transcription (6). The additional observation that Ure2 forms a complex with Gln3 in nitrogen-rich conditions led to the conclusion that formation of this complex is a central step in nitrogen-responsive GATA factor regulation (6–9). In contrast, during nitrogen limitation, Gln3 is not complexed with Ure2 and relocates, along with Gat1, to the nucleus where they activate transcription of the nitrogen catabolic genes (7, 8, 10, 11). This mechanism of regulation was long ago designated NCR (1).

The fact that Gln3 and Gat1 also relocate to the nucleus following rapamycin treatment led to an elegant model concluding that their nitrogen-responsive, i.e., NCR-sensitive, regulation was achieved through nitrogen-responsive control of the TorC1 serine/threonine kinase complex and the PP2A-type phosphatases it regulates (7, 8, 10, 11). In very gross outline, excess nitrogen was envisioned to activate TorC1, which in turn inhibited PP2A-type phosphatases. Loss of the phosphatase activities resulted in increased phosphorylation of Gln3, which complexed with Ure2. As a result of either phosphorylation or complexation with Ure2, Gln3 was unable to enter the nucleus and activate catabolic gene transcription. On the other hand, in limiting nitrogen, TorC1 was envisioned to be down-regulated, the PP2A-type phosphatases were thereby activated, and Gln3 was dephosphorylated and dissociated from Ure2. Dephosphorylated Gln3 could then enter the nucleus and activate transcription. Consistent with this model: (i) rapamycin treatment results in Gln3 dephosphorylation, (ii) TorC1 pathway phosphatase Sit4 is required for nuclear GATA factor localization in nitrogen-limited and rapamycin-treated cells (5, 7, 8, 12–19), and (iii) Tor1 has been shown to interact with the C-terminal region of Gln3 in a two-hybrid assay and to phosphorylate it in vitro (8, 12).

However, more recent evidence contends that NCR-sensitive and TorC1-dependent regulation of GATA factor localization and function may, in fact, be intersecting pathways of a larger network. Among the observations supporting this view: (i) nitrogen limitation-induced nuclear GATA factor localization exhibits markedly different PP2A requirements than does rapamycin treatment (20–31), and (ii) demonstrable Gln3 phosphorylation/dephosphorylation does not always correlate with its intracellular localization (20–31). Therefore, a major unresolved issue concerning the regulation of intracellular GATA factor localization is whether it unambiguously involves one mechanistic pathway or two. Stating the question in another way, is rapamycin-inhibited TorC1-mediated Gln3 regulation the means through which nitrogen-responsiveness (NCR) is exerted? One reason these questions remain somewhat open is that most existing work depends heavily upon inhibitors and mutations situated at various distances upstream of the Gln3 molecule whose localization is being measured, increasing the challenge of separating primary and secondary effects generated by the experimental perturbations. The problem is further exacerbated by the lability and complexity of the Gln3 molecule itself (20% of its residues are serine or threonine), which have largely frustrated direct mass spectroscopic analyses of its phosphorylation (32–35).

Given the conceptual difficulty of cleanly separating primary and secondary responses of GATA factor localization to upstream perturbations, as well as the technical problems involved in biochemical approaches, we elected to take a systematic genetic approach to the question of Gln3 regulation. The rationale is predicated on the reasoning that if Gln3 localization and function involve multiple regulatory pathways, the individual pathways might possess distinct mutable and hence potentially identifiable targets within the Gln3 molecule itself. In the present work we report serine substitutions that diminish the ability of nitrogen replete growth conditions to sequester Gln3 in the cytoplasm. These substitutions completely abolish the response of Gln3 to rapamycin but leave the NCR-sensitive response to limiting nitrogen untouched. We also show that Gln3 possesses at least one additional Tor1 interaction site beyond the one originally described.

MATERIALS AND METHODS

Strains and Culture Conditions

The S. cerevisiae strain used as the transformation recipient in which gln3 mutant plasmids were assayed was JK9-3da (Table 1). The protein interaction assays were performed in transformants of PJ69-4a (Table 1). Growth conditions were identical to those described in Tate et al. (28). Cultures (50 ml) were grown to mid-log phase (A_{600 nm} = 0.5) in YNB (without amino acids or ammonia) minimal medium containing the indicated nitrogen source at a final concentration of 0.1%. Appropriate supplements (120 μg/ml leucine and 20 μg/ml histidine and tryptophan) were added to the medium as necessary to cover auxotrophic requirements. Where indicated, the cells were treated with 200 ng/ml rapamycin for 20 min as described earlier.

TABLE 1.

Saccharomyces cerevisiae strains used in this work

Strain	Pertinent genotype	Parent	Complete genotype	Reference
JK9-3daa	Wild type		MATa, leu2-3,112, ura3-52, trp1, his4, rme1, HMLa	Refs. 45 and 46
TB50	Wild type	JK9-3da	MATa, leu2-3, 112, ura3-52, trp1, his3, rme1, HMLa	Ref. 7
TB123	WT Gln3-Myc13	TB50	MATa, leu2-3, 112, ura3-52, trp1, his4, rme1, HMLa, GLN3-MYC13[KanMX]	Ref. 7
RR215	ure2Δ	JK9-3da	MATa, leu2-3,112, ura3-52, trp1, his4, rme1, HMLa, ure2Δ::[KanMX]	This work
PJ69-4a	Transformation recipient	DGY63::171b	Mata, trp1-901, leu2-3,112, his3–200, gal4Δ, gal80Δ, GAL2-ADE, lys::GAL1-His3, met2::GAL7-lacZ	Ref. 47

^a JK9-3d was constructed by Jeanette Kunz (Michael Hall's laboratory). Joseph Heitman isolated MATa and MATα strains isogenic to JK9-3d by mating type switching. JK9-3da is a hybrid strain containing contributions from the following strains: S288c, a strain from the Oshima lab, and an unidentified strain from the Herskowitz lab. It was chosen because of its robust growth, sporulation efficiency, and good growth on galactose (GAL+). It may have a SUP mutation that allows translation through premature STOP codons and therefore produces functional alleles with many point mutations.

^b DGY63::171 is in the W303 strain background.

Plasmid Construction and Protein-Protein Interaction Assays

gln3 deletion and amino acid substitution mutants were constructed using standard PCR-based methods and the primer sets in Table 2. The Myc¹³ and ADH1 transcriptional terminator were derived from pKA62 (36). The template for all of the constructions was pRR536, which contained the wild type GLN3 gene, including its native promoter, fused in frame with Myc¹³ at the GLN3 translational stop codon.

TABLE 2.

Primer sets used to construct plasmids employed in this work

Plasmids	Sequence changes	Primer sets
pRR536	Gln31–730 (full-length wild type)	5′-CGCGGATCCTATACCAAATTTTAACCAATCCAATTCGTCAGCAATTGCT-3′
		5′-ATCCCCGCGGGACGTCAACTCCATAGAAGTGACTTTTCCG-3′
pRR622	Gln31–584	5′-CGCGGATCCCGATGAGGAGTACGATGCATTGCGCGAC-3′
		5′-ATCCCCGCGGGACGTCAACTCCATAGAAGTGACTTTTCCG-3′
pRR621	Gln31–565	5′-CGCGGATCCACGTTGGAAAGAATTTGATAATAGATTCTG-3′
		5′-ATCCCCGCGGGACGTCAACTCCATAGAAGTGACTTTTCCG-3′
pRR613	Gln31–542	5′-CGCGGATCCAATTCTTGGTGAGGATGCGACACTATTTCC-3′
		5′-ATCCCCGCGGGACGTCAACTCCATAGAAGTGACTTTTCCG-3′
pRR614	Gln31–600	5′-CGCGGATCCGTCCACTTGTTGCTGTTCGTGCAGTTGTTG-3′
		5′-ATCCCCGCGGGACGTCAACTCCATAGAAGTGACTTTTCCG-3′
pRR850a	Gln3S656D,S659D,S662D	5′CAATCGGCCGTCTTCAAGAAAAgatCATACCgacTTGTTAgacCAACAATTGCAGAACTCGGAG-3′
		5′-CGCGGATCCTATACCAAATTTTAACCAATCCAATTCGTCAGCAATTGCT-3′
		5′-CAATCGGCCGTGAGACACTCGGTGAATCTACAGGAG-3′
		5′-ATCCCCGCGGGACGTCAACTCCATAGAAGTGACTTTTCCG-3′
pRR928a	Gln3T641D,S645D,S647D,S649D,S652D,S653D	5′-caatcggccgatcGACatcCGGatcATCTACAGGatcATTAAAAATATCAAAATTTGGCTTTTGAGAC-3′
		5′-ATCCCCGCGGGACGTCAACTCCATAGAAGTGACTTTTCCG-3′
		5′-CAATCGGCCGgatgacAGAAAATCACATACCTCATTGTTATCACAAC-3′
		5′-CGCGGATCCTATACCAAATTTTAACCAATCCAATTCGTCAGCAATTGCT-3′
pRR922	Gln3S616D,S617D,S619D, S621DS624D,S631D	5′-CATCCAGATCTGTTGTTCCGATATTACCAAAACC-3′
		5′-TCTTGAGACACTCGGTGAATCTACAGGAGTATTAAAAATATCAAAATTTGGCTTTTGatcGACAAAATTTGATGATCTgtcATTT GTgtcAACgtcATTgtcATCATTCCAATTCTGTCTATTCGAATTCG-3′
pRR960	Gln3S616A,S617A,S619A,S621A,S624A,S631A	5′-TCTTGAGACACTCGGTGAATCTACAGGAGTATTAAAAATATCAAAATTTGGCTTTTGagcGACAAAATTTGATGATCTtgcATT TGTtgcAACggcATTggctgcATTCCAATTCTGTCTATTCG-3′
		5′-CATCCAGATCTGTTGTTCCGATATTACCAAAACC-3′
pRR966a	Gln3T641A,S645A,S647A,S649A,S652A,S653A	5′-CAATCGGCCGtgcGACagcCGGtgcATCTACAGGagcATTAAAAATATCAAAATTTGGC-3′
		5′-ATCCCCGCGGGACGTCAACTCCATAGAAGTGACTTTTCCG-3′
		5′-CAATCGGCCGgctgcaAAGAAAATCACATACCTCATTGTTATCACAAC-3′
		5′-CGCGGATCCTATACCAAATTTTAACCAATCCAATTCGTCAGCAATTGCT-3′
pRR1014	Gln3S603D,T605D,T607D,S609D	5′-CGCGGATCC TATACCA AATTTTAACC AATCCAATTC GTCAGCAATT GCT-3′
		5′-CGCAATGCATCGTACTCCTCATCGTTTATGGCTGCGTCTTTGCAACAACTGCACGAACAGCAACAAGTGGACGTGAATgacAA CgacAACgacAATgacAATAGACAGAATTGGAC-3′
pRR1038	Gln3S656A,S659A,S662A	5′-CGCGGATCC TATACCA AATTTTAACC AATCCAATTC GTCAGCAATT GCT-3′
		5′-GTAGATTCACCGAGTGTCTCAAGACCTTCTTCAAGAAAAgcaCATACCgcaTTGTTAgcaCAACAATTGCAGAACTCG-3′
pRR1065	Gln31–730	5′-CATGCCATGGAGATGCAAGACGACCCCGAAAATTCGAAGCTG-3′
		5′-CGCGGATCC TATACCA AATTTTAACC AATCCAATTC GTCAGCAATT GCT-3′
pRR1067	Gln31–600	5′-CATGCCATGGAGATGCAAGACGACCCCGAAAATTCGAAGCTG-3′
		5′-CGCGGATCCGTCCACTTGTTGCTGTTCGTGCAGTTGTTG-3′
pRR1069	Gln31–730 S656D,S659D,S662D	5′-CATGCCATGGAGATGCAAGACGACCCCGAAAATTCGAAGCTG-3′
		5′-CGCGGATCC TATACCA AATTTTAACC AATCCAATTC GTCAGCAATT GCT-3′
pRR1098	Gln3654–670	5′-CATGGAGAGAAAATCACATACCTCATTGTTATCACAACAATTGCAGAACTCGGAGTCGGG-3′
		5′-GATCCCCGACTCCGAGTTCTGCAATTGTTGTGATAACAATGAGGTATGTGATTTTCTCTC-3′
pRR1101	Gln3654–670 S656D,S659D,S662D	5′-CATGGAGAGAAAAgacCATACCgacTTGTTAgacCAACAATTGCAGAACTCGGAGTCGGG-3,
		5′-GATCCCCGACTCCGAGTTCTGCAATTGTTGgtcTAACAAgtcGGTATGgtcTTTTCTCTC-3′
pRR1160	Gln3654–670 S656A,S659A,S662A	5′-CATGGAGAGAAAAgcaCATACCgcaTTGTTAgcaCAACAATTGCAGAACTCGGAGTCGGG-3,
		5′-GATCCCCGACTCCGAGTTCTGCAATTGTTGtgcTAACAAtgcGGTATGtgcTTTTCTCTC-3′

^a Sequence Gln3_1947–1952, 5′-AGACCT-3′, was changed to 5′-CGGCCG-3′ to generate a unique EagI restriction site for cloning purposes. These changes did not alter the amino acid sequence of the protein.

PCR fragments using pRR536, pRR614, and pRR850 as template were generated and cloned into interaction vectors pACT-2 (Clontech) to yield GAL4p(AD)-gln3 fusion plasmids, pRR1065, pRR1067, and pRR1069, respectively. Three plasmids (pRR1098, pRR1101, and pRR1160) containing the 17-amino acid sequence of the Gln3_654–670 peptide were constructed by cloning synthetic double-stranded olignonucleotides into the pACT-2 vector. All of the constructs were confirmed by DNA sequencing (University of Tennessee Health Science Center Molecular Resource Center DNA sequencing facility).

GAL4

BD pASTOR(1–2470) and pASTOR1 (1–1764) were generously provided by Prof. Stephen Zheng (12). Positive control plasmids pTD1-1 (SV40 large T antigen) and pLAM5 (human lamin C) were obtained from Clontech. The Gal4p-BD and AD fusion proteins were expressed together in yeast strain PJ69-4a. This strain was constructed in the W303 genetic background. We used it for our Tor1 interaction assays rather than a JK9-3da-derived strain so that Tor1-association data obtained in the present work could be directly compared with that from Carvalho and Zheng (12).

The protein-protein interaction assays we used were those generously provided by Carvalho and Zheng and measured growth of PJ69-4a transformants on synthetic complete medium (SC) lacking leucine, tryptophan, and adenine as the positive control or synthetic complete medium lacking leucine, tryptophan, and histidine plus 3 mm (final concentration) 3-amino-1,2,4-triazole (3AT) to assess whether or not a positive interaction occurred.

Western Blot Analyses

Western blot analyses were performed using a combination of methods described earlier (22, 28, 29, 37). The antibodies used to visualize Gln3-Myc¹³ were primary monoclonal antibody 9E10 (c-Myc) (Covance MMS-150P) and secondary goat anti-mouse IgG (H-L)-horseradish peroxidase conjugate (Bio-Rad). Western blot results were recorded on Kodak BioMax XAR film. A wide range of exposures were collected for each sample, and the brightness settings of the final images were uniformly changed to ensure that no minor bands were lost and that the final images reproduced the actual x-ray films as closely as possible.

Indirect Immunofluorescence Microscopy

Cell collection and immunofluorescent staining were performed as previously described (21, 23, 26, 28). The cells were fixed by the addition of 0.55 ml of 1 m potassium phosphate buffer (pH 6.5) and 0.5 ml of 37% formalin to a 5-ml aliquot of the culture to be assayed. This was followed by incubation for 80 min at 30 °C. Gln3-Myc¹³ was visualized using 9E10 (c-Myc) (Covance MMS-150P) monoclonal antibody and Alexa Fluor 594 goat anti-mouse IgG (Invitrogen Molecular Probes) antibodies. DNA was stained with DAPI. The cells were imaged using a Zeiss Axioplan 2 imaging microscope with a 100× Plan-Apochromat 1.40 oil objective at room temperature. The images were acquired using a Zeiss Axio camera and AxioVision 3.0 and 4.8.1 (Zeiss) software and processed with Adobe Photoshop and Illustrator programs. Level settings (shadow and highlight only) were altered where necessary to avoid any change or loss in cellular detail relative to what was observed in the microscope; changes were applied uniformly to the image presented and were similar from one image to another.

Determination of Intracellular Gln3-Myc¹³ Distribution

It is impossible to obtain a quantitative estimate of Gln3-Myc¹³ localization viewing a single image containing 4–10 cells unless Gln3 is overwhelmingly in a single cellular compartment. This is due to both uneven distributions of cells exhibiting mixed Gln3 localizations (e.g., cytoplasmic and nuclear-cytoplasmic) and subjectivity in image selections. Therefore, we quantitated intracellular Gln3-Myc¹³ localization by manually scoring Gln3-Myc¹³ localization in 200 or more cells in multiple, randomly chosen fields from which each image presented was taken. Scoring was performed exclusively using unaltered, primary .zvi image files viewed with Zeiss AxioVision 3.0 and 4.8.1 software.

Cells were classified into one of three categories in the Gln3-Myc¹³ localization figures (Figs. 1–8) of this work: cytoplasmic (cytoplasmic Gln3-Myc¹³ fluorescence only; red bars), nuclear-cytoplasmic (Gln3-Myc¹³ fluorescence appearing in the cytoplasm, as well as co-localizing with DAPI-positive material; yellow bars), and nuclear (Gln3-Myc¹³ fluorescence co-localizing only with DAPI-positive material; green bars). A representative collection of “standard” images demonstrating the differences in these categories is shown in Fig. 2 of Ref. 28 along with a description of how the criteria were applied. Following recent recommendations (38), we assessed the precision of our scoring by analyzing the data from 10 different experiments performed over a 9-month period; all appear in this work. The average values and standard deviations observed for the five conditions we assayed appear in Fig. 1. These data add additional support to previous measurements by Tate et al. (23, 25, 29). Similar experiments were repeated two or more times with similar results. Images accompanying the histograms were selected on the basis that they exhibited intracellular Gln3-Myc¹³ distributions as close as possible to those observed in the quantitative scoring data.

FIGURE 1.

Evaluation of new assays used to assess the intracellular distribution of wild type Gln3_1–730-Myc¹³ expressed from CEN-based plasmids. The histograms for each of the conditions assayed in these experiments are the averages of 10 experiments performed with CEN-based pRR536 over a 9-month period (A) or four experiments performed in wild type TB123 in which Gln3-Myc¹³ is situated at its native chromosomal position over a 3-year period (B). The error bars represent plus or minus one standard deviation. pRR536 transformants and TB123 were grown in YNB medium with glutamine (Gln) or proline (Pro) as the sole nitrogen source. Where indicated (+Rap), the cultures were treated with rapamycin. Red bars indicate Gln3-Myc¹³ staining in the cytoplasm only, yellow bars indicate both cytoplasmic and nuclear Gln3-Myc¹³ staining, and green bars indicate Gln3-Myc¹³ staining in the nucleus only. Panels of standard images illustrating the characteristics of each scoring category appear in Fig. 2 of Ref. 28. Strain JK9-3da was the transformation recipient. Nucl.-Cyto., nuclear-cytoplasmic.

FIGURE 2.

Localization of full-length Gln3_1–730-Myc¹³ or Gln3_1–600-Myc¹³, Gln3_1–584-Myc¹³, Gln3_1–565-Myc¹³, and Gln3_1–542-Myc¹³ truncations in untreated and rapamycin-treated, glutamine-grown transformants. A depicts representative images from which the corresponding histograms in B were generated. C depicts the ratio of cytoplasmic staining untreated versus rapamycin-treated (+Rap), glutamine-grown cells for each of the truncation proteins analyzed in B. Red bars indicate Gln3-Myc¹³ staining in the cytoplasm only, yellow bars indicate both cytoplasmic and nuclear staining, and green bars indicate staining in the nucleus only. Strain JK9-3da was the transformation recipient. Nucl.-Cyto., nuclear-cytoplasmic.

FIGURE 3.

Localization of full-length Gln3_1–730-Myc¹³ or Gln3_1–600-Myc¹³, Gln3_1–584-Myc¹³, Gln3_1–565-Myc¹³, and Gln3_1–542-Myc¹³ truncations in glutamine-grown (Gln) or proline-grown (Pro) transformants. The experimental format was the same as in Fig. 2 except for the addition of a third growth condition: proline provided as the sole nitrogen source. Nucl.-Cyto., nuclear-cytoplasmic. Panel A depicts representative images from which the corresponding histograms in B were generated.

FIGURE 4.

Intracellular localization of full-length Gln3_1–730-Myc¹³ (pRR536) and Gln3_{S603D,T605D,T607D,S609D}-Myc¹³ (pRR1014) in untreated glutamine-grown (Gln) or proline-grown (Pro) transformants or in rapamycin-treated (+Rap) glutamine-grown cells. A, Gln3 amino acid sequence 601–670 showing the locations of serine/threonine substitutions in the plasmids we used in this study, as well as the location of sequences with the potential to form an α-helix. B and C, the experimental format was the same as in Fig. 2 except for the addition of a third growth condition: proline provided as the sole nitrogen source. Nucl.-Cyto., nuclear-cytoplasmic.

FIGURE 5.

Localization of full-length Gln3_1–730-Myc¹³ (pRR536), Gln3_{S616D,S617D,S619D,S621D,S631D}-Myc¹³ (pRR922), and Gln3_{S616A,S617A,S619A,S621A,S624A,S631A}-Myc¹³ (pRR960) in untreated glutamine-grown (Gln in A–D) or proline-grown (Pro in C and D) transformants or in glutamine-grown, rapamycin-treated (+Rap in A and B) transformants. The experimental format was the same as in Fig. 2 except for the addition of a third growth condition: proline provided as the sole nitrogen source. Nucl.-Cyto., nuclear-cytoplasmic.

FIGURE 6.

Localization of full-length Gln3_1–730-Myc¹³ (pRR536), Gln3_{T641D,S645D,S647D,S649D,S652D,S653D}-Myc¹³ (pRR928), and Gln3_{T641A,S645A,S647A,S649A,S652A,S653A}-Myc¹³ (pRR966) in untreated glutamine-grown (Gln in A–D) or proline-grown (Pro in C and D) transformants or in rapamycin-treated (+Rap in A and B) transformants. The experimental format was the same as in Fig. 2 except for the addition of a third growth condition: proline provided as the sole nitrogen source. Nucl.-Cyto., nuclear-cytoplasmic.

FIGURE 7.

Localization of full-length Gln3_1–730-Myc¹³ (pRR536), Gln3_{S656D,S659D,S662D}-Myc¹³ (pRR850), and Gln3_{S656A,S659A,S662A}-Myc¹³ (pRR1038) in untreated glutamine-grown (Gln in A–D) and proline-grown (Pro in C and D) transformants or in rapamycin-treated (+Rap in A and B) transformants. The experimental format was the same as in Fig. 2 except for the addition of a third growth condition: proline provided as the sole nitrogen source. Nucl.-Cyto., nuclear-cytoplasmic.

FIGURE 8.

Effect of deleting URE2 on the localization of full-length Gln3_1–730-Myc¹³ (pRR536), Gln3_1–600-Myc¹³ (pRR614), and Gln3_{S656D,S659D,S662D}-Myc¹³ (pRR850) in untreated and rapamycin-treated (+Rap), glutamine-grown transformants. The experimental format was the same as in Fig. 2. Strain RR215 was the transformation recipient. Nucl.-Cyto., nuclear-cytoplasmic. Panel A depicts representative images from which the corresponding histograms in B were generated.

RESULTS

Use of Gln3 Structural Mutants to Dissect Gln3 Regulation

To investigate the individual contributions of TorC1 and NCR to the regulation of Gln3 localization, we decided to use the Gln3 molecule itself not only as a reporter but also as a more direct probe to separate these regulatory pathways. However, the intractability of Gln3 to mass spectral analysis of its phosphorylation (32–35) prompted an alternative genetic approach to pursue this objective. The magnitude of the overall task—Gln3 contains 146 potentially phosphorylated serine/threonine residues—made it untenable to study the entire molecule at once. Therefore, we have first focused on its C-terminal region. This is the region reported to be associated with Tor1 and the one about which the least is known. It also possesses the additional asset of being distanced from known Gln3 basic functional domains, i.e., nuclear import and export, DNA binding, transcriptional activation, and Ure2 interaction, situated in the N-terminal portion of the protein (8, 12, 19).

Before such genetic analyses could be confidently pursued, three important technical issues required evaluation. Taking full advantage of Gln3 amino acid substitution mutant proteins now and in the future required the capability of expeditiously introducing them into multiple wild type, TorC1 pathway, and other mutant strains. This goal is most efficiently accomplished using CEN plasmid-borne constructs. Even though CEN-based expression systems are relatively stable and the Myc¹³ tag has been repeatedly shown not to adversely affect Gln3 function, two pertinent questions remained: (i) Would plasmid-borne mutants in which GLN3 was expressed from its own promoter yield sufficiently reproducible data to be useful? (ii) Would the results obtained be the same as when GLN3 was situated in its native chromosomal location? To address these questions, we constructed a CEN plasmid-borne wild type GLN3_1–730 gene driven by its native promoter (pRR536) and surveyed the range of conditions that regulate Gln3 localization: (i) highly repressive, nitrogen-replete conditions (glutamine-grown) where Gln3_1–730-Myc¹³ is completely cytoplasmic, (ii) derepressive, nitrogen-limited conditions (proline-grown) where it is completely nuclear, and (iii) following rapamycin treatment in nitrogen-rich media where it is localized to both cellular compartments.

First, we assessed the reproducibility of our assay. The data depicted in Fig. 1A represent the averaged results of 10 experiments performed over a 9-month period, measuring Gln3_1–730-Myc¹³ localization in wild type JK9-3da transformed with wild type Gln3_1–730-Myc¹³ pRR536. The maximum standard deviation observed with these data was approximately ±7–8%. It is important to note that the most challenging conditions to assay are those in which Gln3_1–730-Myc¹³ localizes to all three scoring categories, i.e., cytoplasmic, nuclear-cytoplasmic, and nuclear. These data supported the contention that our plasmid-based assays exhibited sufficient precision to be useful.

Second, we queried whether the plasmid-borne Gln3_1–730-Myc¹³ reporter (pRR536) yielded results similar to those obtained with a strain in which Gln3-Myc¹³ was situated at its native chromosomal position (TB123). Fig. 1B depicts data from four experiments performed over 3 years with TB123. Note that TB123 was derived from and is nearly isogenic to JK9-3da (Table 1). The only difference noted in data obtained by the two methods was a small but consistent nuclear shift of plasmid-derived Gln3_1–730-Myc¹³ compared with that from chromosomal Gln3_1–730-Myc¹³ (Fig. 1A, pRR536, and Fig. 1B, TB123).

The final technical issue was the nature of the transformation recipient. Two choices existed: wild type cells containing an unaltered chromosomal GLN3 allele in addition to the plasmid borne gln3-Myc¹³ mutant allele or a gln3Δ in which the plasmid-borne gln3-Myc¹³ mutant allele was the only one present in the cell. From extensive previous experience, we were well aware that diminishment or loss of Gln3 activity has far reaching secondary effects on many cellular processes. Moreover, there is no evidence that we know of demonstrating that regulation of Gln3 function is exerted through the amount of Gln3 produced unless it is ectopically overproduced at very high levels (39).³ This derives from the fact that Ure2, the titratable negative regulator of Gln3 and Gat1 (9), is present in the cell at much higher concentrations than Gln3. Given this information, we decided it was most prudent to use a wild type GLN3 strain (JK9-3da) as the transformation recipient, thereby divorcing potential secondary effects caused by Gln3 deficiency in some mutants but not others from primary effects on its localization. Comparison of data in Fig. 1 supports the validity of this decision. Together, the results demonstrated that we could confidently make meaningful measurements of Gln3-Myc¹³ localization in this new genetic context.

Analysis of the Gln3 Region Reported to Bind Tor1

In well performed experiments, Carvalho and co-workers (8, 12) reported that Gln3_510–730-Myc⁹ and Gln3_600–730-Myc⁹ peptides were sufficient to support association with Tor1 in a two-hybrid interaction assay, whereas a Gln3_667–730-Myc⁹ peptide was not. Micrographic data positively correlated with the two-hybrid results (12). Together these data led to the conclusion that Tor1 kinase interacted with the Gln3_600–667-Myc⁹ region and was responsible for nitrogen-responsive Gln3 regulation. However, we noted that the reported localization results were qualitative being derived from overall evaluation of isolated micrographs in which Gln3-Myc⁹ was concluded to be either nuclear or cytoplasmic. Such qualitative evaluation would have limited or abrogated any ability to detect intermediate changes in Gln3-Myc⁹ localization if they occurred. These limitations prompted us to re-evaluate the extent to which Tor1 could account for the totality of nitrogen-responsive Gln3 regulation. If inactivation of Tor1, either as a result of nitrogen limitation or rapamycin-mediated inhibition, was the sole determinant responsible for nitrogen-responsive Gln3 regulation, then elimination of or specific substitutions in Gln3 residues 600–667 should result in highly nuclear Gln3 localization similar to levels that we and others have reported in derepressed, proline-grown cells. If, on the other hand, control of nitrogen-responsive Gln3 localization derived from the combined effects of two mechanistic pathways acting together, then truncation of or amino acid substitutions in Gln3 might be expected to damage one of these regulatory mechanisms, but not the other.

We initially tested the above predictions rather crudely by sequentially truncating the C terminus of Gln3. We included in our analyses a construct, Gln3_1–600-Myc¹³ (pRR614), analogous to one previously characterized by Carvalho and Zheng (12); we considered it to be a negative control. The data we obtained were both intriguing and provocative. From the data of Carvalho and Zheng, we expected Gln3 to become exclusively nuclear in wild type cells containing the truncated, glutamine-grown negative control (Gln_1–600-Myc¹³). This did not occur. Although Gln3_1–600-Myc¹³ was more nuclear than wild type Gln3_1–730, thus supporting the qualitative observation of Carvalho and Zheng, a substantial amount of it, roughly two-thirds to three-quarters, remained either cytoplasmic or nuclear-cytoplasmic when glutamine was provided as the sole nitrogen source (Fig. 2, A and B). Even when Gln3 was more drastically truncated to residues 584 (pRR622), 565 (pRR621), and 542 (pRR613), no further nuclear Gln3-Myc¹³ localization occurred (Fig. 2, A and B). In other words, these truncations were insufficient to relocate Gln3-Myc¹³ completely into the nucleus, thus arguing that at least two regulatory mechanisms were required to account for total control of Gln3 localization.

This conclusion was further supported when the data were viewed from the opposite perspective. The localizations of Gln3_1–600-Myc¹³, Gln3_1–584-Myc¹³, and Gln3_1–542-Myc¹³ were much less cytoplasmic than observed for glutamine-grown, full-length Gln3_1–730-Myc¹³ (Fig. 2, A and B). This indicated that although Gln3_600–730 materially contributed to sequestering Gln3-Myc¹³ in the cytoplasm, these residues and their reported Tor1 interaction could account for only a portion of the cytoplasmic Gln3-Myc¹³ sequestration observed with Gln3_1–730-Myc¹³.

We observed a second, equally important characteristic of the Gln3_1–600-Myc¹³, Gln3_1–584-Myc¹³, and Gln3_1–542-Myc¹³ truncations. Treating them with rapamycin failed to further increase their nuclear localization. In all cases, the ratio of cytoplasmic Gln3-Myc¹³ in the presence versus the absence of rapamycin decreased from 3:1 in wild type to 1:1 in all of the mutants (Fig. 2C). We also noted that the overall intracellular distributions of the Gln3_1–600-Myc¹³, Gln3_1–584-Myc¹³, and Gln3_1–542-Myc¹³ truncations in the presence or absence of rapamycin were the same as that seen with full-length Gln3_1–730-Myc¹³ in wild type glutamine-grown, rapamycin-treated cells (Fig. 2B).

From both perspectives, the above observations suggested that high level cytoplasmic sequestration of wild type Gln3-Myc¹³ likely derived from two regulatory mechanisms. One depended on Gln3 residues 600–730, which were also responsible for the rapamycin responsiveness of Gln3-Myc¹³ localization. The second mechanism was not rapamycin-responsive and remained unaffected even when Gln3 was truncated by as much as ∼25%.

We hypothesized that if this second mechanism derived from NCR, then the cytoplasmic Gln3-Myc¹³ sequestration observed with the truncated forms of Gln3-Myc¹³ in Fig. 2 should continue to be nitrogen-responsive, i.e., when cells were provided with a derepressive nitrogen source (proline), the cytoplasmic and nuclear-cytoplasmic Gln3-Myc¹³ should relocate into the nucleus. We tested this expectation by comparing localization of the truncated versions of Gln3-Myc¹³ in repressive versus derepressive conditions, i.e., glutamine-grown cells versus proline-grown cells (Fig. 3). All four Gln3 truncations exhibited wild type NCR-sensitive responses. Derepressive growth conditions, using proline as the sole nitrogen source, resulted in nearly complete nuclear Gln3_1–600-Myc¹³, Gln3_1–584-Myc¹³, Gln3_1–565-Myc¹³, and Gln3_1–542-Myc¹³ localizations. These observations supported the existence of a second regulatory mechanism and additionally demonstrated that the Gln3 target for it was N-terminal of Gln3₅₄₂.

Identification of Gln3 Residues Required for the Rapamycin Response

Although the above truncations dissected rapamycin-inhibitable, Tor1 association-dependent responses of Gln3-Myc¹³ localization away from responses that were NCR-sensitive, they represented rather blunt analytical probes. Removing up to one-quarter of the Gln3 protein (130–188 residues) would not permit us to rigorously conclude that both diminished cytoplasmic sequestration of Gln3-Myc¹³ in nitrogen-replete conditions and the loss of rapamycin-responsiveness derived solely from the loss of a Tor1-association site; much higher resolution was necessary.

Using systematic amino acid substitutions, we achieved the desired degree of resolution and addressed whether substituting nonphosphorylatable or phosphomimetic alanine or asparate residues, respectively, for serine/threonine residues in the C-terminal region of Gln3 affected its localization or the ability of TorC1 and/or rapamycin treatment to influence it. We sequentially replaced small groups of serine/threonine residues in the Gln3_600–667 region that Carvalho and Zheng (12) reported to contain a Tor1-binding domain (Fig. 4A). All of our substitutions were situated in full-length Gln3 proteins whose production was regulated by the native GLN3 promoter. We reasoned that: (i) substituting phosphomimetic aspartate for serine/threonine residues directly participating in TorC1 control of Gln3 would elicit constitutively cytoplasmic Gln3-Myc¹³ localization, (ii) substituting nonphosphorylatable alanine residues would shift Gln3-Myc¹³ constitutively into the nucleus, and (iii) in instances where Gln3 structure per se, rather than specific phosphorylation, was altered by the substitutions, the aspartate and alanine substitutions would generate similar results.

The first cluster of substitutions, Gln3_{S603D,T605D,T607D,S609D}-Myc¹³ (pRR1014) had little demonstrable effect on Gln3-Myc¹³ localization either in the presence or in the absence of rapamycin (Fig. 4, B and C; compare wild type Gln3_1–730-Myc¹³ (pRR536) with Gln3_{S603D,T605D,T607D,S609D}-Myc¹³ (pRR1014), Gln versus Gln +Rap). Further, the NCR-sensitive response was similarly unaffected (Fig. 4, B and C; compare wild type and mutant responses to glutamine versus proline). We next assayed Gln3 localization in Gln3_S616D,S617D,_{S619D,S621D,S624D,S631D}-Myc¹³ (pRR922) and Gln3_{S616A,S617A,S619A,S621A,S624A,S631A}-Myc¹³(pRR960) (Fig. 5). Again, there was virtually no change from the wild type phenotypes (Fig. 5). Moving further toward the C terminus, we then assayed Gln3_{T641D,S645D,S647D,S649D,S652D,S653D} (pRR928) and Gln3_{T641A,S645A,S647A,S649A,S652A,S653A}-Myc¹³ (pRR966), and yet again, there was no change from the wild type phenotype (Fig. 6). This was remarkable. Substituting aspartate for 25% or 16 of 53 residues, four to six at a time, and thereby substantially increasing the acidity of the region yielded no demonstrable phenotype.

Finally, we assayed Gln3_{S656D,S659D,S662D}-Myc¹³ (pRR850) and Gln3_{S656A,S659A,S662A}-Myc¹³ (pRR1038) (Fig. 7). The response was crystal clear; the phenotypes of both the aspartate (Gln3_{S656D,S659D,S662D}, pRR850) and alanine (Gln3_{S656A,S659A,S662A}, pRR1038) substitutions were indistinguishable. They mirrored the phenotype of the original Gln3_1–600-Myc¹³ truncation (compare Figs. 2 and 3 with Fig. 7). Cytoplasmic Gln3-Myc¹³ localization decreased from over 90% in the wild type Gln3_1–730-Myc¹³ pRR536 to 30–40% in the mutants. Rapamycin treatment was without effect, yet the NCR response was wild type in Gln3_{S656D,S659D,S662D} (pRR850) and Gln3_{S656A,S659A,S662A} (pRR1038), i.e., Gln3-Myc¹³ shifted completely into the nuclei of proline-grown cells (Fig. 7).

In sum, only the very end of Gln3_600–667 was required and, within the resolution we could measure, accounted for all of the phenotypic characteristics observed in our Gln3_1–600-Myc¹³, Gln3_1–584-Myc¹³, Gln3_1–565-Myc¹³, and Gln3_1–542-Myc¹³ truncations, as well as those reported by Carvalho and Zheng (12) (Gln3_1–600-Myc⁹, Gln3_510–730-Myc⁹, and Gln3_600–667-Myc⁹). The fact that Gln3_{S656D,S659D,S662D}-Myc¹³ (pRR850) and Gln3_{S656A,S659A,S662A}-Myc¹³ (pRR1038) exhibited the same intracellular distribution profiles indicated that the effects most likely derived from alteration of the Gln3_656–662 structure itself rather than from specific phosphorylation or dephosphorylation of the substituted residues acting as the direct determinant of Gln3 localization. It was additionally significant that one could dramatically increase the acidity of Gln3 over a 50-residue region without effect on its regulated localization.

Cytoplasmic Sequestration of Gln3-Myc¹³ in Strains Altered in Gln3_600–666 Retain an Absolute Ure2 Requirement

Substitution of just three serine residues appeared to dissect the contribution of rapamycin-inhibited, i.e., TorC1-dependent regulation of Gln3-Myc¹³ localization away from that imposed by NCR. If the Tor1-independent, second regulatory mechanism controlling Gln3 localization was truly NCR, then the cytoplasmic sequestration of the Gln3 truncation and Gln3_{S656,S659,S662} substitution mutants should be Ure2-dependent. As noted earlier, Ure2 had long been known to: (i) be a negative regulator of Gln3 function (1–5), (ii) form a Gln3-Ure2 complex (6–9), and (iii) be required to maintain cytoplasmic Gln3 localization in nitrogen-replete medium (6–9). To test this expectation, CEN plasmids expressing Gln3_1–730-Myc¹³ (pRR536), Gln3_1–600-Myc¹³ (pRR614), and Gln3_{S656D,S659D,S662D}-Myc¹³ (pRR850) were transformed into ure2Δ RR215, and Gln3-Myc¹³ localization was assayed. The loss of Ure2 resulted in complete nuclear localization of Gln3-Myc¹³ carried in all three plasmids transformed into glutamine-grown cells (Fig. 8). Thus, not only was the second Gln3 regulatory mechanism nitrogen source responsive, it also exhibited an absolute requirement for Ure2, both hallmarks of NCR-sensitive control.

Changes in Rapamycin-Elicited Gln3_1–730-Myc¹³_, Gln3_1–600-Myc¹³ Gln3_{S656D,S659D,S662D}-Myc¹³ and Gln3_{S656A,S659A,S662A}-Myc¹³ Phosphorylation

Treating glutamine- or YPD-grown wild type cells with rapamycin results not only in nuclear Gln3 localization, but also in gross Gln3 dephosphorylation thereby increasing its migration rate in SDS-PAGE gels (7, 8, 22). This was one of the foundational correlations from which the earliest model for Tor1-mediated Gln3 regulation derived (7). Therefore, the inability of Gln3_1–600-Myc¹³ (pRR614), Gln3_{S656D,S659D,S662D}-Myc¹³ (pRR850) and Gln3_{S656A,S659A,S662A}-Myc¹³ (pRR1038) to relocate to the nucleus following rapamycin addition prompted us to determine whether or not rapamycin-elicited Gln3 dephosphorylation was also lost in these mutants. If gross Gln3 dephosphorylation is a determinant of its localization, then the two processes should respond in parallel to alterations in Gln3.

As expected, full-length wild type Gln3_1–730-Myc¹³ (pRR536) exhibited strong dephosphorylation following the addition of rapamycin (Fig. 9). The normal slower migrating Gln3_1–730-Myc¹³ species characteristic of hyperphosphorylated Gln3 (upper black dots between lanes 1 and 2) was minimally detectable in the rapamycin-treated cells, whereas the lower, hypophosphorylated Gln3_1–730-Myc¹³ (pRR536) species (lower black dots between lanes 1 and 2) was prominent. In contrast, rapamycin elicited a smaller difference in Gln3_1–600-Myc¹³ (pRR614) phosphorylation levels. In the untreated culture, the hyperphosphorylated species (pRR614; upper black dots between lanes 1 and 2) predominated with little evidence of a clear band corresponding to the hypophosphorylated species (lower black dots). In Gln3_1–600-Myc¹³ rapamycin-treated cells, both hyper- and hypophosphorylated forms appear. Although the hyperphosphorylated species still predominated, the hypophosphorylated species was now clearly present and in greater abundance than observed in the untreated cells. Thus, rapamycin-elicited Gln3 dephosphorylation had been damaged but not destroyed in the truncation mutant.

FIGURE 9.

Gln3-Myc¹³ phosphorylation profiles observed with wild type (pRR536), Gln3_1–600-Myc¹³ truncation (pRR614), Gln3_{S656D,S659D,S662D}-Myc¹³ (pRR850), and Gln3_{S656A,S659A,S662A}-Myc¹³ (pRR1038) amino acid substitution proteins. The conditions in which transformants containing these plasmids were grown appear above the lanes. The plasmids used in the experiments are shown at the right of each panel. Growth conditions were as in Fig. 3. Rap, rapamycin.

When a more focused alteration was made in Gln3, the resulting protein, Gln3_{S656D,S659D,S662D}-Myc¹³ (pRR850), exhibited a phosphorylation profile with even greater similarity to that of the wild type Gln3_1–740-Myc¹³ (pRR536) than did Gln3_1–600-Myc¹³ (pRR614). The more slowly migrating species dominated in untreated cells, whereas the more rapidly migrating species dominated following rapamycin addition. The same predominance of the faster migrating species was also observed with rapamycin-treated Gln3_{S656A,S659A,S662A}-Myc¹³ (pRR1038). The more slowly migrating species also predominated in untreated cells. Together, these data indicated that Gln3 sequences 600–730 were substantially, although not absolutely, required for high level, rapamycin-elicited Gln3 dephosphorylation and that within this region, Gln3 residues 656, 659, and 662 did not play a particularly important role. Therefore, Gln3_{S656D,S659D,S662D}-Myc¹³ and Gln3_{S656A,S659A,S662A}-Myc¹³ dephosphorylation and localization did not parallel one another in rapamycin-treated cells. These data also further demonstrated that rapamycin-elicited gross dephosphorylation of Gln3 is not sufficient to cause its relocation into the nucleus.

Decreased Cytoplasmic Gln3-Myc¹³ Localization Fails to Correlate with Tor1 Association

The phenotypes of Gln3 amino acid substitutions in Gln3_{S656D,S659D,S662D}-Myc¹³ (pRR850), Gln3_{S656A,S659A,S662A}-Myc¹³ (pRR1038), the Gln3_1–600-Myc¹³ truncation, and the Gln3_510–731-Myc⁹, Gln3_600–731-Myc⁹, and Gln3_660–667-Myc⁹ truncations of Carvalho and Zheng (12) suggested that only Gln3 residues between 653 and 667 were required for high level cytoplasmic sequestration in nitrogen excess. If this region coincided with the Tor1-binding domain reported earlier (12), we expected that the serine substitutions in Gln3_{S656D,S659D,S662D}-Myc¹³ (pRR850) would eliminate the Tor1 association.

We tested this prediction with the two-hybrid assay system used and generously provided by Carvalho and Zheng (12). First, we confirmed that the assay system behaved the same in our hands as in the Zheng laboratory; growth in the presence of 3AT signified positive in vivo protein-protein interaction. The positive (Fig. 10A, images B and E, minus and plus 3AT, respectively), negative (Fig. 10A, images A and D), and empty vector (Fig. 10A, images C and F) controls all behaved as expected.

FIGURE 10.

Two-hybrid assessment of Tor1 association with full-length Gln3_1–730 (pRR1065), Gln3_1–600 (pRR1067), and Gln3_{S656D,S659D,S662D} (pRR1069). The assays were performed as described under “Materials and Methods.” The transformation recipient in all cases was strain PJ69-4a. The positive control was constructed by sequentially transforming PJ69-4a with pAV3-1 and pTD1-1. The negative control was a transformant containing pASTOR1–2470 (Gal4 DNA-binding domain-Tor1 fusion) and pACT2 (corresponding empty vector). pRR1065, pRR1067, and pRR1069 consisted of the Gal4 activation domain fused N-terminal of the respective Gln3 genes. Viability and the test for a two-hybrid association of the transformants was determined by streaking them on the same medium in the presence and absence of 3 mm 3AT. The procedures, strain, fusion vectors, and control plasmids were those used by Carvalho and Zheng (12). Panel A contains results yielded from control plasmids, whereas panel B depicts results from wild type and mutant test plasmids.

Confident in the assay, we transformed each of three Gln3 “prey” plasmid constructs: (i) wild type Gln3_1–730 (pRR1065), (ii) Gln3_1–600 (pRR1067), and (iii) Gln3_{S656D,S659D,S662D} (pRR1069), into a strain containing either pASTOR1–2470 (full-length Tor1_1–2470, images M–O) or pASTOR1–1764 (truncated Tor1_1–1764, images G–L) as the bait plasmid (Fig. 10B) (12). The three prey plasmids contained wild type and mutant gln3 genes identical to those in pRR536, pRR614, and pRR850, respectively, except for elimination of the Myc¹³ tag. We then assayed 36 independent transformants for each pair of plasmids in the absence (Fig. 10B, images G–I) or presence (images J–O) of 3AT. Only pASTOR1–1764 transformants grown in the absence of 3AT, the positive control for cells containing both plasmids, are shown in the figure (Fig. 10B, images G–I); analogous pTOR1–2470 transformants yielded similar data (data not shown).

To our great surprise, all three bait plasmids supported growth in the presence of 3AT, indicating positive associations with full-length Tor1_1–2470 (Fig. 10B, images M–O). Fig. 10A (image F) depicts what a lack of association with Tor1_1–2470 would look like. This indicated that wild type Gln3_1–730, Gln3_1–600, and Gln3_{S656D,S659D,S662D} all formed positive associations with full-length Tor1_1–2470. Based on the earlier results of Carvalho and Zheng (12), we minimally expected Gln3_1–600 (pRR1067) to be incapable of associating with Tor1 and hence had included it as a negative control for our two-hybrid assays, but that was not the case.

The exceptional nature of the above results prompted us to repeat the two-hybrid assays using a Tor1_1–1764 truncation containing the HEAT repeats reported to interact with Gln3 as the two-hybrid “bait” plasmid (pASTOR1–1764) (8). Results identical to those observed with bait plasmid pASTOR1–2470 were obtained, i.e., all three types of transformants grew in the presence (Fig. 10B, images J–L) and absence (images G–I) of 3AT.

A Short, Potential α-Helix Interacts with Tor1

The striking incongruities between our results and those reported by Carvalho and Zheng (12) required explanation. We reasoned the differences might have conceivably derived from: (i) technical defects in our assays or (ii) a fundamental, but unrecognized difference between the previous and present experimental strategies. Indeed, the latter explanation turned out to be the source of the disparities. Based on positive interactions between Gln3_510–730, Gln3_600–730, Gln3_600–667, and Tor1 and a lack of an interaction with the Gln3_667–730 peptide, Carvalho and Zheng (12) rightly concluded that Tor1-interacted with the Gln3_600–667 peptide. The key feature of their strategy was the use of isolated Gln3 peptides lacking Gln3 sequences N-terminal of residues 510, 600, and 667, respectively. We, on the other hand, employed C-terminal truncations, Gln3_1–600, and the Gln3_{S656,S659,S662} substitutions in full-length Gln3 molecules. In other words, all of our constructions contained all residues N-terminal to Gln3₆₀₀. Therefore, if Gln3 contained more than one Tor1-interacting site N-terminal of Gln3₆₀₀, it would have been missed in the experiments of Carvalho and Zheng (12), which accounts for the lack of congruency between our data and theirs.

Our functional assays had already demonstrated that minimally, Gln3 residues 656–662 were required for full cytoplasmic sequestration of Gln3-Myc¹³ and that those in Gln3_600–653 were not (Figs. 5–8). Therefore, we focused our attention on a 17-amino acid Gln3_654–670 peptide that contained Gln3_656–666, a sequence predicted by the Psipred secondary structure prediction method to potentially fold into an α-helix (Figs. 4A and 11B) (40, 41). Further, the substituted serine residues in Gln3_{S656D,S659D,S662D}-Myc¹³ (pRR850) were all situated on the same face of this putative Gln3_654–670 α-helix peptide, if it existed in this form (Fig. 11, substituted serine residues appear as white letters). We reasoned, based on these results, that the Gln3_654–670 peptide potentially contained the Tor1-interacting site situated in the C-terminal region of Gln3. This in turn predicted that the wild type Gln3_654–670 peptide (pRR1098) would interact with Tor1, whereas this peptide containing aspartate or alanine substitutions (Gln3_{S656D,S659D,S662D} (pRR1101) and Gln3_{S656A,S659A,S662A} (pRR1160)) would not. These predictions were validated experimentally. The wild type Gln3_654–670 peptide interacted with Tor1_1–2470 (Fig. 12B, images M and Q), whereas the mutant peptides did not (Fig. 12B, images N and R). Note that wild type Gln3_1–730 and Gln3_{S656D,S659D,S662D} (pRR1065 and pRR1069, respectively) again both yielded positive interactions with Tor1_1–2470 (Fig. 12B, images G and H in the absence and images K, L, O, and P in the presence of 3AT). Images A–F in Fig. 12A depict the two-hybrid assay controls, all of which yielded the expected results. Finally, we noted that when the entire set of independent replicate transformants, containing either wild type or mutant forms of the Gln3_654–670 peptide, were viewed as a whole, it appeared that the alanine substitutions exhibited a slightly greater interaction with Tor1_1–2470 (growth on +3AT) than the aspartate substitutions. Together, these data indicated that Gln3 possessed at least two sites that were able to associate, directly or indirectly, with Tor1 and that one of them was situated between Gln3 residues 654 and 670.

FIGURE 11.

Wild type Gln3 residues 656–667 depicted as situated on an α-helical wheel. The serine residues substituted in Gln3_{S656D,S659D,S662D}-Myc¹³ (pRR850) and Gln3_{S656A,S659A,S662A}-Myc¹³ (pRR1038) are indicated as white letters.

FIGURE 12.

Two-hybrid assessment of Tor1 association with full-length Gln3_1–730 (pRR1065), Gln3_{1–730 S656D,S659D,S662D} (pRR1069), Gln3_654–670 wild type peptide (pRR1098), Gln3_{654–670 S656D,S659D,S662D} peptide (pRR1101), and Gln3_{654–670 S656A,S659A,S662A} peptide (pRR1160). The format and strategy of this experiment were the same as that in Fig. 10. Panel A contains results yielded from control plasmids, whereas panel B depicts results from wild type and mutant test plasmids.

DISCUSSION

Parsing Tor1- and NCR-associated Regulation of Gln3 Localization

Experiments interrogating the C-terminal region of Gln3 have genetically separated TorC1- and NCR-associated regulation of Gln3 localization. Specifically, elimination of/or amino acid substitutions in a 11-amino acid peptide with a theoretical potential to fold into an α-helix, Gln3_656–666, abolished the ability of: (i) Gln3 to associate with Tor1 in a two-hybrid assay, (ii) nitrogen-rich conditions to completely sequester Gln3 in the cytoplasm, and (iii) rapamycin to elicit increased nuclear Gln3 localization in glutamine-grown cells. The Tor1-Gln3 association requirement for complete cytoplasmic sequestration is consistent with a requirement of the Tor1 kinase activity to achieve this end. This interpretation is consistent with the in vitro demonstration by Carvalho and Zheng (12) of Tor1 being able to phosphorylate Gln3. On the other hand, the ability of Gln3 localization to exhibit a normal response to the nitrogen source provided, a hallmark of NCR-sensitive regulation, remained intact and responsive to the negative regulation of Ure2 in the absence of this Tor1 association and inability to respond to rapamycin addition. These observations argue that Tor1 association-dependent, rapamycin-elicited Gln3 regulation is a distinct and genetically separable pathway from nitrogen source-responsive, NCR-sensitive Gln3 regulation.

Additivity of TorC1-dependent and NCR-sensitive Regulation

Beyond genetically dissecting the contributions of Tor1- and NCR-associated Gln3 regulation, the data also demonstrate the required collaboration of the two regulatory mechanisms to achieve the overall nitrogen-responsive control of Gln3 localization. NCR-mediated regulation alone in cells cultured in nitrogen-replete glutamine medium was insufficient to completely sequester Gln3 in the cytoplasm; a Tor1 association-dependent contribution was also needed. On the other hand, loss of the Tor1 contribution resulted in a nuclear shift in Gln3 localization, but again one that was far from complete. Even under the best conditions, high concentrations of rapamycin were incapable of achieving more than ∼30% nuclear Gln3-Myc¹³ localization with similar amounts remaining completely cytoplasmic and nuclear-cytoplasmic, respectively (Fig. 1). Complete nuclear Gln3 localization could only be achieved when cells, devoid of the Gln3_656–666-Tor1 association, were cultured under derepressive conditions, i.e., in nitrogen-limiting proline medium, or in an ure2Δ. This is the first time that the combined action of Tor1- and NCR-associated participation has been demonstrated to be additively required for overall wild type regulation.

It may be mere coincidence that the Gln3-Myc¹³ intracellular distribution profiles observed in mutants where the Tor1-Gln3_656–666 association was eliminated are remarkably similar to those observed with rapamycin-treated, wild type, glutamine-grown cells (compare Fig. 1 wild type Gln3_1–730-Myc¹³ (pRR536) +Rap with Fig. 2 truncations ± Rap or Fig. 8, Gln3_{S656D,S659D,S662D}-Myc¹³ (pRR850) and Gln3_{S656A,S659A,S662A}-Myc¹³ (pRR1038) ± Rap). However, this is a lot of positive correlation to be mere coincidence. Rapamycin treatment could mimic the loss of the Tor1-Gln3 interaction but could go no further. In other words, rapamycin effects were restricted to reversing the outcome of Tor1 association-dependent actions. To maintain proper perspective, however, it is important to point out that the data neither indicate that residues putatively phosphorylated by Tor1 are the same as those putatively dephosphorylated by rapamycin-elicited phosphatase action nor that rapamycin treatment is preventing Tor1-dependent phosphorylation. Putative Tor1 association-dependent phosphorylation and cytoplasmic sequestration may well be a completely independent event from rapamycin-elicited phosphatase action and the partial shift of Gln3 into the nucleus.

Although previous data strongly suggested that two separate regulatory systems were responsible for regulating Gln3 localization (20–23, 25–31), they suffered from the unavoidable caveats that observed results could potentially have derived from indirect effects of the conditions employed. This caveat derived from the fact that inhibitors functioned at remote sites from the reporter protein whose localization was being influenced, Gln3-Myc¹³, and mutant studies involved molecules other than Gln3 itself. In the present work, these caveats have been greatly minimized or eliminated because separation of TorC1- and NCR-associated components of Gln3 regulation derived from direct alteration of its target protein, the Gln3-Myc¹³ molecule itself.

With What Does Gln3 Interact?

Data in this work and the earlier study of Carvalho and Zheng (12) clearly demonstrate a positive Tor1-Gln3 interaction using the standard two-hybrid interaction assay. However, this assay has the caveat of not specifying whether the physiologically significant, in vivo interaction(s) visualized are direct or indirect. One need look no further than the data of Carvalho and Zheng (12) for such a precedent. Their two-hybrid results visualized Tor1 interactions with Gln3, Ure2, Dal81, Dal82, Gat1, and Gzf3 (8). Gln3 is known to interact with Ure2 and Dal82 (6–8, 42–44). However, Dal81 interacts with Gln3 only in a Dal82-dependent manner (44). Therefore, it may be premature to conclude that Gln3 binds directly to Tor1. Although in vitro Tor1 clearly mediates Gln3 phosphorylation (8), it is worth noting that Gln3 contains 146 serine/threonine residues providing a wide in vitro selection of possible phosphorylation candidates that may or may not be physiologically significant. However, the present study provides an excellent highly defined probe to search for other molecules with which Gln3_656–666 interacts. This information, which we are attempting to acquire, may contribute to a more accurate and detailed view of the mechanism through which TorC1 regulates Gln3.

Gln3 Contains More than One Tor1-interacting Site and Possible Implications

The data in this work also demonstrate that, based on two-hybrid assays, Gln3 contains one or more Tor1-interacting sites beyond the one situated at Gln3_656–666. The most compelling evidence is that Gln3_1–600, lacking the demonstrated site at Gln3_656–666, exhibits a positive interaction with both Tor1_1–2470 and Tor1_1–1764. Further, reasoning described under “Results” provides a consistent explanation of how such a putative additional Tor1-interacting site(s) may have escaped earlier detection. If additional Tor1-interacting site(s) do, in fact, exist in Gln3, they open an important question: What is their function with respect to Gln3 regulation? Our future investigations will be directed toward locating additional putative Tor1-association site(s) in Gln3 and addressing this question.

Finally, two additional items merit emphasis: (i) the observations from which our conclusions derive would not have been possible without the ability to reliably quantify Gln3-Myc¹³ localization, and (ii) it is quite remarkable that systematic introduction of extensive alterations throughout the Gln3_600–655 region (25% of the residues) had no significant effect on rapamycin-influenced or NCR-sensitive regulation of Gln3 localization; all but one of the multiple substitutions failed to generate a mutant phenotype. This suggests that beyond direct or indirect interaction with Tor1, the bulk of this region either plays only a limited or minimally direct role in the regulated localization of Gln3 or that the role it plays tolerates significant alteration.