Methionine sulfoximine treatment and carbon starvation elicit Snf1-independent phosphorylation of the transcription activator Gln3 in Saccharomyces cerevisiae.

Tor proteins are global regulators situated at the top of a signal transduction pathway conserved from yeast to humans. Specific inhibition of the two Saccharomyces cerevisiae Tor proteins by rapamycin alters many cellular processes and the expression of hundreds of genes. Among the regulated genes are those whose expression is activated by the GATA family transcription activator, Gln3. The extent of Gln3 phosphorylation has been thought to determine its intracellular localization, with phosphorylated and dephosphorylated forms accumulating in the cytoplasm and nucleus, respectively. Data presented here demonstrate that rapamycin and the glutamine synthetase inhibitor, methionine sulfoximine (MSX), although eliciting the same outcomes with respect to Gln3-Myc13 nuclear accumulation and nitrogen catabolite repression-sensitive transcription, generate diametrically opposite effects on Gln3-Myc13 phosphorylation. MSX increases Gln3-Myc13 phosphorylation and rapamycin decreases it. Gln3-Myc13 phosphorylation levels are regulated by at least three mechanisms as follows: (i) depends on Snf1 kinase as observed during carbon starvation, (ii) is Snf1-independent as observed during both carbon starvation and MSX treatment, and (iii) is rapamycin-induced dephosphorylation. MSX and rapamycin act additively on Gln3-Myc13 phosphorylation, but MSX clearly predominates. These results suggest that MSX- and rapamycin-inhibited proteins are more likely to function in separate regulatory pathways than they are to function tandemly in a single pathway as thought previously. Furthermore, as we and others have detected thus far, Gln3 phosphorylation/dephosphorylation is not a demonstrably required step in achieving Gln3 nuclear localization and nitrogen catabolite repression-sensitive transcription in response to MSX or rapamycin treatment.

In addition to the two type 2A phosphatases, three protein kinases, Tor1/2 and Snf1, are also associated with Gln3 regulation. Both the Tor proteins and Snf1 phosphorylate Gln3 in vitro and interact with Gln3 in two-hybrid interaction and co-immunoprecipitation experiments (16,33). However, the actions of these kinases appear to regulate Gln3 localization differently. Tor1/2-dependent, i.e. rapamycin-sensitive, Gln3 phosphorylation correlates with its cytoplasmic localization, whereas Snf1-dependent Gln3 phosphorylation correlates with its nuclear accumulation during glucose starvation. Phosphorylation events regulated by the two kinases were concluded to be distinct because glucose starvation-induced Gln3 phosphorylation could be elicited in the presence of rapamycin (16,33).
Although much of the available data support the existing view of Tor regulation, several observations suggest that we do not yet fully understand how Tor1/2 regulate Gln3 intracellular localization. (i) The results of genetic studies with tap42 mutants are more consistent with the suggestion of Tap42 functioning as a positive rather than a negative regulator of Sit4 activity (7, 8 13, 14). (ii) Gln3 phosphorylation levels do not correlate with intracellular Gln3 localization in cells provided with a range of nitrogen sources and during nitrogen and carbon starvation in the same way as they do with rapamycin treatment (34). (iii) Nuclear accumulation of phosphorylated Gln3 during carbon starvation derives not from starvation for glucose but for nitrogen. Nitrogen starvation occurs because glucose-starved cells cannot produce the ␣-ketoglutarate needed to assimilate the nitrogen source (ammonia) used in the experiment. Therefore, both phosphorylated Gln3, generated by ␣-ketoglutarate-mediated nitrogen starvation, and dephosphorylated Gln3, in response to rapamycin addition, accumulate in the nucleus (35). This occurs despite the fact that nitrogen starvation and rapamycin are both thought to regulate Gln3 through their inhibition of Tor1/2 (9,16,29,30).
This study was initiated in an effort to rectify instances in which the available data cited above are not congruent with expectations derived from the current model describing how the S. cerevisiae Tor proteins regulate Gln3. The report of MSX triggering dephosphorylation and nuclear localization of Gln3 provided a new way of perturbing Gln3 regulation (32) and prompted us to employ MSX in our investigations. Here we confirm the earlier report that MSX treatment elicits nuclear localization of Gln3-Myc 13 and high level NCR-sensitive gene expression. But to our surprise, MSX markedly increased rather than decreased Gln3-Myc 13 phosphorylation. Our results demonstrate Gln3-Myc 13 phosphorylation occurs by at least two mechanisms: one is Snf1-dependent and the other Snf1-independent. MSX-induced Gln3 phosphorylation is Snf1independent, whereas both mechanisms function to increase Gln3 phosphorylation during carbon starvation. Furthermore, the effects of MSX and rapamycin treatment on Gln3 phosphorylation are additive, but those of MSX clearly predominate. Although MSX and rapamycin treatment generate the same final outcome, nuclear localization of Gln3, the phosphorylation data are more consistent with MSX and rapamycin acting in parallel pathways to achieve that outcome than they are with their inhibiting tandem steps in a single regulatory pathway as proposed previously. These observations also suggest that something more or other than the Gln3 phosphorylation/dephosphorylation levels, detected thus far, determine the intracellular localization of Gln3.
Strain Construction-Strain RR181 was constructed as follows. A fragment containing the SNF1 gene and its flanking sequences was amplified by high fidelity PCR (Stratagene PFU Turbo DNA polymerase) using DNA from TCY1 as template. An EcoRI-SpeI fragment from the PCR product was cloned into pBS(KSϩ), yielding pRR576. The 2.2-kb, SNF1-containing, PstI-HpaI fragment from pRR576 was replaced with the 0.8-kb, TRP1-containing, PstI-HpaI fragment from pRR496, yielding pRR578. The 1.7-kb EcoRI-SpeI fragment from pRR578 was used to transform strain TB123. Trp ϩ recombinants were selected by growth in minimal medium. Genomic Southern blot and mRNA analyses were used to confirm the presence of the snf1⌬ in RR181.
Western Analysis-Cells were harvested in one of two ways. In the first method, the flask containing the culture was swirled in an icewater bath for 1 min (9). The culture was then transferred to a sterile 40-ml Nalgene tube, and the cells were pelleted at 3,000 rpm for 5 min (4°C). The supernatant was removed; the cell pellet was resuspended in 300 l of ice-cold protein extraction buffer (120 mM NaCl, 50 mM Tris, pH 7.5, 2 mM EDTA, 2 mM phenylmethylsulfonyl fluoride, 10 mM NaF, 10 mM p-nitrophenyl phosphate, 10 mM sodium pyrophosphate, 10 mM ␤-glycerophosphate, and a commercial mixture of protease inhibitors (Minicomplete, Roche Applied Science) (9,34)) and transferred to a microcentrifuge tube, and the cells were collected by centrifugation (30 s at 3000 rpm). The cells were resuspended in 200 l of fresh, ice-cold extraction buffer and further processed as described earlier (34).
Alternatively, cells were harvested by filtration onto a type HA, 0.45-m Millipore filter. Cells were quickly scraped from the filter, placed in a sterile 1.5-ml microcentrifuge tube, and flash-frozen by submerging the microcentrifuge tube and cells in liquid nitrogen for 20 -30 s. The total time for cell harvest was 45-50 s. The tube, still containing liquid nitrogen, was then quickly transferred to Ϫ80°C until further processing of the cells was required. As with the centrifugation method, the frozen cells were resuspended in 200 l of cold protein extraction buffer; cold glass beads were added, and the cells were processed as described by Cox et al. (34). Except where otherwise indicated, cells were harvested throughout this work using the filtration flash-freezing method. Conditions of electrophoresis, immunochemical staining, and visualization were also as described earlier (34).
Although the method of cell harvest was not indicated in previous reports (9,16), the data presented are the same as we obtained when we harvested cells using the centrifugation method (34). It may also be pertinent that centrifugation, along with chilling the cells on ice for ϳ10 min, was used when cells in one report were transferred to new nutrient conditions (9).
In some experiments we observed spurious lane-to-lane variation in signal strength that we were unable to eliminate. Therefore, we felt it more prudent to assess the relative amounts of various species within a lane and to compare the pattern of data observed in that lane to the pattern observed in another lane than to compare quantitatively the amounts of specific Gln3 species observed in one lane to those in another. Also in many cases, control extracts derived from glutaminegrown cells were placed in the outside lanes of the gel as done earlier (34).
Phosphatase Treatment-Crude protein extracts were prepared as described above with the following extraction buffer: 120 mM Tris-HCl, pH 7.5, 50 mM NaCl, 2 mM EDTA, and 2 mM phenylmethylsulfonyl fluoride and a commercial mixture of protease inhibitors (Roche Applied Science). Extracts were incubated with 20 units of calf intestinal alkaline phosphatase (CIP) (Roche Applied Science) or 20 units of CIP ϩ phosphatase inhibitor, 40 mM sodium pyrophosphate, for 15 (ammonia) or 30 min (urea) at 30°C. Just prior to CIP addition, we added 0.9 mM MgCl 2 and 0.1 mM ZnCl 2 (final concentrations) to alleviate inhibition of CIP by EDTA that was present in the extraction buffer. Following incubation, samples were boiled 5 min and then loaded onto the gel. We noted that the presence of pyrophosphate in a sample adversely affected the strength of the immunofluorescent signal.
Indirect Immunofluorescence-Immunofluorescence staining was carried out as described by Cox et al. (34,35). 13 Mobility Suggesting Increased Phosphorylation-Rapamycin inhibition of Tor1/2 has been useful both in the study of Tor function and Gln3 regulation. Analogously, MSX appears to be an even more useful perturbant for investigating control of Gln3. This inhibitor is more specific than rapamycin in that it altered nuclear localization of Gln3 and Rtg1 but not Gat1, Msn2, Msn4, and an unknown factor involved in the regulation of ribosomal protein gene expression (32).

MSX Induces Decreased Gln3-Myc
Our studies were initiated with control experiments testing whether MSX elicited the same response in our hands as reported previously (32). In support of that earlier work, we observed that treating cells with MSX markedly increased both Gln3-and Rtg3-mediated expression of DAL5 and CIT2, respectively ( Fig. 1). Moreover, increased expression occurred with each nitrogen source tested except glutamine. The data, however, exhibited a somewhat unexpected characteristic. DAL5 expression increased more in glutamate-grown cells treated with MSX than when induced by rapamycin (Fig. 1A, lanes C-E). This was exceptional because rapamycin inhibits proteins (Tor1,2) considered to be downstream of the nutrientsensing signal altered by MSX inhibition of glutamine synthetase in the current model of Tor regulation (32). Given these positive results, we assessed the effect of MSX on Gln3 phosphorylation, and we were surprised that MSX treatment decreased Gln3-Myc 13 mobility rather than increasing it ( Fig. 2A,  lanes B and C). This change in mobility suggested Gln3-Myc 13 phosphorylation was increased rather than decreased by MSX treatment, a finding unaccountably opposite to the observations of Crespo et al. (32).

Effects of Culture Medium and Method of Cell Harvest on
Gln3-Myc 13 Mobility-The striking difference in MSX-induced Gln3-Myc 13 mobility we observed compared with that reported earlier prompted us to look more carefully at the experimental conditions used in present and past experiments. The first difference was the concentration of ammonia provided in the medium; we used 0.1% compared with 0.5% used in the earlier analysis (32). This, however, did not account for the differing results. MSX treatment decreased Gln3-Myc 13 mobility irrespective of the ammonia concentration ( Fig. 2A, lanes B-E). Minor differences did, however, occur in the migration rates of minor Gln3-Myc 13 species derived from untreated cells grown at the two ammonia concentrations. Gln3-Myc 13 migrated slightly faster with glutamine than with low (0.1%) ammonia as nitrogen source; this difference was greater at the higher (0.5%) concentration of ammonia (compare Fig. 2A, lanes A and B with lanes E and F).
Present and past experiments technically differed in two further respects. YNB medium was used in current experiments, whereas those reported earlier used SD medium. Although the two media are similar in composition, they are not identical, and differences in results obtained with them have been noted in the past. 2 In the present instance, however, media differences were not important; MSX treatment elicited decreased Gln3-Myc 13 mobility in both media (Fig. 2B, lanes B and C and D and E). The second difference was the method of cell harvest. In previous experiments, centrifugation was used to harvest the cells (34), whereas filtration and flash-freezing was used here. We changed our protocol because Wilson et al. (41) reported Snf1 kinase is activated by centrifugation, and Snf1 is required for Gln3 phosphorylation during glucose starvation (33). As shown in Fig Fig. 2C, lanes A and D, where centrifugation was used, more of the protein is situated in the uppermost of these two bands (lower black dot). In cells treated with rapamycin, the method of harvest seemed to make less difference in Gln3-Myc 13 mobility ( Fig. 2C, lanes C and F).
Decreased MSX-induced Gln3-Myc 13 Mobility Derives from Increased Phosphorylation-Since differences we could identify between current and past experimental techniques did not account for differences in our results with MSX compared with those reported earlier, we determined whether the electrophoretic behavior of Gln3-Myc 13 from MSX-treated cells (i) was an artifact created during extract preparation or (ii) was derived from verifiable hyperphosphorylation. We tested the first possibility by using an in vitro mixing experiment. Four wild type cultures were prepared. The first two were treated with rapamycin for 30 min, while the third and fourth were similarly incubated with MSX. The first and third cultures were harvested by filtration and processed normally (Fig. 3A, lanes B and C, respectively). The second culture (rapamycin-treated) was harvested, flash-frozen, and held in liquid nitrogen until the fourth cell sample (MSX-treated) was harvested, added to it, and frozen as well (Fig. 3A, lane A and repeated in D). In other words, the third experimental sample contained two sets of cells that were harvested and frozen separately but were contained in the same tube. This protocol prevented the cell samples from being exposed to a second drug before or during harvest. The three samples were subsequently thawed, the cells immediately broken, and the extracts subjected to electrophoresis. Gln3-Myc 13 from the extract of rapamycin-treated cells exhibited significantly greater mobility than the one from MSX-treated cells (Fig. 3A, lanes B and C). The third sample, containing a mixture of cells separately treated with the two inhibitors, yielded a profile that was very nearly the sum of the profiles observed in lanes B and C (Fig. 3A, lanes A and D). This indicated it was unlikely that significant artifacts occurring after the cells were broken had generated the profile seen in extracts from MSX-treated cells. If there was a difference of the profiles in Fig. 3A, lanes B and C versus A and D, it was a small amount of artifactual, in vitro dephosphorylation of the slowly migrating Gln3-Myc 13 species generated by MSX treatment in lanes A and D.
To determine whether decreased Gln3-Myc 13 mobility observed with extracts from MSX-treated cells derived from increased phosphorylation, we tested its sensitivity to CIP. Incubating extracts from urea-grown, MSX-treated cells with CIP prior to electrophoresis increased the mobility of Gln3-Myc 13 such that it nearly co-migrated with that from rapamycintreated cells (Fig. 3B, lanes D-F). The CIP-dependent increase in Gln3-Myc 13 mobility was prevented by including 40 mM sodium pyrophosphate, a CIP inhibitor, in the reaction mixture (Fig. 3B, lanes C-E). Extracts derived from similarly treated ammonia-grown cells yielded the same results (Fig. 3C). It is not clear why Gln3-Myc 13 was dephosphorylated by CIP to a level one band above that observed with rapamycin treatment rather than to the same level, but we did encounter some inhibition of CIP by our extraction buffer.
Three conclusions were derived from these experiments. (i) Gln3 phosphorylation levels are sensitive to even minor differences in culture conditions and method of cell harvest. (ii) Centrifugation, as reported for other substrates of Snf1 (41), increases Gln3 phosphorylation. (iii) Treating cells with MSX markedly increases rather than decreases Gln3 phosphorylation irrespective of technical differences evaluated above.
Increased Nuclear Localization of Gln3-Myc 13 Induced by MSX Occurs with All Nitrogen Sources Tested but Glutamine-Increased Gln3-Myc 13 phosphorylation in MSX-treated cells prompted two questions. (i) Did MSX and rapamycin treatment differentially affect Gln3 nuclear localization? (ii) Were the effects of MSX treatment influenced by the nitrogen source? The pertinence of these questions derived from the fact that MSX and rapamycin were reported to affect Gln3 phosphorylation, nuclear localization, and Gln3-mediated gene expression in the same way, i.e. through their influence on Tor1/2 activity (32).
As reported previously, treating ammonia-grown cells with MSX resulted in strong nuclear accumulation of Gln3-Myc 13 similar to that seen in rapamycin-treated cells (Fig. 4A, microscopic images). This accumulation correlated well with high level Gln3-mediated DAL5 expression under the same conditions (Fig. 4A, images, and Fig. 1A, lanes H and I). On the other hand, little Gln3-Myc 13 nuclear localization or DAL5 expression occurred in MSX-treated, glutamine-grown cells (Fig. 4F, images, and Fig. 1A, lanes J and K). This contrasts with the response of glutamine-grown cells to rapamycin (Fig. 4F, images, and Fig. 1A, lanes J and L). Glutamine, however, was unique in its ability to prevent nuclear accumulation of Gln3-Myc 13 . With all of the other nitrogen sources we assayed (urea, proline, and glutamate), Gln3-Myc 13 nuclear accumulation and DAL5 expression was elicited by MSX treatment, albeit with some quantitative differences (Fig. 4, B, D and E, and Fig. 1A).
The close correlation described above was not, however, observed in Gln3-Myc 13 phosphorylation. Gln3-Myc 13 phosphorylation, in preparations derived from ammonia-or urea-grown FIG. 3. A, effect of in vitro mixing of frozen cells treated with MSX or rapamycin on the electrophoretic mobility of Gln3-Myc 13 . Wild type cells were cultured in YNB/urea (0.1%) in the presence (ϩ) or absence (Ϫ) of MSX or rapamycin. In lanes B and C, cells were harvested by filtration and processed normally for Western blot analysis. The combined cell samples in A, lanes A and D (one from rapamycin-treated cells and the other from MSX-treated cells), were harvested and frozen separately, but with freezing taking place in the same microcentrifuge tube as described in the text. Thereafter, all of the cell samples were processed normally for Western blot analysis. By this procedure, the extracts in lanes A and D derived from cells and extracts that were mixed only during the process of cell breakage. B and C, effects of phosphatase treatment on the electrophoretic mobility of Gln3-Myc 13 derived from cells cultured under various conditions. Wild type cells were cultured in YNB/urea (0.1%) or YNB/ammonia (0.1%) ( ϩ NH 4 ) media to mid log phase. Cultures were then incubated in the presence (ϩ) or absence (Ϫ) of MSX or rapamycin for 30 min and harvested by filtration and flash-freezing. Cells were broken in the phosphatase buffer described under "Materials and Methods," and the indicated samples (ϩ) were incubated with either CIP or CIP plus 40 mM sodium pyrophosphate (Inh). Following this incubation, the extracts were processed for Western blot analysis as described under "Materials and Methods." cells treated with MSX versus rapamycin, behaved oppositely. Gln3-Myc 13 phosphorylation increased upon MSX treatment and, overall, decreased in the presence of rapamycin (Fig. 4, A  and B, lanes B and C and F and G). To ensure that we were not misinterpreting the behavior of Gln3-Myc 13 phosphorylation, we compared experimental extracts flanked on either side by controls (Fig. 4C). MSX treatment increased Gln3-Myc 13 phosphorylation relative to wild type (Fig. 4C, lanes A-C), whereas rapamycin treatment decreased it (lanes D-F). The difference was even more dramatic when Gln3-Myc 13 from MSX-treated cells was electrophoresed between samples from rapamycintreated cells (Fig. 4C, lanes G-I).
Gln3-Myc 13 phosphorylation in untreated versus MSXtreated cells provided with proline or glutamate as nitrogen source overall behaved similarly to ammonia-and urea-grown cells, with the exception that the increase in phosphorylation, relative to untreated cells, was not as great (compare Fig. 4, A and B with D and E). This derives from the fact that Gln3-Myc 13 in untreated cells already exhibits greater phosphorylation with both proline and glutamate than observed with ammonia or urea (compare with Gln3-Myc 13 derived from control Gln-grown cells, Fig. 4, A, B, D, and E, lanes A, D, E, and H) as reported earlier (34). MSX had no demonstrable effect on phosphorylation levels in glutamine-grown cells (Fig. 4F).
Beyond major changes of Gln3-Myc 13 phosphorylation in MSX-treated cells, additional, more subtle differences were also noted in Fig. 4. When using the filtration method of cell harvest, the most rapidly migrating Gln3-Myc 13 species seen in rapamycin-treated cells frequently migrated similarly to the most rapidly migrating species of Gln3-Myc 13 from untreated samples.
In other words, we did not always see the marked increase in electrophoretic mobility in Gln3-Myc 13 species that was often, but not always, observed when we used the centrifugation method of harvest (34) and in earlier reports of phosphorylation experiments (9,16). This is clearly observed in cells provided with ammonia, urea, and glutamine as nitrogen source (Fig. 4,  A and B, lanes F-H, and F, lanes C and D). The effect is not seen with glutamate or proline, because rapamycin treatment of cells provided with these nitrogen sources does not increase mobility of Gln3-Myc 13 ; this is in agreement with our earlier observations (34).
Because MSX treatment of urea-grown cells produced the largest change in Gln3-Myc 13 mobility, we chose this nitrogen source for subsequent experiments. However, experiments using ammonia-grown cells were also performed in most cases to ensure the results we observed did not derive from using urea as the nitrogen source.
Snf1-dependent and -independent Gln3-Myc 13 Phosphorylation during Carbon starvation-Increased Gln3-Myc 13 phosphorylation following MSX treatment was similar to that observed earlier during carbon starvation of ammonia-grown cells (33,34). Moreover, as occurs with MSX treatment above, Gln3-Myc 13 localizes to the nuclei of ammonia-grown, carbonstarved cells (33)(34)(35). These correlations raised the possibility that carbon starvation-activated Snf1 kinase (42-44) might play a role in MSX-mediated Gln3-Myc 13 phosphorylation. To evaluate this possibility, we determined whether deletion of SNF1 affected NCR-sensitive Gln3-mediated gene expression, and whether all of the increased Gln3-Myc 13 phosphorylation observed during carbon starvation was Snf1-dependent. Wild type and snf1⌬ strains are equally NCR-sensitive (Fig. 5A,  lanes A-D). However, strains in this genetic background are overall less NCR-sensitive than M970 in the ⌺1278b background; this is more clearly evident in the case of DAL80 expression (compare data in this work with analogous experi-ments in Refs. 39 and 40). Similarly, the CIT2 expression profile with various conditions is not identical to that in M970 (compare Fig. 1B to data in Refs. 39 and 40). SNF1 expression was not demonstrably affected by NCR, and the message was totally absent, as expected, in the snf1⌬ (Fig. 5A, lanes E-H). Furthermore, Gln3-Myc 13 accumulated in the nuclei of snf1⌬ cells treated with either rapamycin or MSX in nearly the same ways as wild type (Fig. 5B). There did appear, however, to be greater Gln3-Myc 13 nuclear localization in the glutaminegrown, MSX-treated snf1⌬.
We next measured Gln3-Myc 13 phosphorylation in carbonstarved wild type and snf1⌬ cells. As quickly as 30 min following onset of carbon starvation, Gln3-Myc 13 phosphorylation in the wild type strain increased (Fig. 6A, compare lane B to lanes C-F) as reported by Bertram et al. and subsequently ourselves (33,34). In addition, the heterogeneity of Gln3-Myc 13 species increased compared with the unstarved sample. When the above experiment was repeated in a snf1⌬, carbon starvation increased Gln3-Myc 13 phosphorylation but to a lesser degree than wild type (compare Fig. 6B, lanes A-F with equivalent lanes in A). In addition, phosphorylated Gln3-Myc 13 species were less heterogeneous in the snf1⌬; there were three major species (Fig. 6B, lane  D, black dots). Characteristic differences in Gln3-Myc 13 phosphorylation between carbon-starved wild type and snf1⌬ cells are more clearly seen in Fig. 6C, lanes C-E. These results suggest that carbon starvation elicits both Snf1-dependent and -independent Gln3-Myc 13 phosphorylation.
The expected phenotype of deleting a protein kinase gene is decreased phosphorylation of its target substrates, as observed with Gln3-Myc 13 phosphorylation in carbon-starved snf1⌬ versus wild type (Fig. 6C, lanes C-E). However, in addition, we observed a slight increase in Gln3-Myc 13 phosphorylation (by one band, black dot) in the snf1⌬ grown in high glucose (Fig.  6C, lanes A and B). The effect is most clearly seen by comparing

FIG. 5. The effects of MSX and rapamycin treatment on Gln3mediated transcription in wild type (TB123) and snf1⌬ (RR181) as well as intracellular Gln3-Myc 13 localization in a snf1⌬. Experimental details for Northern blots in A and microscopic images in B
were as described in Figs. 1 and 4, respectively. Images of 4,6-diamidino-2-phenylindole-positive material were collected but were not merged for publication with those of the Gln3-Myc 13 signal in B (data not shown). W.T., wild type.

FIG. 6. The effects of carbon starvation on the mobility of Gln3-Myc 13 in wild type (W.T.) (TB123) and snf1⌬ (RR181) cells.
Cells were cultured in YNB medium containing the indicated nitrogen source (0.1%). Extracts from glutamine-grown cells (GLN) were used as a control. In instances of starvation, cells were permitted to grow to an A 600 nm ϭ 0.5; the cultures were quickly filtered, and the cells transferred to pre-warmed, pre-aerated YNB-urea medium devoid of glucose. Following starvation for the times indicated (in min), cells were harvested by filtration and processed as described under "Materials and Methods" for Western blot analysis.
Gln3-Myc 13 phosphorylation in glutamine-versus urea-grown wild type cells (Fig. 6A, lanes A and B) with those in the snf1⌬ (Fig. 6C, lanes A and B and F and G). This suggests Snf1, in high glucose medium, indirectly influences Gln3-Myc 13 phosphorylation in a way opposite that seen in carbon-starved cells.
MSX-induced Gln3-Myc 13 Phosphorylation Is Snf1-independent-Given that both Snf1-dependent and -independent Gln3-Myc 13 phosphorylation occurred, we wanted to ascertain which mechanism was associated with MSX-elicited Gln3-Myc 13 phosphorylation. Therefore, an extract from MSXtreated cells was placed between two extracts from carbonstarved, wild type (Fig. 7A, lanes C-E) or snf1⌬ (Fig. 7B, lanes  C-E) cells. The Gln3-Myc 13 phosphorylation profile in MSXtreated wild type cells resembled that observed in a carbonstarved snf1⌬ more than it did in carbon starved wild type cells, suggesting phosphorylation was Snf1-independent. This was confirmed by demonstrating Gln3-Myc 13 phosphorylation increased in unstarved snf1⌬ cells treated with MSX compared with untreated snf1⌬ cells (Fig. 7D, lanes A-C; also observed in Fig. 7B, lanes B and D). Consistent with these results, the Snf1-dependent heterogeneity of hyperphosphorylated Gln3-Myc 13 species observed in carbon-starved wild type cells was lacking in MSX-treated cell extracts (Fig. 7, A, lanes C-E and  B, lanes C-E). The question remained, however, whether or not Gln3-Myc 13 phosphorylation levels observed in MSX-treated wild type and snf1⌬ cells were the same. Data in Fig. 7C (lanes  D and E), argue that Gln3-Myc 13 species from the two cell samples co-migrate. The slight apparent lag in mobility of the upper-most band in Fig. 7C, lane E (black dot), relative to that in lane D derives from "blooming" of the more intense signal in lane E. On the other hand, the small increase in Gln3-Myc 13 phosphorylation previously seen in snf1⌬ relative to wild type is also present in Fig. 7C, lanes B and C. Note that the bottom band in lane B of Fig. 7C is missing in lane C. Together these data suggest that (i) Gln3 is phosphorylated during carbon starvation by both Snf1-dependent and -independent mechanisms, and (ii) MSX elicits Snf1-independent Gln3 phosphorylation.
During these and other experiments, we also noticed that Gln3-Myc 13 phosphorylation did not respond in quite the same way to rapamycin treatment of wild type versus snf1⌬ cells (Fig.  7D, lanes D-F; data not shown). The extent of Gln3-Myc 13 dephosphorylation in response to rapamycin was greater (by one band) in the wild type than in a snf1⌬, i.e. there was less dephosphorylation in the snf1⌬. Recall the same observation was made above in untreated snf1⌬ extracts (Fig. 6C, lanes A and B, and F  and G versus A, lanes A and B). This again argued in favor of Snf1 kinase indirectly influencing Gln3-Myc 13 dephosphorylation.
Data in Fig. 4, A and B, indicated that increased Gln3-Myc 13 phosphorylation was similarly observed in both ammonia-and urea-grown cells. To ensure this similarity extended to the responses elicited by carbon starvation and snf1⌬, we carbonstarved ammonia-grown cells. MSX-induced Gln3-Myc 13 phosphorylation occurred in ammonia-grown cells just as it had with urea (Fig. 8A). In both cases, the bottom band in extracts from MSX-treated cells was lost, and the signal in the top band increased (carefully compare black dots in Fig. 8A, lanes A and  C, with Fig. 7A, lanes B and D). Similarly, the three phosphorylated species observed in MSX-treated, urea-grown snf1⌬ cells were also observed in snf1⌬ ammonia-grown samples (Fig.  8B, lanes B and E). Increased heterogeneity of Gln3-Myc 13 phosphorylation observed when ammonia-grown wild type cells were starved for carbon was similar to that seen with urea, as was disappearance of the heterogeneity in the snf1⌬ (Fig. 8B,  lanes C and D).
Although results with ammonia as nitrogen source were very similar to those observed with urea, one observation in the experiment with ammonia-grown cells was unsettling. In our earlier studies, Gln3-Myc 13 phosphorylation increased much more dramatically in ammonia-grown, carbon-starved cells (180 min) than we observed here (compare the right panel of Fig. 7B, lanes C and D, in Ref. 34 with Fig. 8A, lanes A and D of the present work). However, upon comparison, we found the two experiments differed in one important respect, i.e. filtration was used to harvest cells in the present case, whereas we previously used centrifugation (34). This technical difference clearly accounts for the markedly different observations. Gln3-Myc 13 was phosphorylated to a much greater extent in the centrifuge-harvested, carbon-starved cells than those collected by filtration (Fig. 8C, compare lanes A, C, D, and F).
MSX and Nitrogen Starvation-induced Gln3-Myc 13 Phosphorylation-MSX is suggested to elicit increased NCR-sensitive transcription by starving cells of the signal for nitrogen excess (32), predicting the Gln3-Myc 13 phosphorylation profiles of MSX-treated and nitrogen-starved cells should be the same. Testing this prediction, we observed that nitrogen starvation of ammonia-grown cells only modestly increased Gln3-Myc 13 phosphorylation levels (Fig. 9B, lanes A, F, C, and D). Similar results were observed when cells were pre-grown in YNB-urea medium (Fig. 9A, lanes A, F, C, and D). In contrast, Gln3-Myc 13 phosphorylation increased substantially in MSX-treated cells (Fig. 9, A and B, lanes B and E). There was a modest increase in Gln3-Myc 13 phosphorylation after 60 min of nitrogen starvation, but it was still significantly less than observed with MSX (Fig. 9, A and B, lanes B-E). Similarly modest nitrogen source-dependent differences were observed earlier (34). Finally, deletion of SNF1 had no detectable effect on the Gln3-Myc 13 phosphorylation profile generated by nitrogen starvation (data not shown).
Effect of Sequential MSX and Rapamycin Treatment on Gln3-Myc 13 Phosphorylation-Extracts of MSX-and rapamycin-treated cells exhibit clearly distinguishable Gln3-Myc 13 phosphorylation profiles, permitting us to test whether MSX and rapamycin inhibit tandem steps in a single regulatory pathway. In the best circumstance, treating cells with these inhibitors might generate the same effects as null mutations of the proteins they inhibit. MSX treatment is thought to generate the signal that is received by and inhibits Tor1/2 function just as does rapamycin (32). Therefore, rapamycin-induced dephosphorylation of Gln3 should be downstream of and hence predominate over MSX-generated Gln3 phosphorylation.
Two experiments were performed to test this prediction. In the first experiment, MSX and rapamycin were added either alone or together, and the cultures were harvested after 30 min of incubation. The combination of inhibitors generated nearly the same result as MSX alone (Fig. 10A, lanes B-E), arguing that (i) the MSX-inhibited step was downstream of the one inhibited by rapamycin, (ii) the inhibited steps did not occur in tandem with the MSX-inhibited step being either the later or predominant of the two, or (iii) MSX somehow inhibited rapamycin action by preventing uptake or interfering with its inhibition of Tor1/2.
The second experiment consisted of treating cells sequentially with rapamycin and MSX. In the first phase, MSX was added to a log phase culture, and following a 30-min incubation, rapamycin was added. 30 min later, the cells were harvested, and Gln3-Myc 13 phosphorylation was assayed. In the second phase, the order of inhibitor addition was reversed. We could not perform the ordered addition experiment by transferring cells from a medium containing one inhibitor to medium containing the other, because merely transferring cells to a fresh sample of the same medium, with no other change at all, generates marked differences in Gln3-Myc 13 localization (see Fig. 1B of Ref. 45).
Since, in these experiments, the first inhibitor added to the cells would be present for 60 min, it was necessary to ensure that treating cells with MSX or rapamycin for 30 min versus 60 min generated the same Gln3-Myc 13 phosphorylation profiles. This was found to be the case, except that Gln3-Myc 13 phosphorylation was a little greater at 60 min of MSX treatment than 30 min (Fig. 10, B and C, lanes B-D; note greater mobility of minor species in Fig. 10C, lanes C and E). In any case, we felt these control experiments were sufficiently positive to proceed. Two control extracts were electrophoresed on either side of samples from the ordered addition protocol: (i) cells treated with rapamycin alone, and (ii) a 1:1 mixture of cells that had been treated separately with rapamycin or MSX and frozen separately in the same tube as occurred for the experiment in Fig. 3A, lanes A and D. These control samples appear in Fig.  10D, lanes A, B, E, and F.  D) using the protocol described in the text. The numbers 1 and 2, at the top of the panel, indicate the order in which the inhibitors were added to the culture, as described in the text. For lanes A and F, cells were treated with only rapamycin. For lanes B and E, cells were treated separately with MSX or rapamycin and then separately harvested but frozen in the same microcentrifuge tube as described for Fig. 3A. These samples were included as controls.
Results from the first phase of the ordered inhibitor addition experiment, i.e. MSX added before rapamycin, are shown in Fig. 10D, lane C. Gln3-Myc 13 was highly phosphorylated, and the two faster migrating, dephosphorylated species of Gln3-Myc 13 characteristic of rapamycin-treated cells (two black dots in Fig. 10D, lanes A, E, and F) were absent. In the second phase of the experiment, i.e. rapamycin was added before MSX, hyperphosphorylated Gln3-Myc 13 was still observed as was the continued absence of dephosphorylated Gln3-Myc 13 species characteristic of rapamycin-treated cells (Fig. 10D, lane D). These results occurred in spite of the fact that Gln3-Myc 13 had been dephosphorylated by rapamycin treatment (Fig. 10, B,  lane C and D, and Fig. 10D, lanes A and F) prior to MSX being added to the cells. In other words, Gln3-Myc 13 was hyperphosphorylated even though Tor1/2 kinase activities were inhibited by rapamycin. This result argued that Tor1/2 was not responsible for the MSX-induced increase in Gln3-Myc 13 phosphorylation (Fig. 10D, lane D versus A and F).
Although results of the ordered addition experiment were clear, they were not absolute, i.e. Gln3-Myc 13 was not as extensively phosphorylated in the second phase of the experiment as when MSX was added before or in combination with rapamycin (Fig. 10D, lane D). Note that Gln3-Myc 13 was more uniformly distributed between the bands in Fig. 10D, lane D than in lane C. The species in Fig. 10D, lane D, that slightly predominated (black dot) was two species lower than the predominant pair of species in lane C (black dot). Also note as a point of reference that the uppermost species observed during rapamycin treatment alone co-migrates with the lowermost species observed after MSX treatment (Fig. 3A, lanes A-D). In the first experiment (Fig. 10A), the effects of MSX clearly predominated over those of rapamycin, whereas in the second experiment (Fig. 10D) there was detectable additivity of the effects even though the MSX-generated Gln3-Myc 13 species again predominated. In neither experiment were the data as expected if the proteins inhibited by rapamycin were downstream of the nitrogen excess signal whose generation was posited to be inhibited by MSX. DISCUSSION The above data demonstrate that MSX and rapamycin, although eliciting the same outcomes with respect to Gln3-Myc 13 nuclear accumulation and NCR-sensitive transcription, generate opposite effects on Gln3-Myc 13 phosphorylation, i.e. MSX increases it and rapamycin decreases it. The additive effects of the two inhibitors on Gln3-Myc 13 phosphorylation favor them functioning nontandemly in different pathways or in different branches of one pathway more than they support the current model positing them to function tandemly in a single pathway (32). Furthermore, inhibitors oppositely affecting Gln3-Myc 13 phosphorylation, but eliciting the same outcomes on its intracellular localization, suggest that Gln3-Myc 13 phosphorylation/ dephosphorylation, detected thus far in this and the work of others, is likely associated with something other than determining Gln3 intracellular localization. This suggestion is also consistent with the lack of correlation observed between Gln3-Myc 13 phosphorylation and intracellular localization in cells provided with various nitrogen sources (34).
We also showed that Gln3-Myc 13 phosphorylation levels are regulated by at least three separate mechanisms. The first depends on the carbon regulatory kinase, Snf1. The second is independent of it; MSX-induced phosphorylation occurs via the Snf1independent mechanism. Third, Gln3-Myc 13 is dephosphorylated in response to rapamycin inhibition of Tor1,2. The existence of Snf1-dependent Gln3-Myc 13 phosphorylation is not surprising as it had been observed earlier in response to carbon starvation of ammonia-grown cells (33,34). The novel observation of the pres-ent work is that carbon starvation also triggers Gln3-Myc 13 phosphorylation in a snf1⌬. This phosphorylation level increases to even higher levels in the presence of wild type Snf1. However, the profile of Snf1-dependent and -independent phosphorylation responses differ. Snf1-independent phosphorylation appears to generate three characteristic Gln3-Myc 13 species, whereas the Snf1-dependent Gln3-Myc 13 species are more heterogeneous; appearing largely as a fainter, slower mobility smear. This would be an expected result if the Snf1-independent mechanism possessed greater substrate specificity. However, given the number of phosphorylated Gln3-Myc 13 species we can detect, in addition to those in the smear that we can not individually resolve, such an interpretation is probably speculative.
Although our experiments demonstrate increased Gln3-Myc 13 phosphorylation derives from at least two distinguishable regulatory pathways, they cannot identify the kinases directly responsible for it. The literature supports Snf1 and Tor1/2 as the most likely candidates: (i) both kinases interact with Gln3 in two-hybrid assays (16,33) and (ii) both Tor1/2 and Snf1 phosphorylate Gln3 in vitro (16,33). Consistent with Snf1 phosphorylating Gln3-Myc 13 during carbon starvation is the correlation that Snf1 activity is activated by centrifugation (41), a characteristic also observed in our Snf1-dependent data (Fig. 8C). To conclude that Tor1/2 phosphorylate Gln3-Myc 13 is more difficult. First, the data in Fig. 10D demonstrate that MSX induces Gln3-Myc 13 phosphorylation in the presence of rapamycin, which inactivates Tor1/2. Second, if Tor1/2 are responsible for Gln3-Myc 13 phosphorylation, a dilemma arises in trying to reconcile current data with the earlier model of Tor regulation (9,32). The dilemma arises from the fact that both MSX treatment and carbon starvation of ammonia-grown cells (urea is also degraded solely to ammonia) are thought to generate nitrogen starvation, but yet unlike rapamycin treatment, they increase Gln3-Myc 13 phosphorylation. According to the existing model, nitrogen starvation and MSX and rapamycin treatment are all envisioned to inhibit Tor1/2 kinase activities, which in turn should all result in dephosphorylation of Gln3-Myc 13 (32). By this reasoning, we can account for present data only by suggesting (i) that Tor1/2 mediate Gln3 phosphorylation in their inactive forms or (ii) that Tor1/2 are not directly responsible for MSX-induced and/or nitrogen starvation-induced Gln3 phosphorylation. Although we do not favor it, one possibility that the data in Fig. 10D do not eliminate is that Tor1/2 were responsible for the most highly phosphorylated species of Gln3-Myc 13 in lane C (where Tor1/2 were not inhibited by rapamycin for the first 30 min) that were missing in lane D.
The final dilemma generated by the above experiments is that MSX and rapamycin treatment, in our hands, elicit effects on Gln3-Myc 13 phosphorylation opposite to those reported previously (32). Since the strains used here were the same as those used earlier (9, 32), we searched for differences in culture conditions. Although such differences were found, they do not account for the results obtained. Our failed attempts to rectify the differing observations notwithstanding, there is clearly one or more differences between present and earlier experiments that remain to be identified.
Another question that arose during this work was whether MSX treatment elicited Gln3-Myc 13 phosphorylation profiles similar to those that occur in response to nitrogen starvation as expected if MSX treatment inhibits glutamine formation, the signal of excess nitrogen proposed earlier to be relayed to Tor1/2 (32). Our MSX-induced Gln3-Myc 13 phosphorylation profiles share characteristics with profiles generated both by carbon and nitrogen starvation. To conclude the Gln3-Myc 13 phosphorylation profile resembles that of nitrogen starvation, however, would require that MSX treatment generates a more extreme case of nitrogen starvation than incubating cells in nitrogen-free medium for 60 min. Although certainly not difficult to imagine, if this were the case, then one would also expect Gat1 to be accumulated in the nucleus following MSX treatment, because like Gln3, Gat1-mediated transcription is NCR-sensitive. Furthermore, Gat1 has been proposed to be subject to the same phosphorylation/dephosphorylation regulatory mechanism as Gln3 (9). This expectation contrasts with the report that MSX does not elicit nuclear localization of Gat1 (32). On the other hand, MSX-elicited Gln3-Myc 13 phosphorylation does possess a profile similar to the one seen in carbonstarved, wild type cells and identical to the one caused by carbon starvation of a snf1⌬ mutant. This suggests MSX treatment directly or indirectly triggers responses shared by those that occur with carbon starvation.
Most importantly, the MSX data reported above argue that both phosphorylated and dephosphorylated forms of Gln3-Myc 13 can be accumulated in the nucleus, a conclusion also reported in studies of the effects of carbon starvation on Gln3 regulation (33). These observations make it difficult to understand how data concerning Gln3 phosphorylation levels, which we and others have detected thus far, can be rectified with the current regulatory model. At the same time, we must hasten to emphasize that these difficulties would quickly disappear if the phosphorylated species detected so far are not the ones associated with nitrogen-dependent Gln3 regulation. Current data also prompt a new question, if phosphorylation levels we detect are not regulating Gln3 intracellular localization, then what are their functions? The importance of this question to future work is emphasized both by the variety of environmental perturbations that can trigger Gln3 phosphorylation and the number of regulatory systems that are involved in the process.
Finally, the data emphasize the importance of the ability of Snf1 to influence Gln3 phosphorylation both in the presence of high glucose and during carbon starvation. The ramifications of this observation may extend to the technical design of phosphorylation/dephosphorylation experiments in general, especially those involving carbon starvation or limitation. Centrifugation-mediated activation of Snf1 kinase was originally reported with respect to carbon regulation (41), but the above data suggest cell harvest methods could influence the data obtained with other proteins as well. Such artifacts appear to be avoided, or at least decreased, when cells are harvested by filtration rather than centrifugation.