Snf1 Phosphorylates Adenylate Cyclase and Negatively Regulates Protein Kinase A-dependent Transcription in Saccharomyces cerevisiae *

Background: The Snf1/AMPK and PKA pathways are crucial for nutrient sensing and utilization in yeast. Results: A novel cross-talk mechanism between Snf1/AMPK and PKA is proposed. Conclusion: Snf1 and Cyr1 interact in a nutrient-independent manner. Active Snf1 phosphorylates Cyr1 and negatively regulates cAMP content and PKA-dependent transcription. Significance: This is the first evidence of regulation of PKA pathway by Snf1/AMPK. In eukaryotes, nutrient availability and metabolism are coordinated by sensing mechanisms and signaling pathways, which influence a broad set of cellular functions such as transcription and metabolic pathways to match environmental conditions. In yeast, PKA is activated in the presence of high glucose concentrations, favoring fast nutrient utilization, shutting down stress responses, and boosting growth. On the contrary, Snf1/AMPK is activated in the presence of low glucose or alternative carbon sources, thus promoting an energy saving program through transcriptional activation and phosphorylation of metabolic enzymes. The PKA and Snf1/AMPK pathways share common downstream targets. Moreover, PKA has been reported to negatively influence the activation of Snf1/AMPK. We report a new cross-talk mechanism with a Snf1-dependent regulation of the PKA pathway. We show that Snf1 and adenylate cyclase (Cyr1) interact in a nutrient-independent manner. Moreover, we identify Cyr1 as a Snf1 substrate and show that Snf1 activation state influences Cyr1 phosphorylation pattern, cAMP intracellular levels, and PKA-dependent transcription.

All organisms need to match growth and metabolism with the availability of nutrients. In the budding yeast Saccharomyces cerevisiae, several mechanisms of sensing of the carbon source present in the environment have evolved (see Refs. 1-3 for extensive review). Nutrients, and in particular glucose, influence the yeast cell metabolic behavior through the activa-tion of signaling pathways, which in turn finely tune transcription and enzymatic activities (4), as well as cell growth and cell cycle (5).
The main signaling pathway activated by glucose in yeast is the PKA pathway, which shows extensive involvement in growth, proliferation, and metabolism (2,6). The second messenger responsible for the activation of PKA is cAMP, which is synthesized by adenylate cyclase (Cyr1) upon stimulus from the G-proteins Gpa2 and the RAS pathway (1,7). cAMP binds to a homodimer of PKA regulatory subunits Bcy1, causing its dissociation from the dimer of catalytic subunits (composed by two of the partially redundant proteins Tpk1, Tpk2, and Tpk3), thus activating the kinase (8). Active PKA has a broad influence on transcription. During exponential growth and in the absence of stress, PKA phosphorylates the transcription factors Msn2,4, thus causing their confinement to the cytoplasm (9). Msn2,4 activate the transcription of genes with promoters containing stress response element sequences, comprising genes coding for chaperones, antioxidant proteins, and proteases and, more in general, for elements necessary to respond to various stress types (10). Besides Msn2,4, PKA controls several other transcription factors including Rim15 (11) and Rap1 (12). Furthermore, PKA directly regulates important metabolic pathways such as glycolysis (6). It phosphorylates proteins catalyzing key enzymatic steps, such as both isoforms of pyruvate kinase, Cdc19 and Pyk2 (13)(14)(15).
On the contrary, the signaling pathway centered on the kinase Snf1 is required to adapt to nutrient limitation and to utilize alternative carbon sources, such as sucrose or ethanol (16). Snf1 shares evolutionary conserved functions with the AMP-activated kinase (AMPK) 4 in higher eukaryotes, constituting the well defined family of regulators of cellular energy homeostasis in eukaryotes (17). Snf1 is activated through the phosphorylation of Thr 210 by one of the three constitutively active upstream kinases Sak1, Tos3, and Elm1 (18,19). In response to high glucose concentrations, Snf1 is inactivated through dephosphorylation of Thr 210 by the Glc7 protein phosphatase (also known as PP1), which is targeted to Snf1 by the adaptor subunit Reg1 (20,21). However, recent studies highlighted Snf1 function also in glucose-repressed conditions, as well as its partial activation and its role in the regulation of cell cycle (22)(23)(24)(25). The most studied function of Snf1 is the regulation of transcription, involving more than 400 genes (26). Active Snf1 causes the translocation to the cytoplasm of the inhibitor Mig1, responsible for the repression of over 90 genes (27). Besides Mig1, Snf1 regulates the activity of various transcription factors. Cat8 and Sip4, which bind carbon source response elements, mostly regulate the expression of gluconeogenic genes (28), and are activated by Snf1-dependent phosphorylations (29,30). Furthermore, Snf1 regulates through phosphorylation several metabolic enzymes, such as Pfk27, Gpd2, and Acc1 (6), to promote carbon saving in conditions of nutrient scarceness.
In S. cerevisiae, several examples of cross-talk between the Snf1 and PKA pathways have been reported. PKA phosphorylates and contributes to the regulation of the Snf1-activating kinase Sak1 (31) and the ␤ subunit Sip1, preventing the localization of the Snf1 complex to the vacuole (32). Snf1 and PKA share common downstream targets. PKA phosphorylates and deactivates Adr1, the transcriptional activator of several glucose repressed genes, whereas Snf1 indirectly causes its dephosphorylation and activation (33,34). Surprisingly, both PKA and Snf1 act as repressors of the transcription factor Msn2, but, whereas PKA is the main regulator of this factor, Snf1 targets it only to shut off transcription of the stress response element regulon after prolonged stationary phase (9,35,36). Remarkably, a direct regulation of the Snf1 orthologue AMPK by PKA has been demonstrated in human adipocytes, where PKA phos-phorylates Ser 173 of AMPK preventing the phosphorylation of residue Thr 172 , requested for activation (37). Both residues are conserved in S. cerevisiae (Thr 210 and adjacent Ser 211 ), thus a similar mechanism has been hypothesized in yeast (31).
Here, we propose a novel cross-talk mechanism between Snf1/AMPK and PKA pathways. Seeking for new Snf1 targets, we identify adenylate cyclase (Cyr1) as a Snf1-interacting protein. We show that Cyr1 is a Snf1 target in vitro and that its phosphorylation pattern is altered in vivo in the snf1⌬ mutant compared with the wild type. Strikingly, we demonstrate that constitutive activation of Snf1 reduces cAMP intracellular content and functionally impairs PKA activity as a regulator of glucose-repressed genes.

Experimental Procedures
Yeast Strains and Growth Conditions-S. cerevisiae strains used in this study are listed in Table 1. Synthetic medium contained 2, 5, or 0.05% glucose or 2% ethanol (as indicated in the figures), 6.7 g/liter of yeast nitrogen base (Difco), 50 mg/liter of required nutrients at standard pH (5.5).
Cell density of liquid cultures grown at 30°C was determined with a Coulter counter on mildly sonicated and diluted samples or spectrophotometrically at 600 nm. All experiments were performed with cells in exponential phase of growth, at cell densities between A 600 nm ϭ 0.1 (2 ϫ 10 6 cells/ml) and A 600 nm ϭ 0.7 (1.3 ϫ 10 7 cells/ml).
Recombinant and Genetic Techniques-DNA manipulation and yeast transformations were carried out according to standard techniques. To obtain the tagged strains, proteins were tagged with a C-terminal 4HA or 9myc epitope by an in locus 3Ј in-frame insertion. Escherichia coli DH5␣ and BL21 (DE3) [pLysE] were used in cloning experiments and for expression of recombinant proteins, respectively. The HpaI-XhoI Protein Extraction, Immunoblotting, and Immunoprecipitation-Samples of cells were harvested and lysed using icecold lysis buffer (50 mM Tris, pH 7.5, 150 mM NaCl, 0.1% Nonidet P-40, 10% glycerol) plus 1 mM PMSF, proteases inhibitor mix (Complete EDTA free protease inhibitor mixture tablets; Roche), and phosphatase inhibitor mix (Sigma). An equal volume of acid-washed glass beads (Sigma) was added, and cells were broken by 20 vortex/ice cycles of 1 min each. Extracts were transferred to new tubes and clarified by centrifugation. Protein concentration was determined using the Bio-Rad protein assay. Crude protein extracts were incubated 2 h at 4°C with anti-HA affinity matrix (Roche) to immunoprecipitate Snf1-HA protein complexes or with IgG-agarose beads (Sigma) to immunopurify Cyr1-TAP. The immunocomplexes obtained after removal of supernatants were washed three times with lysis buffer and once with 25 mM MOPS, pH 7.5. SDS sample buffer was then added to the beads, and the samples were heated for 5 min at 95°C. Western blot analysis was performed using anti-TAP antibody (1:2500; Thermo Scientific, Open BioSystems), anti-HA antibody (1:1000, 12CA5; Roche), anti-myc antibody (1:1000, 9E10; Santa Cruz), anti-Cyr1 antibody (1:1000, yC-20, Santa Cruz), and anti-Swi6 (1:1000, kindly provided by L. Breeden). For the analysis of Snf1 interactors, Snf1-HA protein complexes were immunopurified from 270 mg of total protein extract and resolved by SDS-PAGE followed by staining with GelCode TM blue stain reagent (Pierce) for MS analysis. Bands were excised and treated as reported in the section "Sample Preparation and Mass Spectrometry Techniques." For the analysis of Cyr1 phosphorylation, Cyr1-TAP was immunopurified from 9 mg of total protein extract and resolved by SDS-PAGE followed by silver staining, according to the manufacturer's instructions (GE Healthcare).
In Vitro Phosphorylation Assays-In vitro phosphorylation of recombinant His 13 -Mig1(207-413) was carried out as in Ref. 39. In vitro phosphorylation of recombinant His 6 -Cyr1(335-1066) (4 g of purified protein) was performed in a buffer containing 20 mM HEPES, pH 7, 100 mM NaCl, 0.5 mM EDTA, 0.5 mM DTT, 5 mM MgAc, using protein kinase Snf1-HA or Snf1-T210A-HA immunopurified from yeast cells growing in 2% glucose. The reaction was started by adding 0.24 M [␥-32 P]ATP (specific radioactivity, 2000 cpm/pmol) and incubated at 30°C for 30 min. The reaction was stopped by adding 4ϫ SDS sample buffer and then heated for 5 min at 95°C, and proteins were separated by SDS-PAGE. Phosphorylated bands were visualized by autoradiography.
Sample Preparation and Mass Spectrometry Techniques-The protein content of IP Snf1-HA sample and IP untagged control sample were resolved by SDS-PAGE, analyzed, and compared using a GS-800 TM densitometer and Quantity One analysis software. Each protein band was quantified by densitometric analysis. Only bands exclusively present on IP Snf1-HA sample or bands whose A value differed at least 0.1 in the comparison between IP Snf1-HA sample and IP untagged control sample were excised and analyzed by mass spectrometry. These data generate the list of Snf1-interacting proteins reported in Table 2.
Bands were excised after staining, reduced with DTT, and alkylated with 55 mM iodoacetamide at room temperature for 45 min. Bands were dried, soaked with ammonium bicarbonate 0.1 M, and digested overnight with trypsin sequence grade, at 37°C using a protease: protein ratio (1:10). Tryptic digests were extracted with 50% acetonitrile (ACN) in 0.1% TFA, desalted/ concentrated on a ZipTipC18 (Millipore), and analyzed by mass spectrometry. LC-electrospray ionization-MS/MS analysis was performed on a Dionex UltiMate 3000 HPLC System with a PicoFrit ProteoPrep C18 column (200 mm, internal diameter of 75 m) (New Objective). Gradient: 1% ACN in 0.1% formic acid for 10 min, 1-4% ACN in 0.1% formic acid for 6 min, 4 -30% ACN in 0.1% formic acid for 147 min, and 30 -50% ACN in 0.1% formic for 3 min at a flow rate of 0.3 l/min. The eluate was electrosprayed into an LTQ Orbitrap Velos (Thermo Fisher Scientific) through a Proxeon nanoelectrospray ion source (Thermo Fisher Scientific). Data acquisition was controlled by Xcalibur 2.0 and Tune 2.4 software (Thermo Fisher Scientific). The LTQ-Orbitrap was operated in positive mode in data-dependent acquisition mode to automatically alternate between a full scan (m/z 350 -2000) in the Orbitrap (at resolution, 60,000; AGC target, 1,000,000) and subsequent CID MS/MS in the linear ion trap of the 20 most intense peaks from full scan (normalized collision energy of 35%, 10-ms activation). Data Base searching was performed using the Sequest search engine contained in the Proteome Discoverer 1.1 software (Thermo Fisher Scientific). The following parameters were used: 10 ppm for MS and 0.5 Da for MS/MS tolerance, carbamidomethylation of Cys as fixed modification; oxidation of Met and phosphorylation of Ser, Tyr, and Thr, as variable modifications, trypsin (two misses) as protease. To generate the list of phosphosites reported in Table 3, we considered only the sites with the highest X Correlation value (Xcorr) in Sequest (Ն1.5), the rank value of 1 and the best fragmentation pattern, selected manually after visual inspection of the MS/MS spectra. Three different tools for phosphorylation site prediction where applied: NetPhos Yeast 1.0, NetPhosK 1.0 Server-CBS, and Phosida.
cAMP Assay-cAMP was measured using the cAMP Biotrak EIA Assay (GE Healthcare) following the manufacturer's instructions. Sample preparation was carried out as in Ref. 40.

Results
Snf1 Interacts with Adenylate Cyclase-Snf1 is known to regulate glucose metabolism and to promote the utilization of alternative carbon sources mainly through the derepression of a large transcription regulon and the phosphorylation of metabolic enzymes (16). However, because the cellular metabolic behavior could be the result of the action of different and somehow opposite signaling pathways, we wondered whether Snf1 could directly regulate such pathways. To detect possible new targets of Snf1, we chose to use a co-immunoprecipitation/MS approach, immunoprecipitating myc-tagged Snf1 and detecting co-immunoprecipitated proteins with mass spectrometry after resolution with SDS-PAGE. We performed this experiment with protein extracts of exponentially growing cells in 2% glucose, because we already demonstrated that in this condition Snf1 is at least partially functional (22,25).
Interestingly, we identified a total of 50 potentially interacting proteins of which only 16 have been already identified as Snf1 interactors (Fig. 1A and Table 2) (41)(42)(43). We detected most of the components of the complex of Snf1 and of the Snf1  1B and Table 2). Remarkably, we identified many new Snf1 potential interactors involved in protein synthesis, such as components of the large ribosomal subunit, and in important metabolic processes, such as glycolytic enzymes (Fig.  1B and Table 2). Strikingly, one of the newly identified interactors was adenylate cyclase (Cyr1), the essential enzyme responsible for the synthesis of cAMP, activator of PKA ( Fig. 1 and Table 2). To validate the interaction, we performed co-immunoprecipitation experiments, clearly detecting myc-tagged Snf1 in Cyr1-TAP immunocomplexes ( Fig. 2A). Moreover, the interaction was not indirectly caused by Snf1 and Cyr1 being common substrates of the phosphatase complex Reg1-Glc7, because it was detectable also in extracts from reg1⌬ cells (Fig. 2B). Interestingly, the interaction was not dependent on the glucose concentration nor the carbon source (Fig. 2, C and D) and was detectable in the presence of a catalytic-deficient mutant Snf1, Snf1-K84R, or in the presence of a constitutively active mutant Snf1-G53R (Fig. 2E). Furthermore, switching bait and prey proteins and detecting Cyr1-TAP in Snf1-HA immunocomplexes, we were able to demonstrate that the interaction was particularly stronger in the presence of a Snf1-K84R mutant (Fig. 2F). This finding was of particular interest, because it was previously shown that in glucose-repressed conditions, the catalytic-deficient Snf1 mutant interacts more strongly than the WT with Mig1, a well known Snf1 substrate (44). Taken together, these data demonstrate the physical interaction of Snf1 with Cyr1 and support the possibility of adenylate cyclase being a substrate of Snf1.
Loss of Snf1 Causes an Alteration in the Phosphorylation Pattern of Adenylate Cyclase-Cyr1 is a phosphoprotein, and several phosphorylated residues have been identified in proteomic studies (45)(46)(47)(48). Moreover, the sequence of Cyr1 presents five phosphorylatable residues (Ser 241 , Ser 554 , Thr 736 , Ser 1594 , and Ser 1978 ), two of which were identified as phosphosites in vivo, that are contained in perfect Snf1 phosphorylation consensus motifs (Ser 241 and Ser 554 ). Remarkably, Ser 241 has been recently identified as being phosphorylated in a Snf1-dependent fashion (24). However, no specific function until now has been assigned to any phosphorylation site.
Because the physical interaction with Snf1 suggested Cyr1 as a possible target of the kinase, we investigated whether the presence or the absence of Snf1 could influence the phosphorylation status of adenylate cyclase in vivo. We therefore immunoprecipitated TAP-tagged Cyr1 from WT and snf1⌬ cell extracts and, after resolution on SDS-PAGE and silver staining, identified the phosphorylated residues by MS analysis (Fig. 3, A and B, and Table 3). Remarkably, in a WT strain we identified 19 phosphosites, of which only 7 were previously reported, providing an extensive map of adenylate cyclase phosphorylation. Strikingly, a subset of these phosphosites, almost completely localized in the region flanking the RAS-associating domain, turned out to be Snf1-dependent, not being detectable in the snf1⌬ strain. Surprisingly, none of these sites was one of the five contained in Snf1 consensus sequences discussed above. Interestingly, we also identified seven additional phosphosites that were exclusively present in a snf1⌬ strain, which could be dependent on kinases normally inhibited by Snf1 (Fig. 3A). When we checked whether the identified sites were contained in consensus sequences of a specific kinase, we failed to identify any obviously responsible for the phosphorylation events. These data provide a substantial advance in mapping possible regulatory post-translational modifications of adenylate cyclase and demonstrate the necessity of Snf1 to preserve the proper phosphorylation pattern of Cyr1. Snf1 Phosphorylates the RAS-associating Domain of Adenylate Cyclase in Vitro-Because we demonstrated that a lack of Snf1 could influence the phosphorylation pattern of Cyr1 in vivo, we sought for direct kinase activity of Snf1 on adenylate cyclase in vitro. Because we wanted to demonstrate a direct kinase activity of immunopurified Snf1, discerning from phosphorylation events dependent on kinases that co-immunoprecipitate with Snf1 (Table 2), we searched for the most suitable catalytic-deficient mutant of Snf1 to be included as a control. We therefore tested different Snf1 mutants for their ability to phosphorylate recombinant Mig1 in vitro. Strikingly, the unphosphorylatable mutant Snf1-T210A turned out to be more impaired than the Snf1-K84R mutant (Fig. 4A) and was chosen as a control for subsequent analysis. We then purified from E. coli a His 6 -tagged region of Cyr1 containing the whole RASassociating domain (RAD), four of five (Ser 435 , Ser 527 , Thr 631 , and Ser 645 ) residues whose phosphorylation was found to be Snf1-dependent (Fig. 3A) and two residues contained in Snf1 phosphorylation consensus sites (Ser 554 and Thr 736 ). We then performed an in vitro kinase assay with ␥-32 P-labeled ATP, using the purified protein as substrate and immunoprecipitated HA-tagged Snf1 and Snf1-T210A as kinases. In the presence of the WT kinase, but not in the presence of the Snf1-T210A mutant, the His 6 -Cyr1(335-1066) recombinant protein incor-porated ATP, demonstrating that this domain of adenylate cyclase is a bona fide substrate of Snf1 in vitro (Fig. 4B).
To determine which residues of His 6 -Cyr1(335-1066) were phosphorylated in vitro by Snf1, we carried out a phosphorylation assay, followed by SDS-PAGE resolution and MS analysis. Remarkably, all the sites whose phosphorylation was Snf1-dependent in vivo ( Fig. 3A and Table 3) turned out to be phosphorylated by Snf1 in vitro (Ser 435 , Ser 527 , Thr 631 , and Ser 645 ) ( Fig. 4C and Table 4), and among these only Ser 435 and Ser 527 were slightly phosphorylated in the presence of Snf1-T210A, possibly evidencing a residual activity of the mutant kinase (Table 4). Remarkably, two phosphosites that turned out to be Snf1-independent in vivo (Ser 483 and Ser 524 ) and two sites that were phosphorylated upon SNF1 deletion (Thr 1043 and Thr 1046 ) were also found to be phosphorylated in vitro by Snf1 (Fig. 4C and Table 4). These data demonstrate that although the in vivo Snf1-dependent phosphosites are not contained in perfect Snf1 consensus sequences, they are potential targets of direct phosphorylation by Snf1.
Catalytic Activation of Snf1 Negatively Regulates cAMP Content-The function of adenylate cyclase is to catalyze the synthesis of cAMP, the second messenger responsible for the activation of PKA, in response to stimuli from the RAS and GPCR systems (1). Having shown that a lack of Snf1 influences the Cyr1 phosphorylation pattern and that Snf1 can phosphorylate Cyr1 in vitro, we wondered whether activation of Snf1 could influence the activity of adenylate cyclase. cAMP is rapidly synthesized at high levels after glucose refeeding to starved cells but is present at detectable basal levels also in exponentially growing cells (40). Therefore we measured cAMP levels in log phase cells growing in 2% glucose, either lacking Snf1 or
Strikingly, although a lack of Snf1 did not influence the cAMP content of cells growing in glucose-repressed conditions, the presence of the Snf1-G53R mutant nearly halved the cAMP levels, as compared with the WT (Fig. 5). These data demonstrate that the catalytic activity of Snf1 can reduce the intracellular content of cAMP in glucose-repressed condition.
Catalytic Activation of Snf1 Negatively Influences PKA-dependent Transcriptional Expression-PKA regulates several cellular processes including nutrient sensing, energy metabolism, cell cycle progression, thermotolerance, osmotic shock tolerance, sporulation, pseudohyphal growth, aging, and autophagy (1). Some of these processes, such as thermotolerance, are usually used to indirectly measure the activity of the PKA pathway (8). However, Snf1 regulates these processes independently from PKA (50), and the use of these methods to measure the influence of Snf1 catalytic activity on the PKA pathway was therefore discarded.
Thus, we chose to indirectly measure the activity of the PKA pathway through the effect on the transcription of genes which are regulated by transcription factors that are PKA targets. Some transcription factors are shared downstream targets of Snf1 and PKA (9,(32)(33)(34)(35)(36); thus the strategy required the identification of Snf1-independent target genes.
Several PKA-regulated genes have been mapped (51), and we checked the dependence of selected genes (HXT1, HXT7, and ATR1) during a nutritional shift-up (from ethanol to glucose containing medium) using a cyr1⌬pde2⌬yak1⌬ strain, in which the disruption of adenylate cyclase impairs the activation of PKA (35) (Fig. 6A). Because HXT1, whose expression was PKAdependent, is also known to be regulated by Snf1 (52), it was included in subsequent analyses as a control.   OCTOBER 9, 2015 • VOLUME 290 • NUMBER 41

JOURNAL OF BIOLOGICAL CHEMISTRY 24721
To assess whether Snf1 catalytic activity influenced the expression of PKA-regulated genes, we tested the expression of HXT1, HXT7, and ATR1 in the presence of the aforementioned Snf1-G53R mutant. HXT1 turned out to be down-regulated in a Snf1-G53R strain grown in 2% glucose (Fig. 6B), thus proving that this mutant is constitutively active. The PKA-activated gene ATR1 was slightly down-regulated in the presence of a Snf1-G53R mutant, but more strikingly, in the same strain the PKA-repressed gene HXT7 was 5-fold derepressed (Fig. 6B). HXT7 is known to be Snf1-independent (4), and we also checked its independency from the Snf1-inhibited repressor Mig1 using a mig1⌬ strain (Fig. 6C), demonstrating that this gene is a useful PKA-dependent/Snf1-independent reporter gene. To further characterize the effect of Snf1 activity on PKAdependent genes, we tested the expression of HXT1, HXT7, and ATR1 in the presence of a Snf1-G53R mutant in 0.05, 2, and 5% glucose. In cells growing in 0.05% glucose, where PKA activity is already low and Snf1 activity is high, all three genes were equally expressed in both the WT and the Snf1-G53R strain (Fig. 6D). Strikingly in 5% glucose, where PKA activity should be maximal, activation of Snf1 through the G53R mutation still causes the deregulation of PKA-dependent genes (Fig. 6D). Altogether, these data demonstrate that Snf1 catalytic activity functionally influences the PKA pathway, determining an alteration in the expression of PKA-dependent genes.
Mutation of Snf1-dependent Phosphorylation Sites of Cyr1 Influences PKA-related Phenotypes-We showed that in the presence of active Snf1, cAMP levels were low (Fig. 5). In yeast, two phosphodiesterases, Pde1 and Pde2, degrade cAMP, and Pde1 is subjected to activation by PKA phosphorylation, generating a negative feedback on the pathway (53). Thus, to assess whethertheobservedeffectofSnf1oncAMPcontentandPKA-dependent gene expression was due to activation of the phosphodiesterases, we tested the expression of HXT1, HXT7, and ATR1 in a pde1⌬pde2⌬ strain in the presence of a Snf1-G53R mutant. Interestingly, deregulation of PKA-dependent genes by Snf1 activation was observed even in the absence of the phosphodiesterases (Fig. 7A). Only the derepression of HXT7 was somehow reduced, but the basal levels of the mRNA were already high in the pde1⌬pde2⌬ strain compared with a WT (data not shown).
We therefore tested the possibility that the PKA transcriptional activity was instead dependent upon phosphorylation of adenylate cyclase. Thus, we generated a mutant Cyr1 in which the four Snf1-dependent phosphosites near the RAD (Ser 435 , Ser 527 , Thr 631 , and Ser 645 ), identified in vivo (Fig. 3A and Table
Taken together, these data indicate that Cyr1 phosphorylation, rather than phosphodiesterase activation, is important for the control of HXT7 expression. In particular, the phosphory-lation status of four Snf1-dependent residues seems to directly influence PKA activity.

Discussion
A major challenge in the understanding of a signaling pathway is the discovery of cross-talk with other pathways. Because Snf1 is mainly known for its role in shaping metabolism to guarantee cell growth in nutritional scarceness, it seemed likely that it could regulate other signaling pathways that cause opposite outcomes. In search for new targets of Snf1, we identified a set of previously unknown Snf1-interacting proteins ( Fig. 1 and Table 2), including adenylate cyclase (Cyr1), suggesting a possible cross-talk between Snf1 and PKA pathways. PKA was a possible candidate as a Snf1-regulated pathway, because it is activated in conditions of nutrient abundance. The two pathways share common downstream targets, such as transcription factor Adr1 (9, 33-36), and PKA has been reported to regulate FIGURE 6. Expression of PKA-dependent genes in the presence of constitutively activated Snf1. A, real time PCR quantification of mRNAs of the indicated genes in cells in exponential phase of growth in 2% ethanol and after addition of 5% glucose (final concentration). B, real time PCR quantification of mRNAs of the indicated genes in cells expressing Snf1 or Snf1-G53R in exponential phase of growth in 2% glucose. C, real time PCR quantification of mRNAs of the indicated genes in WT and mig1⌬ cells in exponential phase of growth in 2% glucose. D, real time PCR quantification of mRNAs of the indicated genes in cells in exponential phase of growth in 0.05% glucose, 2% glucose, or 5% glucose. OCTOBER 9, 2015 • VOLUME 290 • NUMBER 41 the recruitment of Snf1 to the vacuole and to contribute to the inactivation of the Snf1-activating kinase Sak1 (31,32). This regulation seem to be evolutionarily conserved, because in mammalian cells PKA phosphorylates and inactivates the Snf1 orthologue AMPK (37). Thus, showing the converse regulation of the PKA pathway by Snf1, we suggest a novel cross-talk mechanism. Although we showed that the interaction of Snf1 and Cyr1 is independent from nutritional conditions (Fig. 2, C and D) and from Snf1 activity (Fig. 2, E and F), we demonstrated that the lack of Snf1 could alter the phosphorylation status of  . Map of the cross-talk between the Snf1 and PKA pathways. Schematization of the basic circuitry of the cross-talk between the Snf1 and PKA pathway, applied to the expression of the PKA-repressed gene HXT7. Subsidiary pathways were omitted for simplification. Dotted lines and solid lines indicate inefficient and efficient processes respectively. In high glucose condition, Cyr1 produces high cAMP levels, which activate PKA that can inhibit Msn2,4 translocation into the nucleus and Sak1 activity. Therefore, expression of HXT7 is low and Sak1 inefficiently phosphorylates Snf1. In low glucose, Snf1 is active and inhibits Cyr1, thus resulting in lower cAMP levels and lower PKA activity; Msn2,4 is nuclear and activates HXT7 expression.

Snf1/AMPK Regulates cAMP/PKA Pathway
Cyr1 in vivo and that Snf1 can phosphorylate an important regulatory domain of adenylate cyclase in vitro (Figs. 3 and 4B). Moreover, by mapping the in vitro phosphorylated sites, we demonstrated that Snf1 directly phosphorylates the phosphosites, which turned out to be Snf1-dependent in vivo ( Fig. 4C and Table 4). Strikingly, by using a constitutively activated Snf1 mutant, we highlighted an inverse correlation between the activation state of the kinase and the intracellular content of cAMP (Fig. 5). To determine whether the activity of Snf1 had actual influence on PKA activity, we established a method based on the expression of an endogenous gene used as a reporter. The gene HXT7, identified as a target of the PKA-repressed transcription factors Msn2 and Msn4 (51), was chosen for this purpose. In fact, it showed PKA-dependent repression in a strain lacking adenylate cyclase (Fig. 6A), and it was independent from direct Snf1 activity in a mig1⌬ strain (Fig. 6C). Moreover, HXT7 was already reported to be Snf1-independent (4), and no interaction of Snf1-regulated transcription factors with the promoter of HXT7 was reported in specialized databases. HXT7 turned out to be dramatically derepressed in a Snf1-G53R strain (Fig. 6B), thus linking Snf1 activation to the transcription of PKA-dependent genes. Moreover, the phosphomimetic mutation of Snf1-dependent Cyr1 phosphosites induced alteration of heat shock resistance and caused HXT7 derepression. HXT7 encodes for a high affinity glucose transporter, specifically expressed in low glucose media (55), and the dependence of its expression from an interplay between PKA and Snf1, which act as opposite pathways, seems therefore biologically consistent. The regulation of PKA pathway by Snf1 seems likely to occur in nutrient limiting conditions, but the study in glucose abundant medium allowed the use of the otherwise lethal mutant snf1⌬ and the activated mutant Snf1-G53R in the absence of a high basal activity of Snf1. Altogether, these findings (summarized in Fig. 8) added a new link in the circuitry involving these two major signaling pathways, with Snf1 regulating both directly and indirectly the content of cAMP and thus the activation state of PKA, consequently enhancing the expression of genes required in conditions of nutrient scarceness and stress.
Author Contributions-R. N. performed experiments, analyzed data, and wrote the paper. F. T., M. G., A. C., V. R., and S. N. performed experiments and analyzed data. G. T. and P. C. analyzed data and wrote the paper.