Snf1 protein kinase, like AMPK, exists in three heterotrimeric forms containing the Snf1 (α) catalytic subunit; the Snf4 (γ) regulatory subunit; and one of the β subunits, Gal83, Sip1, and Sip2 (43), referred to henceforth as Snf1-Gal83, Snf1-Sip1, and Snf1-Sip2, respectively. The three β subunits have largely overlapping functions, and all three must be mutated to confer the Snf⁻ phenotype (29); however, they exhibit distinct subcellular localizations (15, 36) and play distinct roles in various cellular processes (1, 20, 29, 34, 37).

Three kinases that activate Snf1 by phosphorylation of its activation-loop threonine have been identified: Pak1, Tos3, and Elm1 (16, 24, 32). All three Snf1 protein kinase kinases contribute to activation of Snf1 in vivo, as the cognate genes for all three must be deleted to confer the Snf⁻ phenotype of failure to grow on nonpreferred carbon sources (16, 32). In addition, Elm1 has roles in controlling cell morphology, cell cycle progression, and filamentous invasive growth (2, 3, 7, 30, 31) that are not known to be related to Snf1 function. This kinase cascade is conserved in mammals, where the tumor suppressor kinase LKB1, an ortholog of Pak1, Tos3, and Elm1, phosphorylates and activates AMPK (13, 16, 40).

Why do yeast cells have three different activating kinases for Snf1? There is some evidence to suggest that the different kinases play distinct roles with respect to regulation of Snf1 function. In particular, Pak1 has a specific role in the nuclear enrichment of Snf1-Gal83 (14), which is the only form of the kinase that becomes enriched in the nucleus in response to glucose limitation (36). In a pak1Δ mutant, nuclear enrichment is defective following sudden glucose depletion and during long-term growth in glycerol-ethanol or galactose (14). This enrichment requires phosphorylation of the activation-loop Thr of Snf1, but it is not clear why Pak1 is specifically required. Previously, we excluded the possibility that each kinase is dedicated to activation of a specific form of Snf1 protein kinase; both Pak1 and Elm1 activated both Snf1-Gal83 and Snf1-Sip2, as judged by assays of Snf1 catalytic activity in vitro and by growth phenotypes (14). These findings rule out exclusivity, but the activating kinases may exhibit some preferential recognition of different Snf1 protein kinase forms.

Somewhat surprisingly, the tos3Δ mutation caused no phenotype in previous assays (14). When glucose-grown cells were subjected to acute glucose deprivation, Pak1 was the most important of the three with respect to activating Snf1 catalytic activity, and Elm1 contributed significantly, but Tos3 had little or no role. When cells expressing only Snf1-Gal83 or Snf1-Sip2 were grown on raffinose, the absence of Pak1 or Elm1 diminished growth, but Tos3 had no discernible role in cells expressing only Snf1-Gal83, Snf1-Sip2, or Snf1-Sip1. Perhaps Tos3 is simply a minor Snf1 protein kinase kinase activity in vivo, despite its robust catalytic activity towards recombinant Snf1 in vitro (16). However, it is also possible that one explanation for the existence of three Snf1-activating kinases is that the different kinases play roles under distinct growth conditions; according to this idea, Tos3 may play a more important role in regulating Snf1 under different growth conditions.

In this study, we have addressed the possibility that Tos3 has a more substantial role in activating Snf1 during growth on nonfermentable carbon sources, which is important for yeast cells in natural environments. During growth of cells on glucose or other fermentable carbon sources, the cells first exhaust the glucose, accumulating ethanol and other metabolic products. The cells then make a transition to respiratory growth during the diauxic shift and continue to double slowly until they enter stationary phase. Using several assays, we show that the tos3Δ mutation causes defects during growth on ethanol plus glycerol. We also assess the expression and subcellular localization of Tos3 under these conditions.

Strains and media. Table 1 lists S. cerevisiae strains used in this work. All strains have the W303 genetic background. To insert the tandem affinity purification (TAP) tag (27) at the 3′ end of the TOS3 open reading frame, a DNA fragment encoding the TAP tag and the Kluyveromyces lactis URA3 marker was amplified from plasmid pBS1539 (25) with primers containing appropriate TOS3 sequences. The purified PCR fragment was used to transform strain W303-1A to produce strain MCY5533. To construct MCY5510, we used a PCR fragment amplified from a strain carrying SIP2::TAP-HI3MX6 (Open Biosystems, Huntsville, AL) to transform W303-1A. Rich medium was yeast extract-peptone (YEP), and selective synthetic complete medium (SC) lacked appropriate supplements to select for plasmids (28).

TABLE 1. S. cerevisiae strains used in this study

Plasmids. To construct pMK52, which expresses Tos3 fused to green fluorescent protein (GFP) from the native promoter, we used PCR to amplify a DNA fragment containing 600 bp of 5′ sequence and the TOS3 open reading frame. The PCR fragment was inserted into the BamHI/PstI sites of the centromeric, TRP1-marked vector pMK43. pMK43 contains the yeast-enhanced GFP (yEGFP) fragment from pYGFP3 (6) and the CYC1 terminator in the pRS414 backbone and was constructed by inserting the yEGFP fragment into the XhoI/ClaI site of pMK22, which contains the CYC1 terminator in pRS414. pOV22, carrying the CSRE-lacZ reporter, has been described previously (35).

Assay of Snf1 kinase activity by phosphorylation of SAMS peptide. The procedure described previously (14) was followed. Briefly, protein extracts were prepared from yeast cells grown to mid-log phase and harvested by filtration. Cells were broken by being vortexed with glass beads in buffer A (50 mM Tris-HCl, pH 7.5, 50 mM NaF, 5 mM sodium pyrophosphate, 1 mM EDTA, 1 mM dithiothreitol, 0.1 mM phenylmethylsulfonyl fluoride, 10% [vol/vol] glycerol). After brief centrifugation, Snf1 was partially purified by DEAE-Sepharose (Amersham Biosciences) chromatography. Snf1 was eluted from the column with buffer A containing 0.2 M NaCl. Peak fractions (2 ml) were pooled and assayed for phosphorylation of the SAMS peptide (HMRSAMSGLHLVKRR) (8) in the presence of [γ-³²P]ATP (Perkin-Elmer; specific activity, ~300,000 cpm/nmol) in reaction buffer (50 mM HEPES, pH 7.5, 5 mM MgCl₂, 1 mM EDTA, 0.2 mM ATP, 10% [vol/vol] glycerol) containing 0.2 mM SAMS peptide. Kinase activity was expressed as nmol of phosphate incorporated into SAMS peptide per min per mg protein (8).

Immunoblot analysis. Protein extracts were prepared as described above. Proteins were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (8% polyacrylamide) and analyzed by immunoblotting using polyclonal anti-Snf1 (5) or peroxidase antiperoxidase (Sigma-Aldrich). Antibodies were detected by enhanced chemiluminescence with ECL Plus reagents (Amersham Biosciences).

Microscopy. Nuclei were stained by addition of 4′,6′-diamidino-2-phenylindole (DAPI) (0.8 μg/ml) for 5 min. Cells were harvested by brief centrifugation, resuspended in medium, and viewed using a Nikon Eclipse E800 fluorescent microscope. Images were taken with an Orca100 (Hamamatsu) camera, using Open Lab (Improvision) software, and processed in Adobe Photoshop 5.5.

Tos3 affects growth on nonfermentable carbon sources. The tos3Δ mutation conferred no significant defect in growth on plates containing 2% glycerol plus 3% ethanol (glycerol-ethanol) as the carbon source, regardless of whether cells expressed all three forms of Snf1 protein kinase or only Snf1-Sip1, Snf1-Sip2, or Snf1-Gal83 (14, 16) (Fig. 1A). However, comparison of the growth kinetics of tos3Δ mutant and wild-type cultures in rich medium (YEP) containing glycerol-ethanol at 30°C showed a difference in growth rate during mid-log phase. The doubling time for wild-type cultures (W303-1A) was 2.8 h, whereas that for tos3Δ mutant cultures was significantly longer, 3.5 h for MCY5120 and 3.6 h for MCY5114 (Fig. 1B and data not shown). The final cell density was also much lower for tos3Δ mutant cultures than for wild-type cultures (after 72 h of cultivation, optical density at 600 nm was 5.9 for MCY5114 and 8.1 for W303-1A). In contrast, cultures of tos3Δ mutant and wild-type cells showed identical growth kinetics in YEP-2% glucose or YEP-2% raffinose during exponential phase and through entry into stationary phase (data not shown). After incubation in YEP-2% glucose for 10 days, tos3Δ mutant and wild-type cultures showed no difference in cell viability, whereas viability of snf1Δ mutant cultures was more than 10-fold lower (data not shown). The tos3Δ and wild-type cultures also showed the same exponential growth rate in YEP-2% glucose containing 1 M sorbitol (doubling time, 1.2 h).

FIG. 1. Growth of wild-type and tos3Δ mutant cultures. (A) Strains were W303-1A (wild type, WT), MCY4908, MCY5114, MCY4100, and MCY4975. Cultures were grown in YEP-2% glucose to mid-log phase and spotted in twofold serial dilutions on YEP-2% glucose and (more ...)

Tos3 affects Snf1 catalytic activity in cells grown on glycerol-ethanol. To assess the role of Tos3 in activating Snf1 protein kinase, we assayed Snf1 catalytic activity in cells grown to exponential phase in glycerol-ethanol (Fig. 2A). For comparison, we also assayed cells grown in 2% glucose and shifted to 0.05% glucose for 30 min (Fig. 2B). Cells were in all cases collected by filtration. Snf1 protein kinase was partially purified, and catalytic activity was assayed by phosphorylation of the SAMS synthetic peptide substrate (8). Immunoblot analysis of assayed fractions was carried out to determine levels of Snf1 protein (Fig. 2C).

FIG. 2. Assays of Snf1 catalytic activity. (A and B) Cultures of wild-type (WT) and mutant strains of the indicated genotypes either were grown to mid-log phase in YEP-2% glycerol-3% ethanol (A) or were grown to mid-log phase in YEP-2% glucose and were then shifted (more ...)

During growth in glycerol-ethanol, the tos3Δ mutation reduced Snf1 catalytic activity twofold (Fig. 2A) but did not reduce levels of Snf1 protein (Fig. 2C). In the experiment shown here, the tos3Δ mutant fraction appears to contain slightly more Snf1 than the wild-type fraction (Fig. 2C), but this was not confirmed by immunoblot analysis of a different pair of fractions with a series of dilutions of the protein samples (data not shown). In contrast, Tos3 had little or no role during the response to abrupt glucose depletion (Fig. 2B), in accord with previous results (14). Pak1 was previously shown to be the most important of the three Snf1 protein kinase kinases for activating Snf1 upon glucose depletion (14) (Fig. 2B). Pak1 similarly plays an important role during growth in glycerol-ethanol; pak1Δ mutant cells showed much-reduced Snf1 catalytic activity (Fig. 2A) despite normal levels of Snf1 protein (Fig. 2C).

We then tested strains expressing only one of the three forms of Snf1 protein kinase. During growth in glycerol-ethanol, tos3Δ reduced Snf1 activity in the sip1Δ gal83Δ and sip1Δ sip2Δ mutant strains, which express only Snf1-Sip2 and Snf1-Gal83, respectively (Fig. 2A). The sip2Δ gal83Δ strain, which expresses only Snf1-Sip1, had very low activity (about 10% that of wild type; data not shown) and was not further examined. In contrast, tos3Δ had no effect during a shift to 0.05% glucose (Fig. 2B).

During growth in glycerol-ethanol, Snf1 activity in both sip1Δ sip2Δ and sip1Δ gal83Δ strains was half that of wild-type cells expressing all three forms of Snf1 protein kinase, whereas after a shift to 0.05% glucose, activity in the sip1Δ sip2Δ mutant was close to that of wild type and activity in the sip1Δ gal83Δ mutant was much reduced (compare Fig. 2A and B). These findings suggest that Snf1-Sip2 is responsible for a greater fraction of Snf1 activity, compared to Snf1-Gal83, during growth in glycerol-ethanol than after a shift to low glucose. Correspondingly, expression of Sip2, TAP tagged at its C terminus, from the genomic locus was much higher during growth in glycerol-ethanol than during growth in glucose (Fig. 2D). This finding is in accord with microarray studies showing that SIP2 RNA levels are regulated by carbon source (10) and with immunoblot analysis of GFP-tagged β subunits expressed from centromeric plasmids, which showed that Sip2 levels were higher in glycerol-grown cells than in glucose-grown cells and became comparable to those of Gal83 (36). Levels of Sip2-TAP were, however, similar in wild-type and tos3Δ mutant cells grown in either glucose or glycerol-ethanol (Fig. 2D).

Consistent with these findings that Tos3 has a role in activating Snf1 during growth in glycerol-ethanol, we detected Snf1 catalytic activity in cells expressing Tos3 alone (pak1Δ elm1Δ cells; strain MCY4970). The activity was low (0.69 ± 0.06 nmol/min/mg) but significantly higher than that found when cells were grown in 2% glucose (0.28 ± 0.04 nmol/min/mg). Another pak1Δ elm1Δ strain (MCY5121) showed similar Snf1 activity when grown in glycerol-ethanol (0.53 ± 0.06 nmol/min/mg).

Tos3 affects activation of the CSRE. To address the role of Tos3 in Snf1-dependent gene expression, we examined the effects of the tos3Δ mutation on activation of the Snf1-dependent carbon source-responsive element (CSRE) of gluconeogenic genes, which binds the activators Cat8 and Sip4 (26, 35). Expression of gluconeogenic genes is essential for growth on a nonfermentable carbon source. Snf1-Gal83 is the most important form of the kinase for activation of the CSRE (38; data not shown), so we assayed expression of a CSRE-lacZ promoter fusion in sip1Δ sip2Δ cells, which express only Snf1-Gal83. When cells were shifted from 2% glucose to glycerol-ethanol for 5.5 h, the tos3Δ mutation reduced expression of β-galactosidase (Fig. 3A); however, when cells were shifted to 0.05% glucose, the reporter was activated much less strongly, and tos3Δ had no effect (Fig. 3B). In contrast, the pak1Δ mutation reduced activation in both cases. Thus, the effects of tos3Δ on the function of a Snf1-dependent promoter element in vivo correlate well with its effects on activation of Snf1 catalytic activity.

FIG. 3. Effect of tos3Δ on CSRE-lacZ expression. Cells of the indicated genotype were transformed with pOV22, carrying the CSRE-lacZ reporter (35). Transformants were grown selectively in SC-2% glucose to mid-log phase and then were shifted to SC-2% glycerol-3% (more ...)

Expression and localization of Tos3 in glycerol-ethanol. To assess the expression of Tos3, we introduced a C-terminal TAP tag at the genomic TOS3 locus in strain MCY5533. Cultures were grown in YEP containing glucose or glycerol-ethanol, and expression of Tos3-TAP was examined by immunoblot analysis. No differences in protein levels or mobility were observed (Fig. 4A). These findings are consistent with evidence that TOS3 RNA is not strongly regulated in response to carbon source (10).

FIG. 4. Expression and localization of Tos3. (A) Strain MCY5533, expressing Tos3 tagged with the TAP epitope at its C terminus, was grown to mid-log phase in YEP-2% glucose (Glu) or YEP-2% glycerol-3% ethanol (GE). Protein extracts were prepared and subjected (more ...)

The subcellular localization of Tos3 has not been reported previously. To determine its localization, we expressed Tos3-GFP from its native promoter on a centromeric plasmid. This fusion protein was functional, as tos3Δ pak1Δ elm1Δ triple mutant cells expressing Tos3-GFP were able to grow on raffinose. The localization of Tos3-GFP was examined in wild-type cells during growth in 2% glucose, after a 30-min shift to 0.05% glucose, and during growth in glycerol-ethanol. Tos3-GFP was cytosolic and excluded from the nucleus under all conditions (Fig. 4B).

We show that, when cells are grown on glycerol-ethanol, the tos3Δ mutation reduces growth rate, Snf1 catalytic activity, and activation of the Snf1-dependent CSRE of gluconeogenic genes. This is the first demonstration of any phenotype for tos3Δ, beyond the initial finding that the tos3Δ pak1Δ elm1Δ triple mutant exhibited a Snf⁻ phenotype not found in any of the double mutants. These results indicate that Tos3 contributes to activation of Snf1 protein kinase during growth on nonfermentable carbon sources. In contrast, the tos3Δ mutation did not significantly affect Snf1 catalytic activity or CSRE function during abrupt glucose depletion, indicating that Tos3 has a much more substantial role during growth on nonfermentable carbon sources than during acute carbon stress. These findings lend support to the idea that the three Snf1 protein kinase kinases make different contributions to cellular regulation under different growth conditions.

Nonetheless, the defects that we have detected are modest, and even during growth on glycerol-ethanol, Pak1 affects Snf1 activity more profoundly than does Tos3. The abundance of Tos3 was not estimated in the yeast GFP fusion localization database (11), because Tos3-GFP was not detected, and it is possible that Tos3 is simply less abundant than Pak1. We also consider it possible, however, that the primary function of Tos3 involves activation of Snf1 protein kinase in response to some other environmental condition.

In addition, Tos3 may have some role in the cell that is unrelated to Snf1. For example, Tos3 may phosphorylate other protein kinases or other components of a different regulatory pathway. Another of the Snf1-activating kinases, Elm1, has roles in cell cycle progression, cell morphology, and filamentous invasive growth (2, 3, 7, 30, 31) that are apparently unrelated to Snf1. We also note that a mammalian ortholog of Tos3, the LKB1 tumor suppressor kinase, activates not only AMPK (13, 16, 40) but also a dozen other AMPK-like protein kinases (21). Thus, the existence of three Snf1 protein kinase kinases may reflect diverse cellular functions that go far beyond the activation of Snf1.

We thank K. Hedbacker for assistance with microscopy and H. Wiatrowski and K. Hedbacker for strains.

This work was supported by NIH grant GM34095 to M.C. and in part by a KOSEF fellowship to M.-D.K.