Pak1 is the major kinase required for the activity of Snf1-Gal83 kinase in vitro. Gal83 is the most abundant β subunit in glucose-grown cells (9
). Assays of mutants expressing only one β subunit showed that Snf1-Gal83 kinase is, correspondingly, responsible for the majority of Snf1 kinase activity (Fig. ). Cells were grown in glucose, collected by centrifugation, and shifted to 0.05% glucose for 30 min; both centrifugation and carbon deprivation are stresses that activate Snf1 (41
). The kinase was partially purified and catalytic activity was assayed by phosphorylation of the SAMS synthetic peptide substrate (5
). In the sip1
Δ mutant, which expresses only Snf1-Gal83, the level of activity was 75% of wild-type levels. Conversely, in sip2
Δ and sip1
Δ strains, which express only Snf1-Sip1 and Snf1-Sip2, respectively, activity was reduced to 10% of that of the wild type. Part of this decrease can be attributed to the somewhat lower levels of Snf1 protein in the absence of the major β subunit, Gal83 (Fig. ).
FIG. 1. Assay of Snf1 kinase activity. Cells of the indicated genotype were grown in YEP-2% glucose, collected by centrifugation (A) or by filtration (B to D), and shifted to YEP-0.05% glucose for 30 min. Cells were then collected by filtration, extracts were (more ...)
Although deletion of all three upstream kinase genes is necessary to abolish Snf1 kinase activity, mutation of PAK1
causes the greatest decrease (13
) (see Fig. ). To assess the role of Pak1 in the activation of Snf1-Gal83, we assayed the pak1
Δ mutant for in vitro kinase activity (Fig. ). In the triple mutant, activity was reduced 12-fold relative to that of the sip1
Δ mutant, indicating that Pak1 has a major role in the activation of Snf1-Gal83. The pak1
Δ mutation also appeared to cause some reduction of Snf1-Sip1 and Snf1-Sip2 activities, as judged by a comparison of the corresponding double and triple mutants (Fig. ; see the legend for values). The absence of Pak1, or of the other upstream kinases, did not affect levels of Snf1 protein (Fig. ), in accord with previous results (13
Snf1-Gal83 kinase is not exclusively activated by Pak1.
To determine whether Snf1-Gal83 kinase is exclusively activated by Pak1, we assayed Snf1-Gal83 activity in mutants lacking each of the three upstream kinases. Cells were grown in 2% glucose, collected by filtration, and shifted to 0.05% glucose for 30 min (Fig. ). The pak1Δ sip1Δ sip2Δ mutant again showed much lower activity than the sip1Δ sip2Δ double mutant, and the elm1Δ sip1Δ sip2Δ mutant showed a twofold reduction in activity. The activity level of the tos3Δ sip1Δ sip2Δ mutant was not significantly different from that of the sip1Δ sip2Δ mutant, although it remains possible that Tos3 has a minor effect. We conclude that Pak1 is the most important of the three upstream kinases for activating Snf1-Gal83 but that Elm1 also has a role.
We also compared the three upstream kinases with respect to their activation of Snf1-Sip2 and Snf1-Sip1. In this assay, when cells were collected by filtration, Snf1-Sip2 activity was reduced more than threefold by the pak1Δ mutation and twofold by elm1Δ (Fig. ). The activity of Snf1-Sip1 was very low in this assay, so the apparent effects of Pak1 and Elm1 are not compelling (Fig. ).
Pak1 affects Snf1-Gal83 function in vivo.
The in vitro assays suggest that Pak1 has a major physiological role in activating Snf1-Gal83 kinase. To assess the role of Pak1 in vivo, we compared sip1
Δ and pak1
Δ cells with respect to several phenotypes. First, we tested growth on different carbon sources. The absence of Pak1 diminished growth on raffinose (Fig. ) but did not affect growth on glycerol-ethanol (data not shown). We noted no effect of a tos3
deletion and a small effect of an elm1
deletion, as judged by comparisons to the glucose control plate (elm1
Δ strains are clumpy and spot differently from the others). The absence of Pak1 did not affect growth on raffinose when all β subunits were present (references 13
and data not shown).
FIG. 2. Growth phenotypes and glycogen accumulation. Cells of the indicated genotype were grown overnight in YEP-2% glucose. (A) Cells were adjusted to an OD600 of 0.5 and spotted with serial dilutions (one 10-fold dilution followed by successive 5-fold dilutions) (more ...)
We next assayed glycogen accumulation, which requires Snf1 (7
). Cells grown on solid medium containing 2% glucose were exposed to iodine vapor; cells containing glycogen stain dark (Fig. ). A comparison of sip1
Δ and pak1
Δ strains showed that Pak1 is required for Snf1-Gal83 function in glycogen accumulation. A very modest effect for Elm1 was also apparent upon direct examination of the plate.
Activation of the CSRE of gluconeogenic genes, which binds the activators Cat8 and Sip4 (26
), depends on Snf1-Gal83 (40
). Cells carrying a CSRE-lacZ
reporter were induced by a shift from 2 to 0.05% glucose for 5.5 h. β-Galactosidase activity was undetectable prior to the shift. After induction, activity was modestly lower in pak1
Δ cells than in sip1
Δ cells (4 ± 1 and 13 ± 1 U, respectively; values are averages for seven transformants). Activity was <1 U in sip1
Δ cells and was slightly lower in wild-type cells (10 U) than in sip1 sip2
cells, where all the kinase is in the Snf1-Gal83 form that activates the CSRE.
Finally, we used an assay in which LexA-tagged Snf1G53R, a catalytically hyperactive mutant kinase, activates transcription of a lacZ
reporter with LexA binding sites in its promoter (17
). This activation requires Gal83 (38
) and most likely reflects the interaction of the kinase with the transcriptional apparatus (17
). We expressed LexA-Snf1G53R in strain CTY10-5d, which carries the lacZ
reporter, and in pak1
Δ and gal83
Δ derivatives of CTY10-5d. Transformants were grown in 2% glucose, shifted to 0.05% glucose for 3 h, and assayed for β-galactosidase activity. Activation of the reporter was reduced 35-fold in pak1
Δ mutant cells relative to that in the wild type (2.6 ± 0.1 and 92 ± 3 U, respectively; values are averages for three transformants). Activation was virtually abolished in the gal83
Δ mutant (0.4 ± 0.1 U); for all strains, activity was negligible in glucose-grown cells (pak1
Δ mutant, 0.3 U; wild type, 1.0 U; gal83
Δ mutant, 0.1 U). Thus, Pak1 was important for the function of Snf1G53R-Gal83 in this assay.
Pak1 affects Snf1-Sip2 function in vivo.
Assays of Snf1 catalytic activity in vitro suggested that Pak1 contributes to the activation of Snf1-Sip2 and, possibly, Snf1-Sip1 (Fig. ). To confirm that Pak1 affects other forms of the kinase besides Snf1-Gal83, we assayed growth phenotypes. In cells expressing only Snf1-Sip2, the pak1Δ mutation impaired growth on raffinose (Fig. ; compare sip1Δ gal83Δ and pak1Δ sip1Δ gal83Δ cells). The defect is more severe than for pak1Δ sip1Δ sip2Δ cells, suggesting that low levels of Snf1-Sip2 activity are less effective in providing functions required for growth on raffinose than low levels of Snf1-Gal83 activity. In cells expressing Snf1-Sip1, no effect of pak1Δ was evident (Fig. ), but this negative result does not exclude the possibility of effects on other phenotypes not assayed here. These findings, together with the results of the kinase assays, indicate that Pak1 is not dedicated to the activation of Snf1-Gal83 but rather also regulates the activity of at least one other form of Snf1 protein kinase.
FIG. 3. Growth phenotypes. Cells of the indicated genotype expressing only Snf1-Sip2 (A) or Snf1-Sip1 (B) were grown overnight in YEP-2% glucose. Cells were adjusted to an OD600 of 0.25, except for strains carrying elm1, which were adjusted to an OD600 of 2.5, (more ...)
We also examined the effects of the other two upstream kinases. In cells expressing only Snf1-Sip2 or Snf1-Sip1, the elm1Δ mutation also appeared to impair growth on raffinose, whereas a deletion of tos3 had no effect (Fig. ). These results are consistent with a role for Elm1 in activating these kinase forms.
Upstream kinases are required for nuclear enrichment of Gal83 in response to carbon stress.
The β subunits regulate the subcellular localization of Snf1 protein kinase in response to the carbon source, and localization presumably affects the access of the kinase to particular sets of substrates. All β subunits are cytoplasmic during growth in 2% glucose, but upon a shift to glucose-limiting conditions, Gal83 rapidly becomes enriched in the nucleus, Sip1 relocalizes around the vacuole, and Sip2 remains cytoplasmic (38
). Protein kinase A negatively regulates the localization of Sip1 to the vacuolar membrane (12
) but does not affect the localization of Gal83 (38
). The evidence has suggested glucose-6-phosphate as a candidate signal for the nuclear exclusion of Gal83 (38
), but the pathway controlling the localization of Gal83 has not been identified.
To assess the role of the upstream kinases, we expressed functional, GFP-tagged Gal83 from its native promoter on a centromeric plasmid in wild-type and pak1Δ tos3Δ elm1Δ cells. During growth in glucose, Gal83-GFP was cytoplasmic and excluded from nuclei in both wild-type and mutant cells (Fig. ). Upon a shift to glycerol-ethanol, Gal83-GFP became enriched in the nuclei of wild-type cells. Figure shows cells photographed 30 min after the shift, but enrichment was evident within 5 min (data not shown). In contrast, Gal83-GFP remained excluded from the nuclei of triple mutant cells at 30 min (Fig. ). Thus, the triple mutant exhibited a striking defect in the nuclear enrichment of Gal83-GFP.
FIG. 4. Nuclear localization of Gal83-GFP in response to carbon stress depends on upstream kinases. Cells of the indicated genotype expressed Gal83-GFP. Strains used were W303-1A (wild type [WT]) (A) and MCY5123 (B). Cells were grown to mid-log phase in SC medium-2% (more ...) The nuclear localization of Gal83 is defective in pak1Δ mutants.
To identify the upstream kinase(s) responsible for regulating the localization of Gal83, we examined Gal83-GFP in each of the single mutants. Nuclear enrichment was observed in tos3Δ and elm1Δ mutants after a shift to glycerol-ethanol (data not shown), but the pak1Δ mutant was clearly defective (Fig. ), as were the tos3Δ pak1Δ and elm1Δ pak1Δ mutants (data not shown). Expression of HA-tagged Pak1 from a plasmid complemented the pak1Δ mutant defect, thereby confirming that the mutation is responsible for the phenotype (Fig. ). Conversely, nuclear localization occurred normally in the tos3Δ elm1Δ double mutant, indicating that Pak1 is sufficient for localization (Fig. ). To determine whether Pak1 is required for the nuclear enrichment of Gal83 under different conditions, we shifted glucose-grown cells to low (0.05%) glucose for 30 min or to glycerol-ethanol or galactose for extended growth. In all cases, the pak1Δ mutant failed to exhibit a nuclear enrichment of Gal83-GFP (Fig. and data not shown).
FIG. 5. Pak1 is required for the nuclear localization of Gal83-GFP. Cells of the indicated genotype expressed Gal83-GFP. Strains used were MCY5115 (A and D), MCY5115 expressing HA-Pak1 from pRH104 (B), MCY5122 (C), MCY5115 expressing HA-Pak1D277A from pRH119 (more ...)
To determine whether the catalytic activity of Pak1 is required for nuclear enrichment, we introduced mutations into the HA-Pak1 expression plasmid to replace Asp277 with Ala, which inactivates Pak1 with respect to the phosphorylation of Snf1 on Thr210 (25
), and to change the highly conserved Asp295 to Ala. The expression of the wild-type HA-Pak1 served as a control (Fig. ). Gal83-GFP did not become enriched in the nuclei of pak1
Δ mutant cells expressing the kinase-dead HA-Pak1D277A or HA-Pak1D295A upon shifts to glycerol-ethanol or low glucose (Fig. ). Thus, the catalytic activity of Pak1 is required for the nuclear enrichment of Gal83.
Evidence for a minor role for Tos3 in the localization of Gal83.
Although Tos3 and Elm1 do not have primary roles in the nuclear enrichment of Gal83, evidence suggests that in the absence of Pak1 they play minor roles under some conditions. In the pak1Δ mutant, Gal83-GFP was clearly excluded from the nucleus during growth in glycerol-ethanol or galactose, whereas after a shift from 2% glucose to 0.05% glucose or glycerol-ethanol, nuclear exclusion was less easily apparent (Fig. ). In contrast, exclusion was clearly evident in the pak1Δ tos3Δ elm1Δ mutant after a shift (Fig. ). Moreover, overexpression of kinase-dead HA-Pak1 enhanced nuclear exclusion in the pak1Δ mutant (Fig. ). These observations suggested that during acute carbon stress, Tos3 and/or Elm1 partially compensates for the absence of Pak1 and promotes the presence of a very low level of Gal83 in the nucleus. To test this possibility, we overexpressed Tos3 (LexA tagged) from the ADH1 promoter in a pak1Δ mutant. We detected weak nuclear enrichment in some cells after a shift to glycerol-ethanol, indicating that overexpression of Tos3 partially suppressed the pak1Δ defect (Fig. ). A similar experiment with Elm1 gave negative, and hence inconclusive, results. Together, these data suggest a minor degree of overlap of the functions of Tos3 and Pak1 with respect to localization, which is evident in the absence of Pak1.
Pak1 is not localized in the nucleus in response to carbon stress.
One possible explanation for the profound effects of Pak1 on nuclear localization is that Pak1 itself is enriched in the nucleus and is responsible for the colocalization of Snf1-Gal83. To test this idea, we constructed a gene fusion expressing Pak1-GFP from the genomic locus. Glucose-grown cells showed low-level cytosolic fluorescence (Fig. ). After a shift to limiting glucose, many cells exhibited enhanced fluorescence around the vacuolar membrane, similar to that of Sip1-GFP under these conditions (12
); however, we detected no nuclear enrichment of Pak1-GFP (Fig. ).
FIG. 6. Localization of Pak1-GFP. Cells of strain MCY4999, expressing Pak1-GFP from the genomic locus, were grown to mid-log phase in SC medium-2% glucose (Glu) and were shifted to 0.05% glucose (Low Glu) for 30 min. DIC, GFP fluorescence, and DAPI staining are (more ...) The nuclear localization of Gal83 does not depend on Snf1.
A simple model is that Pak1 causes the nuclear localization of Gal83, which is present in Snf1-Gal83 kinase complexes, by activating Snf1. However, previous studies of snf1
Δ mutant cells of the S288C background showed that nuclear enrichment of Gal83-GFP does not require Snf1 (38
). Because differences in genetic backgrounds can have profound effects, we reexamined this issue with a snf1
Δ mutant of the W303 background used in this study. Gal83-GFP was excluded from nuclei in glucose-grown snf1
Δ cells and became enriched in nuclei after a shift to glycerol-ethanol (Fig. ). The fraction of cells showing strong enrichment was somewhat lower than for the wild type, which may simply reflect the general unhealthiness of snf1
Δ cells or may reflect a contributing role of Snf1. Nonetheless, Snf1 is clearly not required, as cells exhibited regulated nuclear enrichment of Gal83 in the absence of Snf1.
FIG. 7. Nuclear localization of Gal83-GFP in snf1Δ and pak1Δ snf1Δ mutant cells. Cells expressed Gal83-GFP. Strains used were MCY4908 (A) and MCY5025 (B). Cells were grown to mid-log phase in SC medium-2% glucose (Glu) and then were shifted (more ...) In the absence of Snf1, the nuclear localization of Gal83 does not depend on Pak1.
The findings described above suggested that Pak1 regulates Gal83 directly; alternatively, it remains possible that in the absence of Snf1, the nuclear localization of Gal83 does not depend on Pak1. We indeed found that in pak1Δ snf1Δ double mutant cells, Gal83-GFP became enriched in nuclei upon a shift to low glucose (Fig. ). Enrichment was not as pronounced as in the wild type but was not noticeably different from that observed in the snf1Δ mutant. Thus, in the absence of Snf1, the nuclear localization of Gal83 is regulated by a mechanism that is independent of Pak1.
Both Pak1 and Gal83 are required for the nuclear enrichment of Snf1.
To examine the localization of Snf1, we used functional, GFP-tagged Snf1 expressed from its native promoter on a centromeric plasmid (38
). In the S288C background, Snf1-GFP was enriched in the nucleus after a shift to glycerol or during growth in glycerol-ethanol, and enrichment was strongly reduced in gal83
Δ mutant cells (38
). In the W303 background, the nuclear enrichment of Snf1-GFP was similarly evident in wild-type cells after a shift to a poor carbon source (Fig. ); enrichment was less striking than for Gal83-GFP because some Snf1 is complexed with Sip1 and Sip2, which do not relocalize to the nucleus (38
). No nuclear enrichment occurred in gal83
Δ cells after a shift to low glucose or glycerol-ethanol (Fig. and data not shown). We also observed no nuclear enrichment of Snf1-GFP in pak1
Δ, or pak1
Δ cells (Fig. and data not shown). Thus, both Pak1 and Gal83 are required for the nuclear enrichment of Snf1.
FIG. 8. Localization of Snf1-GFP and Snf1T210A-GFP. Cells expressed Snf1-GFP (A to C) and Snf1T210A-GFP (D). Strains used were W303-1A (A and D), MCY4093 (B), and MCY5115 (C). Cells were grown in SC medium-2% glucose (Glu) and were shifted to 2% glycerol-3% ethanol (more ...) Mutation of Thr210 of Snf1 abolishes the nuclear localization of Snf1-Gal83.
A parsimonious model is that Pak1 controls the localization of Snf1-Gal83 through phosphorylation of the activation loop threonine (Thr210) of Snf1. This model predicts that the nuclear enrichment of Gal83 would be impaired in the presence of Snf1T210A, which has Thr210 replaced with Ala (6
). To maximize the association of Gal83 with Snf1T210A, we expressed Snf1T210A from the ADH1
promoter in snf1
Δ cells; the native level of Gal83 has been reported to be higher than that of Snf1 (9
). Expression of Snf1T210A (HA or LexA tagged) prevented the nuclear enrichment of Gal83-GFP, whereas expression of HA-Snf1 had no effect (data not shown). As a more direct test of the model, we expressed Snf1T210A-GFP from its native promoter in both wild-type and snf1
Δ cells. Snf1T210A-GFP remained excluded from the cytosol and nucleus after a shift from high to low glucose (Fig. and data not shown). Hence, the nuclear enrichment of Snf1 requires the phosphorylation of Thr210.
Phosphorylation of Thr210 causes conformational changes and activates Snf1. To assess the requirement for Snf1 catalytic activity, we expressed kinase-dead Snf1K84R-GFP, in which Arg replaces the Lys of the ATP-binding site (3
). Snf1K84R-GFP did not become enriched in the nucleus after a shift to low glucose (data not shown). In previous experiments, enrichment of overexpressed Gal83-GFP in snf1-K84R
cells probably reflected Gal83-GFP that was not complexed with Snf1K84R (38
). Together, these findings strongly suggest that the nuclear enrichment of Snf1-Gal83 requires the phosphorylation and activation of Snf1 by Pak1.