Function of mammalian LKB1 and Ca2+/calmodulin-dependent protein kinase kinase alpha as Snf1-activating kinases in yeast.

The Snf1/AMP-activated protein kinase (AMPK) family is important for metabolic regulation in response to stress. In the yeast Saccharomyces cerevisiae, the Snf1 kinase cascade comprises three Snf1-activating kinases, Pak1, Tos3, and Elm1. The only established mammalian AMPK kinase is LKB1. We show that LKB1 functions heterologously in yeast. In pak1Delta tos3Delta elm1Delta cells, LKB1 activated Snf1 catalytic activity and conferred a Snf(+) growth phenotype. Coexpression of STRADalpha and MO25alpha, which form a complex with LKB1, enhanced LKB1 function. Thus, the Snf1/AMPK kinase cascade is functionally conserved between yeast and mammals. Ca(2+)/calmodulin-dependent kinase kinase (CaMKK) shows more sequence similarity to Pak1, Tos3, and Elm1 than does LKB1. When expressed in pak1Delta tos3Delta elm1Delta cells, CaMKKalpha activated Snf1 catalytic activity, restored the Snf(+) phenotype, and also phosphorylated the activation loop threonine of Snf1 in vitro. These findings indicate that CaMKKalpha is a functional member of the Snf1/AMPK kinase family and support CaMKKalpha as a likely candidate for an AMPK kinase in mammalian cells. Analysis of the function of these heterologous kinases in yeast provided insight into the regulation of Snf1. When activated by LKB1 or CaMKKalpha, Snf1 activity was significantly inhibited by glucose, suggesting that a mechanism independent of the activating kinases can mediate glucose signaling in yeast. Finally, this analysis provided evidence that Pak1 functions in another capacity, besides activating Snf1, to regulate the nuclear enrichment of Snf1 protein kinase in response to carbon stress.

The Snf1/AMP-activated protein kinase (AMPK) 1 family is widely conserved in eukaryotes. In mammalian cells, AMPK is a major regulator of glucose and lipid metabolism and serves to maintain the cellular energy balance (reviewed in Refs. [1][2][3][4]. AMPK is activated in response to reduced energy availability (high cellular AMP:ATP ratios) and by hormones, including leptin (5) and adiponectin (6). Once activated, it inhibits ATPconsuming processes, stimulates ATP-generating processes, and regulates transcription. AMPK has been implicated in type 2 diabetes, obesity, and aspects of the metabolic syndrome (1,7,8). AMPK is the target of pharmacological agents that are used to treat diabetes (8,9), and recent evidence indicates that it functions in the hypothalamus to regulate food intake in response to hormonal and nutrient signals (10,11). In addition, mutations in the AMPK ␥2 subunit have been linked to the cardiac diseases hypertrophic cardiomyopathy and Wolff-Parkinson-White syndrome (12,13).
Snf1 protein kinase (14) is the ortholog of AMPK in the yeast Saccharomyces cerevisiae (15,16). Snf1 protein kinase is heterotrimeric, as is AMPK, comprising the Snf1 catalytic subunit, the regulatory subunit Snf4, and one of three alternate ␤ subunits. Mutation of the SNF1 gene causes the Snf Ϫ (for sucrosenonfermenting) phenotype, which is characterized by inability to utilize carbon sources that are less preferred than glucose. Like AMPK, Snf1 protein kinase regulates gene expression and affects the activity of metabolic enzymes and transporters in response to metabolic stress, particularly carbon stress; the glucose signal or signals regulating Snf1 are not known, but AMP does not appear to have a role (reviewed in Refs. 2 and 17).
We and others have identified three upstream kinases, Pak1, Tos3, and Elm1, that phosphorylate the activation loop threonine Thr-210 of the Snf1 catalytic subunit and thereby activate Snf1 protein kinase (18 -20) (Fig. 1). These three kinases have overlapping functions in vivo, and all three genes must be deleted to abolish Snf1 catalytic function and cause a Snf Ϫ phenotype (18,19). Elm1 also has roles in controlling cell morphology, cell cycle progression, and filamentous invasive growth (21)(22)(23)(24)(25) that are not known to be related to Snf1 function.
These yeast kinases led to the identification of a mammalian homolog, the LKB1 tumor suppressor kinase, as an AMPK kinase (18) (Fig. 1). Snf1 and AMPK are highly conserved, and Tos3 and Elm1 activated AMPK in vitro (18,19). Their closest mammalian homolog, Ca 2ϩ /calmodulin-dependent kinase kinase (CaMKK), was known to activate AMPK in vitro but did not correspond to the major AMPK kinase activity detected in the liver (26). We therefore focused on LKB1 and showed that LKB1 phosphorylates Thr-172 in the activation loop of the catalytic subunit and activates AMPK in vitro (18). Subsequent work verified that LKB1 is an AMPK kinase in vivo (27)(28)(29). Mutations in the human LKB1 gene lead to the Peutz-Jeghers familial cancer syndrome (30,31), and knock-out mutations in mice block embryonic development at mid-gestation (32). However, the relationship of AMPK to these phenotypes is not clear, as recent work shows that LKB1 activates 12 AMPK-related kinases (33).
AMPK kinases are potentially important targets in the AMPK pathway for therapeutic intervention. Here we have used the yeast system, which offers the advantages of powerful genetics, for functional analysis of the Snf1/AMPK kinase family. We first tested LKB1 for heterologous function in yeast lacking the native Snf1-activating kinases and show that LKB1 activates Snf1 catalytic activity and confers a Snf ϩ growth phenotype. These findings indicate that the Snf1/AMPK kinase cascade is conserved between yeast and mammals. We therefore used the yeast system to test the function of CaMKK as an activating kinase in vivo. There are two isoforms of CaMKK, ␣ and ␤ (34,35). We show that rat brain CaMKK␣ (36) activates Snf1 protein kinase in pak1⌬ tos3⌬ elm1⌬ mutant yeast, indicating that CaMKK␣ is a functional member of the Snf1/AMPK kinase family. These findings support CaMKK␣ as a likely candidate for an AMPK kinase in mammalian cells. Finally, we took advantage of the heterologous function of LKB1 and CaMKK␣ to probe mechanisms regulating the catalytic activity and the subcellular localization of Snf1 protein kinase in yeast. We present evidence that a mechanism independent of the activating kinases mediates glucose signaling and that Pak1 functions in another capacity, besides activating Snf1, to regulate the nuclear enrichment of Snf1 protein kinase in response to carbon stress. The potential use of the yeast system to identify additional mammalian AMPK kinases by genetic selection is discussed.
Assay of Snf1 Kinase Activity by Phosphorylation of SAMS Peptide-Cells were grown in SC with 2% glucose to an A 600 of 1, collected by filtration, incubated in SC with 0.05% glucose for 30 min, and collected by filtration. Extracts were prepared from two independent cultures, and assays for phosphorylation of a synthetic peptide (HMRSAMS-GLHLVKRR; SAMS peptide) (41) were performed as described (15,38) using different protein concentrations to confirm linearity. Control reactions lacked the SAMS peptide. Kinase activity is expressed as nanomoles of phosphate incorporated into the peptide per minute per milligram of protein (41).
Immunoblot Analysis-Extracts for immunoblot analysis were prepared by boiling cells and vortexing with glass beads (43). Proteins were separated by SDS-PAGE and immunoblotted with anti-Snf1 (14), anti-LexA (Invitrogen), or monoclonal anti-HA antibody (12CA5). Antibody was detected by chemiluminescence using ECL Plus (Amersham Biosciences).

LKB1 Restores Growth of the pak1⌬ tos3⌬ elm1⌬ Mutant on
Raffinose-The pak1⌬ tos3⌬ elm1⌬ triple mutant lacks the native Snf1-activating kinases and therefore fails to grow on raffinose or glycerol-ethanol, the utilization of which requires Snf1 activity (18). To test the ability of LKB1 to function in yeast, we expressed HA-tagged LKB1 in pak1⌬ tos3⌬ elm1⌬ cells. LKB1 restored slow growth on raffinose, whereas the kinase-dead mutant LKB1D194A did not, indicating dependence on the catalytic activity ( Fig. 2A). Expression of LKB1 and LKB1D194A was confirmed by immunoblot analysis (data not shown), and control transformants expressing Pak1 and kinase-dead Pak1D295A (38) showed the expected growth phenotypes. To exclude the possibility that LKB1 simply bypasses FIG. 1. Snf1/AMPK kinase cascades. In yeast cells, three homologous protein kinases, Pak1, Tos3, and Elm1, phosphorylate and activate Snf1 protein kinase. CaMKK and LKB1 are the mammalian kinases that are most similar to Pak1, Tos3, and Elm1. In mammalian cells, LKB1 phosphorylates and activates AMPK. The possibility that CaMKK also activates AMPK in vivo is indicated by a dashed arrow and a question mark.

FIG. 2. Growth of yeast mutants expressing LKB1 or CaMKK␣.
A, HA-tagged proteins or HA alone (vector) were expressed in pak1⌬ tos3⌬ elm1⌬ cells. Cells were spotted with serial 5-fold dilutions on SC medium containing 2% glucose, 2% raffinose plus the respiratory inhibitor antimycin (1 g/ml), or 2% glycerol and 3% ethanol. Plates were incubated at 30°C and photographed after 4 days for glucose, 7 days for raffinose, and 5 days for glycerol-ethanol. B, HA-tagged wild-type (WT) or kinase-dead (D194A) LKB1 was coexpressed with LexA-tagged STRAD␣ and MO25␣ in pak1⌬ tos3⌬ elm1⌬ cells. Negative sign (Ϫ) denotes that HA or LexA was expressed from the vector. Cells were spotted as in panel A. Plates shown in the lower sections were incubated for additional days. C, snf1⌬ cells expressing the indicated proteins were tested for growth as described above. D, extracts were prepared as described for the assay of Snf1 activity from pak1⌬ tos3⌬ elm1⌬ cells expressing the indicated proteins. Immunoblot analysis was performed with anti-HA and anti-Snf1 antibodies.
the requirement for Snf1, we expressed LKB1 in the snf1⌬ mutant. LKB1 did not restore growth (data not shown; see Fig.  2C), indicating that complementation of the triple mutant defect depends on the presence of Snf1.
Coexpression of STRAD␣ and MO25␣ Improves LKB1 Function in Yeast-In mammalian cells LKB1 functions in a complex with two other proteins, the pseudokinase STRAD and MO25 (29,44,45). To determine whether these proteins similarly contribute to the function of LKB1 as a Snf1 protein kinase kinase, we coexpressed LexA-tagged STRAD␣ and/or MO25␣ with LKB1 in the pak1⌬ tos3⌬ elm1⌬ mutant; expression was confirmed by immunoblot analysis (data not shown). Coexpression of STRAD␣ with LKB1 improved growth on raffinose, and coexpression of both STRAD␣ and MO25␣ with LKB1 allowed growth on glycerol-ethanol as well as raffinose (Fig. 2B). Growth was dependent on Snf1 and did not occur in the snf1⌬ mutant (Fig. 2C). The enhancement of growth by STRAD␣ and MO25␣ depended on LKB1 catalytic activity, as no growth was observed with kinase-dead LKB1D194A (Fig.  2B). Immunoblot analysis showed that coexpression of STRAD␣ and MO25␣ resulted in higher levels of LKB1, suggesting that formation of the LKB1-STRAD␣-MO25␣ complex stabilizes LKB1 in yeast (Fig. 2D). Although it remains formally possible that LKB1 functions alone rather than as part of this complex, these results suggest that the ability of the LKB1-STRAD␣-MO25␣ complex to function in the Snf1/AMPK kinase cascade has been conserved from mammals to yeast.
LKB1 Activates Snf1 Protein Kinase in Vivo-To demonstrate biochemically that LKB1 functions in yeast to activate Snf1 protein kinase, we assayed Snf1 catalytic activity by phosphorylation of the SAMS peptide (15,41). Triple mutant cells expressing the mammalian kinase were subjected to glucose limitation to activate Snf1 protein kinase. Snf1 was then partially purified from cell extracts and incubated with SAMS peptide in the presence of [␥-32 P]ATP. This assay is specific for Snf1 activity in yeast, and no activity is detected in the snf1⌬ mutant (15,18). Expression of LKB1 in pak1⌬ tos3⌬ elm1⌬ cells elevated Snf1 activity relative to the vector and LKB1D194A controls (Fig. 3A). The modest effect on catalytic activity was consistent with the modest effect on growth. Moreover, triple mutant cells coexpressing LKB1, STRAD␣, and MO25␣ showed higher Snf1 catalytic activity than cells coexpressing LKB1 and the two respective vectors (Fig. 3B), as expected from their growth phenotypes (Fig. 2B). Immunoblot analysis confirmed the presence of equivalent amounts of Snf1 protein in the assayed fractions (Fig. 3C).
LKB1 Phosphorylates Snf1 on the Activation Loop Threonine in Vitro-LKB1 activates AMPK by phosphorylation of the conserved Thr-172 in the activation loop region (18). To determine whether LKB1 phosphorylates the analogous residue of Snf1, Thr-210, we used as substrates two recombinant inactive forms of the Snf1 catalytic domain, GST-Snf1KD-T210A and GST-Snf1KD-K84R (Lys-84 is the conserved residue of the ATP-binding site) (Fig. 4). HA-LKB1 was immunoprecipitated from extracts of pak1⌬ tos3⌬ elm1⌬ cells that coexpressed STRAD␣ and MO25␣, and the immunoprecipitates were incubated with substrate in the presence of [␥-32 P]ATP. LKB1 phosphorylated GST-Snf1KD-K84R strongly and phosphorylated GST-Snf1KD-T210A much more weakly, indicating that Thr-210 is the major site of phosphorylation (Fig. 4). Coomassie Blue staining of the gel confirmed that the amounts of the two substrates were the same (Fig. 4). In control experiments, HA-Tos3 phosphorylated Snf1KD-K84R but not Snf1KD-T210A, as expected (18), and immunoprecipitates from cells expressing HA alone showed no phosphorylation of either substrate (data not shown). Together, these results suggest that LKB1 functions in yeast to phosphorylate Thr-210 and activate Snf1 protein kinase.
CaMKK␣ Restores Growth of the pak1⌬ tos3⌬ elm1⌬ Mutant on Raffinose and Glycerol-Ethanol-CaMKK, like LKB1, is homologous to Pak1, Tos3, and Elm1 and, in fact, shows greater sequence similarity. Hawley et al. (26) showed many years ago that a preparation of CaMKK from pig brain activated AMPK in vitro but that CaMKK did not correspond to the major AMPK kinase in rat liver, which is now known to be LKB1 (27,29). We reasoned that mammals, like yeast, may have multiple AMPK kinases, and CaMKK seemed a likely FIG. 3. Assays of Snf1 catalytic activity. Proteins were expressed in pak1⌬ tos3⌬ elm1⌬ cells. Negative sign (Ϫ) denotes that HA or LexA was expressed from the vector. Cells were grown to mid-log phase in SC plus 2% glucose, collected by filtration, resuspended in SC plus 0.05% glucose for 30 min, and collected by filtration. Extracts were prepared, and Snf1 was partially purified. A and B, Snf1 was assayed for ability to phosphorylate the SAMS peptide. Extracts were prepared from two independent transformants, and values are averages of 4 -6 assays. C, assayed fractions were analyzed by immunoblotting with anti-Snf1. Lane numbers correspond to the assays numbered in panels A and B. candidate to be a member of the Snf1/AMPK kinase family. We therefore used yeast as a convenient model system to test its function as an activating kinase in vivo. Using a cloned rat brain cDNA encoding the CaMKK␣ isoform (36), we expressed HA-tagged CaMKK␣ in the pak1⌬ tos3⌬ elm1⌬ mutant. We first tested its ability to restore the Snf ϩ growth phenotype. CaMKK␣ very effectively restored growth on both raffinose and glycerol-ethanol ( Fig. 2A). The kinase-dead mutant CaMKK␣D293A was expressed (data not shown; see Fig. 4C) but did not confer growth ( Fig. 2A), indicating that catalytic activity is required. Restoration of growth was dependent on the presence of Snf1, as expression of CaMKK␣ had no effect in the snf1⌬ mutant (Fig. 2C). Immunoblot analysis showed that HA-CaMKK␣ was expressed at substantially higher levels than HA-LKB1 (data not shown), which may contribute to its effective function in yeast.
CaMKK␣ Activates Snf1 Protein Kinase in Vivo and Phosphorylates Snf1 on Thr-210 in Vitro-To show that CaMKK␣ activates Snf1 protein kinase in yeast cells, we assayed Snf1 catalytic activity by phosphorylation of the SAMS peptide, as described above. Expression of CaMKK␣ conferred significant catalytic activity to Snf1 protein kinase (Fig. 3A), roughly half of that found in triple mutant cells expressing HA-Pak1 or in wild-type cells (see Fig. 5A). In control experiments, kinasedead CaMKK␣D293A and Pak1D295A did not activate Snf1. Immunoblot analysis confirmed the presence of equivalent amounts of Snf1 protein in the assayed fractions (Fig. 3C). These findings indicate that CaMKK␣ functions as a Snf1activating kinase in yeast.
Previous studies showed that pig brain CaMKK activated AMPK in vitro but did not determine the site of phosphorylation (26). We tested the ability of CaMKK␣ to phosphorylate Thr-210 of Snf1. HA-CaMKK␣ was immunoprecipitated from extracts of pak1⌬ tos3⌬ elm1⌬ cells and incubated with substrate in the presence of [␥-32 P]ATP. CaMKK␣ phosphorylated Snf1KD-K84R more strongly than it phosphorylated Snf1KD-T210A (Fig. 4) and did not phosphorylate GST alone (data not shown). The kinasedead control CaMKK␣D293A was immunoprecipitated as confirmed by immunoblot but did not phosphorylate a substrate (Fig. 4). Thus, CaMKK␣ primarily phosphorylates Thr-210 but also recognizes another site(s). Together, these genetic and biochemical data strongly suggest that CaMKK␣ is a member of the family of Snf1/AMPK kinases. Snf1 Is Regulated by Glucose Signals When Activated by LKB1 or CaMKK␣-We next took advantage of the heterologous function of these mammalian kinases to address questions regarding regulatory mechanisms in yeast. In wild-type yeast cells, Snf1 catalytic activity is inhibited by glucose signals (15,46,47); however, the glucose sensor/signal(s) and its targets have not been identified. One model is that the activating kinases Pak1, Tos3, and Elm1 are regulated, but it is also possible that Snf1 protein kinase directly senses a glucose signal or that the target is the Reg1-Glc7 form of protein phosphatase 1 (PP1), which dephosphorylates Thr-210 and negatively regulates Snf1 function in various indirect assays (47)(48)(49)(50), or that some combination of the above is involved. To address the role of the activating kinases, we examined the regulation of Snf1 protein kinase in triple mutant cells expressing LKB1 or CaMKK␣. We reasoned that if activation by a heterologous kinase abolished glucose regulation of Snf1 activity, this result would strongly implicate the native Snf1 protein kinase kinases in glucose signaling.
We assayed Snf1 catalytic activity during growth in high (2%) glucose and following a shift to limiting (0.05%) glucose for 30 min. In cells expressing either mammalian kinase, Snf1 activity was significantly inhibited by glucose, as was the case in wild-type cells (Fig. 5A). Levels of Snf1 protein were comparable in all samples (Fig. 5C). Thus, glucose regulation of Snf1 activity was maintained in the presence of two different heterologous activating kinases. Several lines of evidence suggest that LKB1 is constitutively active in mammalian cells (27,33,51), and there is no reported connection between glucose limitation and Ca 2ϩ signaling in yeast (52,53). Thus, although it is conceivable that both LKB1 and CaMKK␣ serendipitously respond to yeast glucose signals, these findings strongly suggest that the native activating kinases are not solely, if at all, responsible for mediating glucose signals and that at least one other regulatory mechanism is operative.
To further explore the role of protein phosphatase 1, we assayed Snf1 catalytic activity in reg1⌬ cells, which lack the subunit that targets protein phosphatase 1 to Snf1 (49,50). Activity was the same during growth in high glucose and after a shift to low glucose, indicating that the reg1⌬ mutation abolishes glucose inhibition of Snf1 activity (Fig. 5B); both Snf1  H) glucose. An aliquot of the culture was collected by filtration for assay (open bars). Another aliquot was collected by filtration, resuspended in SC plus 0.05% (low, L) glucose for 30 min, and collected by filtration (filled bars). Extracts were prepared, and Snf1 was partially purified. A and B, Snf1 catalytic activity was assayed by phosphorylation of the SAMS peptide. Values are averages of 4 -6 assays from two independent cultures. C, assayed fractions from the above samples were analyzed by immunoblotting with anti-Snf1. For comparison, some samples are shown on two blots. activity and Snf1 protein levels were somewhat lower in reg1⌬ fractions than in wild-type fractions (Fig. 5, A and C). The reg1⌬ mutation also abolished glucose regulation of Snf1 activity in reg1⌬ pak1⌬ tos3⌬ elm1⌬ cells expressing LKB1 or CaMKK␣ (Fig. 5B). Importantly, the level of Snf1 activity was different in each case, indicating that the absence of Reg1 did not simply result in maximal phosphorylation of all Snf1 proteins in the cell. Moreover, in glucose-limited cells expressing any of the Snf1-activating kinases, the absence of Reg1 did not increase Snf1 activity (Fig. 5, compare A and B), indicating that Reg1-Glc7 does not negatively regulate Snf1 under these conditions. Together, these findings strongly suggest that the function of Reg1-Glc7 toward Snf1 is positively regulated by glucose signals, but further study of Reg1-Glc7 will be required to establish this point.
LKB1 and CaMKK␣ Do Not Confer Nuclear Enrichment of Snf1-Gal83-We also used the mammalian kinases to address the mechanisms regulating the subcellular localization of Snf1 protein kinase. Activation of Snf1 is required for the nuclear enrichment of Snf1 protein kinase containing the ␤ subunit Gal83 (called Snf1-Gal83) in response to carbon stress (38,54). Nuclear enrichment of Snf1 depends on Gal83; however, enrichment is also abolished by the T210A mutation and by the pak1⌬ single mutation (38). Nuclear enrichment of Gal83 occurs in the snf1⌬ mutant but depends on Pak1 when Snf1 is present (38). The pak1⌬ mutation reduced Snf1-Gal83 catalytic activity much more severely than tos3⌬ or elm1⌬, indicating that Pak1 has a more major role than Tos3 or Elm1 in activating Snf1-Gal83, and overexpression of Tos3 partially restored nuclear enrichment in the pak1⌬ mutant (38). These results suggested that the apparently specific role of Pak1 in regulating the subcellular localization of Snf1-Gal83 could simply reflect its major role in activating Snf1-Gal83.
To address this issue, we tested whether LKB1-STRAD␣-MO25␣ or CaMKK␣ suffices for nuclear localization of Snf1-Gal83. We examined the localization of Gal83 tagged with green fluorescent protein (GFP), expressed from its native promoter on a centromeric plasmid (54), in pak1⌬ tos3⌬ elm1⌬ cells expressing each kinase. In both cases, Gal83-GFP was cytosolic when cells were grown in 2% glucose, as expected; however, when cells were shifted to 0.05% glucose, no relocalization to the nucleus was detected ( Fig. 6 and data not shown). Thus, neither heterologous kinase complemented the localization defect, despite the fact that CaMKK␣ substantially activates Snf1. To exclude the possibility that CaMKK␣ recognizes Snf1 protein kinase containing the ␤ subunit Sip1 or Sip2 rather than Snf1-Gal83, we showed that CaMKK␣ restores growth on glycerol-ethanol in cells containing only Snf1-Gal83 (sip1⌬ sip2⌬ pak1⌬ tos3⌬ elm1⌬ cells; data not shown). These findings suggest that nuclear enrichment of Snf1-Gal83 requires not only activation of Snf1 but also phosphorylation of another target or another site in Snf1 besides Thr-210, which is recognized only by Pak1. Thus, these findings reveal a new function of Pak1 that is not shared by other members of the Snf1/AMPK kinase family. We also note that many of the triple mutant cells expressing LKB1-STRAD␣-MO25␣ or CaMKK␣ exhibited the aberrant elongated cell morphology associated with the elm1⌬ mutation, indicating that neither heterologous kinase restored wild-type morphology. DISCUSSION We show that LKB1, a mammalian AMPK kinase, functions heterologously as a Snf1 protein kinase kinase both in yeast cells and in vitro. In mutant yeast cells lacking the native Snf1-activating kinases, LKB1 restored growth on raffinose and activated Snf1 catalytic activity. Coexpression of its partners STRAD␣ and MO25␣ resulted in elevated levels of LKB1 protein and an improved Snf1-activating function. The ability of a mammalian AMPK kinase to function heterologously in yeast cells indicates that the Snf1/AMPK kinase cascade has been highly conserved. LKB1 function in yeast, moreover, confers an easily selectable growth phenotype. Thus, the power of the yeast genetic system can be exploited for mutational analysis of the structure and function of LKB1 and the LKB1-STRAD-MO25 complex.
We also tested the ability of CaMKK␣, which is homologous to Pak1, Tos3, and Elm1, to activate Snf1 in yeast. CaMKK␣ effectively provided Snf1 protein kinase kinase function in the pak1⌬ tos3⌬ elm1⌬ mutant and phosphorylated the activation loop threonine of Snf1 in vitro. These findings support CaMKK␣ as a candidate for an authentic AMPK kinase in mammalian cells (see Fig. 1). Both CaMKK␣ and CaMKK␤ are highly expressed in brain and are also present in other tissues (34,35).
We took further advantage of the heterologous function of LKB1 and CaMKK␣ in yeast to explore mechanisms regulating the catalytic activity and subcellular localization of Snf1 protein kinase. First, we examined the role of the Snf1-activating kinases in glucose signaling. We found that when the heterologous kinases were responsible for activation, Snf1 activity was still significantly inhibited by glucose. This result indicates that unless both LKB1 and CaMKK␣ fortuitously respond to the yeast glucose signals, at least one regulatory mechanism operates independently of the activating kinases. Additional studies of the effects of the reg1⌬ mutation in cells expressing LKB1, CaMKK␣, or the native Snf1-activating kinases suggest that inhibition of Snf1 activity by Reg1-Glc7, which dephosphorylates Thr-210 (48), is regulated by glucose signals. The regulatory role of protein phosphatase 1 in glucose signaling warrants further investigation.
Second, neither heterologous kinase rescued the defect in nuclear enrichment of Gal83-GFP associated with the pak1⌬ mutation despite substantial activation of Snf1 catalytic activity, at least by CaMKK␣. Thus, although activation of Snf1 is essential for the nuclear enrichment of Snf1-Gal83, it is not sufficient. Pak1 must also have a second role in regulating localization, perhaps involving phosphorylation of another site in Snf1 protein kinase, besides Thr-210 of the catalytic subunit, or phosphorylation of an as yet unidentified protein. These findings indicate that Pak1 has at least one function in addition to its role as a Snf1-activating kinase that is not shared by other members of the Snf1/AMPK kinase family. Similarly, Elm1 has unique roles in other cellular processes (21)(22)(23)(24)(25).
Finally, this demonstration of heterologous function of LKB1 and CaMKK␣ in yeast provides the basis for a convenient and powerful genetic selection for mammalian AMPK kinases, the restoration of the Snf ϩ growth phenotype in pak1⌬ tos3⌬ elm1⌬ mutant yeast. Given that yeast, a relatively simple unicellular organism, has three members of the Snf1/AMPK kinase family, it seems highly likely that mammals have multiple AMPK kinases. The power of this selection lies not only in its simplic-FIG. 6. Localization of Gal83-GFP. Cells of the pak1⌬ tos3⌬ elm1⌬ strain expressing Gal83-GFP and CaMKK␣ were grown in SC plus 2% glucose and shifted to 0.05% glucose for 30 min. Nuclei were stained with 4Ј,6-diamidino-2-phenylindole (DAPI), and cells were viewed using a Nikon Eclipse E800 fluorescent microscope and photographed as described (38). GFP fluorescence, DAPI staining, and differential interference contrast (DIC) are shown. ity but, more importantly, in its sensitivity; the Snf1 pathway is robust, and very little activity is required for growth. Comparison of the growth assay and the biochemical assay for LKB1 function underscores this point; restoration of growth by LKB1 was unmistakable, whereas the elevation of Snf1 catalytic activity was modest. Sensitivity is crucial, because the AMPK kinase catalytic subunit must function in yeast without other mammalian proteins, either alone or in association with yeast orthologs. The identification of the entire family of AMPK kinases is important, because AMPK kinases are potential therapeutic targets in the AMPK pathway that may be specific to particular cell types or to particular metabolic stress and endocrine signals.