Opposite Transcriptional Regulation in Skeletal Muscle of AMP-activated Protein Kinase γ3 R225Q Transgenic Versus Knock-out Mice*

AMP-activated protein kinase (AMPK) is an evolutionarily conserved heterotrimer important for metabolic sensing in all eukaryotes. The muscle-specific isoform of the regulatory γ-subunit of the kinase, AMPK γ3, has an important role in glucose uptake, glycogen synthesis, and fat oxidation in white skeletal muscle, as previously demonstrated by physiological characterization of AMPK γ3 mutant (R225Q) transgenic (TgPrkag3225Q) and γ3 knock-out (Prkag3-/-) mice. We determined AMPK γ3-dependent regulation of gene expression by analyzing global transcription profiles in glycolytic skeletal muscle from γ3 mutant transgenic and knock-out mice using oligonucleotide microarray technology. Evidence is provided for coordinated and reciprocal regulation of multiple key components in glucose and fat metabolism, as well as skeletal muscle ergogenics in TgPrkag3225Q and Prkag3-/- mice. The differential gene expression profile was consistent with the physiological differences between the models, providing a molecular mechanism for the observed phenotype. The striking pattern of opposing transcriptional changes between TgPrkag3225Q and Prkag3-/- mice identifies differentially expressed targets being truly regulated by AMPK and is consistent with the view that R225Q is an activating mutation, in terms of its downstream effects. Additionally, we identified a wide array of novel targets and regulatory pathways for AMPK in skeletal muscle.

AMP-activated protein kinase (AMPK) 2 is a critical regulator of carbohydrate and fat metabolism in eukaryotic cells (reviewed in Refs. 1 and 2). AMPK is a heterotrimer that consists of ␣-, ␤-, and ␥-subunits, all of which are required for its activity. The catalytic ␣-subunit contains a conventional serine/threonine protein kinase domain, and phosphorylation of Thr-172 residue within the activation loop of the ␣-subunit by upstream kinases is essential for the activity of the heterotrimer (3)(4)(5)(6). Once phosphorylated at Thr-172, AMPK can be further activated by allosteric binding of AMP to the evolutionary conserved cys-tathionine ␤-synthase domains in the regulatory ␥-subunit (7). The AMPK ␤-subunit acts as a scaffold for binding of the ␣and ␥-subunits (8). The ␤-subunit also contains a glycogen-binding domain, and recent findings provide evidence that this motif is involved in targeting the AMPK complex to cellular glycogen stores (9,10). The mammalian genome contains seven AMPK genes encoding for two ␣-, two ␤-, and three ␥-isoforms. Thus, there are 12 possible combinations of heterotrimeric AMPK, and the physiological function of the AMPK holoenzyme depends on the particular isoforms present in the complex.
We have provided evidence that AMPK ␥3 is the predominant ␥-isoform expressed in glycolytic (white, fast-twitch type II) skeletal muscle (11). In contrast, it is expressed at low levels in oxidative (red, slowtwitch type I) skeletal muscle and is undetectable in brain, liver, heart, or white adipose tissue (11). Thus, the AMPK ␥3-subunit is the only isoform exhibiting tissue-specific expression. Furthermore, the AMPK ␥3-subunit primarily forms heterotrimers with the ␣2and ␤2-isoforms in glycolytic skeletal muscle (11).
The functional significance of the AMPK ␥3-subunit has been demonstrated by phenotypic analysis of animal models carrying a mutated form of the gene. The dominant Rendement Napole (RN) phenotype identified in Hampshire pigs is caused by a single missense mutation (R225Q) in the AMPK ␥3-subunit (12). RN pigs have a 70% increase in glycogen content in skeletal muscle, whereas liver and heart glycogen content remains unchanged (13). Furthermore, RN carriers are also characterized by a higher oxidative capacity in white skeletal muscle fibers (14,15). Conversely, a second mutation (V224I) identified in pigs at the neighboring amino acid residue of the ␥3-protein is associated with an opposite phenotype compared with the RN allele, resulting in reduced skeletal muscle glycogen content (16). Characterization of transgenic mice with skeletal muscle-specific expression of the mutant (R225Q) form of the mouse AMPK ␥3-subunit, as well as AMPK ␥3-subunit knock-out mice, provided further evidence that the ␥3-subunit plays a key role in skeletal muscle carbohydrate and lipid metabolism. Glycogen resynthesis after exercise was impaired in AMPK ␥3 knock-out mice but was markedly enhanced in transgenic mutant mice. An AMPK-activator failed to increase skeletal muscle glucose uptake in knock-out mice, whereas insulin-mediated glucose uptake was unaltered. When fed with a high fat diet, ␥3 R225Q transgenic mice were protected against excessive triglyceride accumulation and insulin resistance in skeletal muscle, presumably due to an increase in fat oxidation (17). Additionally, skeletal muscle from ␥3 R225Q mutant mice is characterized by enhanced work performance, whereas knock-out mice are fatigue-prone (18).
To further characterize the role of AMPK ␥3 in skeletal muscle and to uncover molecular mechanisms explaining phenotypic consequences of the mutations in this isoform, we have studied AMPK ␥3-dependent gene transcription by a systematic approach, using global analysis of the mRNA expression pattern in the skeletal muscle of ␥3 R225Q mutant and ␥3 knock-out mice. Here we describe distinct biomarker patterns, comprising AMPK ␥3-dependent transcriptional changes of genes involved in glucose and lipid metabolism and skeletal muscle ergogenics.

EXPERIMENTAL PROCEDURES
AMPK Knock-out (Prkag3 Ϫ/Ϫ ) and R225Q Transgenic (TgPrkag3 225Q ) Mice-The Prkag3 Ϫ/Ϫ and TgPrkag3 225Q mice used in this study have been previously described (17). Prkag3 Ϫ/Ϫ mice were created by conventional gene targeting techniques. TgPrkag3 225Q mice express the mutant ␥3 R225Q subunit under the control of mouse myosin-light chain promoter and enhancer elements. Mice used in the study were bred into the C57BL/6 genetic background. Mice were maintained in a 12-h light-dark cycle and were cared for in accordance with regulations for the protection of laboratory animals. The study was performed after prior approval from the local ethical committee. Gene expression profiles were characterized in male mice fasted overnight (food was removed 16 h prior to study). The white portion of the gastrocnemius muscle was dissected from anesthetized mice, cleaned of fat and blood, and quickly frozen in liquid nitrogen as described (17).
Preparation of Total RNA-Total RNA was isolated from the white portion of the gastrocnemius muscle using the RNeasy Fibrous Mini Kit (Qiagen) applying Mixer Mill MM 301 (Retsch) followed by a DNase digestion step using RNase-Free DNase Set (Qiagen) according to the manufacturer's instructions. The RNA yield was quantified by spectrophotometric analysis and the RNA purity was determined based on the A 260 /A 280 ratio. The quality of the RNA was confirmed by Agilent 2100 Bioanalyzer analysis (Agilent Technologies) using the RNA 6000 Nano Assay Kit (Agilent Technologies).
Preparation of cRNA, Gene Chip Hybridization-10 g of total RNA spiked with poly-A controls (pGIBS-TRP, -THR, and -LYS, American Type Culture Collection) was converted to cDNA utilizing a T7 promoter-polyT primer (Affymetrix) and the reverse transcriptase Superscript II (Invitrogen), followed by a second strand cDNA synthesis (Invitrogen). Double-stranded cDNA was in vitro transcribed to biotinylated cRNA (Enzo) and then fragmented (Invitrogen). The fragmented cRNA was mixed with control oligonucleotide B2 (Affymetrix) and a hybridization control cRNA mixture (BioB, BioC, BioD, and Cre, Affymetrix). Aliquots of each sample were hybridized (16 h at 45°C) to GeneChip Mouse Expression Set 430A arrays (Affymetrix). The arrays were subsequently washed, stained, and scanned according the manufacturer's instructions (GeneChip Expression Analysis Technical Manual, Affymetrix).
Quantitative Real-time PCR-Quantification of mRNA levels for selected genes was performed by qRT-PCR as described (19) using acidic ribosomal phosphoprotein P0 (Arbp) as endogenous control (see supplemental Table SI for primer information). qRT-PCR was performed on extended set of samples including 7 Prkag3 Ϫ/Ϫ mice with 8 wild-type littermates and 13 TgPrkag3 225Q mice with 10 wild-type littermates, while RNA from 6 animals in each group was used in gene array analysis.
Histochemistry-Enzyme activity staining for succinate dehydrogenase and cytochrome c oxidase was done on serial cross-sections (10-m thickness) of frozen gastrocnemius muscle as described previously (20,21). For succinate dehydrogenase activity staining, sections were incu-bated for 4 min in a 0.1 M phosphate buffer (pH 7.6) containing 5 mM EDTA, 45 mM disodium succinate, 1.2 mM nitro blue tetrazolium, 1 mM potassium cyanide, and 1 mM phenazine methosulfate. Cytochrome c oxidase activity staining was performed by incubating sections for 1 h in a 50 mM phosphate buffer (pH 7.6) containing 0.22 M sucrose, 2.3 mM 3,3Ј-diaminobenzidine tetrahydrochloride, 1 mM cytochrome c, and 1300 units of catalase.

RESULTS
Microarray Analysis of the mRNA Expression in the Skeletal Muscle of AMPK ␥3 Knock-out (Prkag3 Ϫ/Ϫ ) and R225Q Transgenic (TgPrkag3 225Q ) Mice-To determine the role of ␥3-containing AMPK complexes in regulation of gene expression in the skeletal muscle, we utilized mouse models that either lack the AMPK ␥3-protein (Prkag3 Ϫ/Ϫ ) or express a R225Q mutant form of this protein in skeletal muscle (TgPrkag3 225Q ) (17). In Prkag3 Ϫ/Ϫ mice, AMPK ␥3-protein expression is completely ablated, and importantly, no compensatory increase in ␥1or ␥2-isoform is detected (17). Equally important, AMPK expression in TgPrkag3 225Q mice resembles the expression pattern in wild-type mice, both with regard to tissue distribution and protein expression, with the mutant (R225Q) form replacing the endogenous AMPK ␥3-protein (17). Global gene expression profiles in the white portion of gastrocnemius muscle of Prkag3 Ϫ/Ϫ and TgPrkag3 225Q mice were compared with the corresponding wild-type littermates using oligonucleotide microarrays. The expression of 167 genes was significantly (p Յ 0.05) changed by a factor of 20% or more, in TgPrkag3 225Q and/or Prkag3 Ϫ/Ϫ mice, relative to the wild-type controls ( Table 1). Applying the same filtering criteria on randomly created groups within the Prkag3 Ϫ/Ϫ dataset and TgPrkag3 225Q dataset resulted in six genes determined as differentially expressed. This indicates that the rate of false positives is low. Consequently, the vast majority of the genes appearing differently expressed in Prkag3 Ϫ/Ϫ and/or TgPrkag3 225Q mice can be considered as truly regulated.
Of the 167 differentially expressed transcripts, the identity of 148 genes is known and represents proteins of different functional classes, whereas 19 transcripts only show homology to sequences in the EST or genomic databases. Interestingly, the expression level of 16 genes was significantly changed in both AMPK ␥3 R225Q transgenic and AMPK ␥3 knock-out mice, compared with their respective wild-type littermates (Table 1). For these 16 transcripts, the direction of the observed change was the opposite in knock-out versus mutant transgenic mice. Furthermore, most of the genes that were significantly changed exclusively in Prkag3 Ϫ/Ϫ mice tended to be regulated in an opposite manner in TgPrkag3 225Q mice, even though this difference did not reach statistical significance and/or meet the -fold change criteria. Correspondingly, the vast majority of transcripts, which were differentially regulated exclusively in R225Q transgenic mice, exhibited an opposite trend in knock-out mice. The striking pattern of opposing transcriptional changes in the AMPK ␥3 R225Q transgenic versus knock-out mice, as compared with their wild-type littermates, is illustrated (Fig. 1).
Many of the genes, which were found to be differentially expressed in Prkag3 Ϫ/Ϫ and/or TgPrkag3 225Q mice, are previously undescribed as being regulated by AMPK. Full functional significance of these changes in global transcriptional profile remains to be addressed in further experiments. To determine the possible mechanistic explanations for previously described physiological differences between Prkag3 Ϫ/Ϫ and TgPrkag3 225Q mice (17,18), we performed a more detailed analysis of gene expression changes for targets that are known to be involved in lipid and carbohydrate metabolism and muscle ergogenics. The expression of several genes involved in these functions depends on the skeletal Global mRNA expression pattern was characterized in the white portion of the gastrocnemius skeletal muscle in male mice of C57BL/6 genetic background. The filtering criteria were set to a mean absolute -fold change Ͼ1.2 and a p value Յ0.05. In addition the mean intensity in the group showing highest expression should be Ͼ75.  muscle fiber type, and, correspondingly, any alterations in skeletal muscle fiber type composition might contribute to expression differences. However, relative expression of slow and fast myosin and troponin isoforms remained unchanged in TgPrkag3 225Q or Prkag3 Ϫ/Ϫ mice versus their respective wild-type littermates (data not shown). Moreover, enzyme activity staining for succinate dehydrogenase and cytochrome c oxidase (markers for oxidative energy metabolism that stain red muscle fibers containing high levels of mitochondria more intensively than white fibers with fewer mitochondria) did not show any clear alteration in fiber type composition between the different genotypes (supplementary Fig. S1). Taken together, these data indicate that differences in the transcriptional profile we describe are independent of skeletal muscle fiber type changes. qRT-PCR Validation of Differentially Expressed Genes-To minimize erroneous conclusions due to technical variability of the microarray technology, qRT-PCR analysis was applied to validate expression profiles of 13 genes selected on the basis of biological relevance (Fig. 2). For all the transcripts examined, qRT-PCR data verified the significant differences in gene expression (-fold change Ͼ1.2; p Յ 0.05) detected by gene array analysis. Furthermore, individual animal-to-animal comparison of the expression profiles for these genes showed close to perfect correlation comparing the two techniques (data not shown). The high level of correlation between the expression profiles generated by microarray versus qRT-PCR approach illustrates the reliability of the gene array results. However, for 11 of 13 transcripts examined the microarray data tended to underestimate the expression change compared with qRT-PCR results.

Public ID Gene symbol Gene title
Altered Expression of Components of the Glycogen Synthesis Pathway in AMPK ␥3 Knock-out and R225Q Transgenic Mice-AMPK function is closely connected to glycogen storage. In human and rat skeletal mus-cle, high glycogen content represses AMPK activity (22,23). Concomitantly, there is also genetic evidence that AMPK regulates glycogen levels, because mutations of the ␥3or ␥2-subunit affect glycogen storage in skeletal muscle of RN pigs or in human cardiac muscle, respectively (12,16,24). Furthermore, AMPK ␥3 R225Q transgenic and ␥3 knock-out mice have a respective increase or decrease in the rate of glycogenesis in the recovery phase following exercise (17). Notably, the expression of two transcripts encoding proteins involved in glycogen synthesis, Ugp2 and Gbe1, was significantly up-regulated in TgPrkag3 225Q mice, while being significantly down-regulated in Prkag3 Ϫ/Ϫ mice, as determined by the microarray analysis ( Table 1). The observed change in the expression pattern was further confirmed by a qRT-PCR approach (Fig. 2). Ugp2 codes for UDP-glucose pyrophosphorylase 2 (EC 2.7.7.9), an enzyme catalyzing the synthesis of UDP-glucose, a common substrate for glycogenin and glycogen synthase. Glycogenin catalyzes the first step in glycogen synthesis: a selfglycosylation reaction to form an oligosaccharide chain of around eight residues in length (25). Secondly, glycogen synthase (EC 2.4.1.11), with the participation of the glycogen branching enzyme, Gbe1 (EC 2.4.1.18), elongates the oligosaccharide chain to form a mature glycogen molecule. Thus, coordinated changes in Ugp2 and Gbe1 expression are likely to contribute to the differences in the glycogen synthesis rate observed comparing Prkag3 Ϫ/Ϫ and TgPrkag3 225Q muscle.

Coordinated and Reciprocal Changes in the Expression of Map2k6 and Dusp10 Genes in Prkag3
Ϫ/Ϫ and TgPrkag3 225Q Muscle-Based on correlative evidence, stimulation of glucose uptake has been reported to be partly regulated by Mapk14 (also known as p38 MAPK), a downstream target of AMPK (26 -30). However, the exact mechanism of how the activation of AMPK would lead to an increase in Mapk14 phosphorylation remains obscure. Interestingly, two transcripts encoding proteins involved in regulation of Mapk14 activity, Map2k6 and Dusp10, were coordinately regulated in ␥3 R225Q transgenic and knock-out mice, as shown by microarray analysis as well as qRT-PCR (Fig. 2). A protein encoded by Map2k6 (mitogen-activated protein kinase kinase 6) is known to activate Mapk14 by dual phosphorylation of specific threonine and tyrosine residues (31). Dusp10 (dual specificity phosphatase 10), on the other hand, down-regulates the enzymatic activity of MAPKs by dephosphorylating the threonine and tyrosine residues, with selectivity toward Mapk14 (32). Thus, the finding of an up-regulation of Map2k6 in combination with a suppression of the Dusp10 transcript in TgPrkag3 225Q mice, and the reversed pattern of changes seen in Prkag3 Ϫ/Ϫ mice, suggests that Mapk14 is a target of AMPK ␥3-containing trimers in the skeletal muscle. An up-regulation of Map2k6 mRNA in TgPrkag3 225Q muscle was accompanied by an increase in the protein level, as seen by Western blot analysis (data not shown).

␥3-Containing AMPK Heterotrimers Regulate Lipid Metabolism Gene Expression in Skeletal
Muscle-The AMPK ␥3-subunit has previously been shown to be involved in regulation of fat oxidation. Pigs and mice carrying the R225Q mutation in AMPK ␥3-gene are characterized by increased lipid oxidation in skeletal muscle (14,15,17). In the microarray and qRT-PCR analysis, several genes involved in fat metabolism were differentially expressed in TgPrkag3 225Q and Prkag3 Ϫ/Ϫ mice, suggesting that ␥3-containing AMPK complexes are involved in the regulation of lipid oxidation in skeletal muscle at the transcriptional level. mRNA for Srebf1 (sterol regulatory element binding factor 1), implicated in lipogenic gene expression (33), was down-regulated in ␥3 R225Q mutant mice, whereas mRNA encoding for Ppargc1a (peroxisome proliferative-activated receptor, ␥ coactivator 1 ␣), known to increase the expression of both nuclear and mitochondrial-encoded genes of oxidative metabolism (34), was up-regulated. Additionally, a key gene integral to free fatty acid uptake (Cd36 (35)), as well as genes involved in use of fat-derived energy (3-oxoacid-CoA transferase 1, Oxct1, EC 2.8.3.5 and carboxylesterase 3, Ces3, EC 3.1.1.1) were up-regulated in TgPrkag3 225Q mice. The opposite pattern of changes was observed in ␥3 knock-out mice, with differential expression of mRNA for Cd36, Oxct1, and Ppargc1a reaching statistical significance (Fig. 2).  Possible Role for the AMPK ␥3-Subunit in the Regulation of Cellular Iron Homeostasis-Expression of several genes involved in iron metabolism was coordinately changed in AMPK ␥3 R225Q transgenic mice, compared with the wild-type littermates. The transferrin receptor, ferritin, and aminolevulinic acid synthase (a key enzyme for heme synthesis) regulate the uptake, storage, and use of iron in cells, respectively. Expression of these markers is known to be coordinately controlled by cellular iron supply such that, under conditions of iron starvation, the transferrin receptor is up-regulated, whereas ferritin and aminolevulinic acid synthase are down-regulated (36). In TgPrkag3 225Q mice, the transcription of the transferrin receptor, Tfrc was decreased, while mRNAs for ferritin light chain 1 and aminolevulinic acid synthase 1 were increased, suggesting the possibility of improved iron status compared with the wild-type mice (Fig. 2). Improved iron status may increase skeletal muscle capacity for aerobic metabolism (37,38). In line with this, the level of skeletal muscle myoglobin, a heme-carrying protein, transporting oxygen to the mitochondria, was non-significantly increased in TgPrkag3 225Q mice, as determined by Western blot analysis (data not shown).
AMPK ␥3-Containing Complexes Regulate Transcription of the Nos1 Gene-Nitric oxide synthase 1 (Nos1 also known as nNos) was significantly up-and down-regulated by ϳ2-fold in ␥3 R225Q transgenic and knock-out mice, respectively (Fig. 2). The family of Nos enzymes catalyzes formation of endogenous nitric oxide, a mediator implicated in regulation of skeletal muscle contractility, mitochondrial function, as well as glucose uptake (39,40). Thus, the direction of differential expression of this gene was in complete agreement with the observed phenotype in the mouse models. Although Nos1 is considered a predominant isoform expressed in skeletal muscle fibers, endothelial Nos3 (Nos3 or eNos) and inducible Nos2 (Nos2 or iNos) are also expressed in this tissue (41)(42)(43). Interestingly, Nos2 and Nos3 were represented on oligonucleotide microarray and were detected in the skeletal muscle samples. However, mRNA levels for these two genes were unaltered.

DISCUSSION
The present study provides the first systematic characterization of the role of the ␥3-containing AMPK heterotrimers in transcriptional regulation in skeletal muscle. An oligonucleotide microarray technology was used to compare global transcriptional profiles in white skeletal muscle from AMPK ␥3 R225Q transgenic (TgPrkag3 225Q ) and knockout (Prkag3 Ϫ/Ϫ ) mice and their respective wild-type littermates. Collectively, evidence is provided for an important role of the AMPK ␥3-subunit in the transcriptional regulation of divergent groups of genes in glycolytic skeletal muscle. The expression of 167 genes was significantly changed (-fold change Ͼ 1.2 and p Յ 0.05) in skeletal muscle from ␥3 R225Q transgenic and/or knock-out mice. A number of genes from the same biological pathways was coordinately regulated. Changes in levels of mRNA of particular interest, including genes involved in glycogen and lipid metabolism, as well as iron homeostasis and Mapk14 signaling, were selected to be further verified by qRT-PCR (Fig. 2). For all of 13 selected genes, the qRT-PCR results were consistent with the gene array data, which emphasizes the quality of the microarray results. Of note is that most of the genes differentially regulated in TgPrkag3 225Q mice were changed in an opposing manner in AMPK ␥3 knock-out mice (Fig. 1). The reciprocal and coordinated expression pattern observed increases the confidence in the gene-array data and demonstrates that the majority of differentially expressed genes are true positives.
This reverse transcription pattern seen in TgPrkag3 225Q versus Prkag3 Ϫ/Ϫ mice is intriguing, considering that the mechanism of action of R225Q substitution in AMPK ␥3-subunit is still unresolved. A previ-ous report provided evidence that the kinase activity of AMPK was reduced by ϳ3-fold in skeletal muscle of ␥3 R225Q mutant pigs when measured in the presence and absence of the allosteric activator, AMP (12). Nevertheless, in resting muscle from TgPrkag3 225Q mice, AMPK activity was unaltered (18). Excessive glycogen content characterizing skeletal muscle of mice and pigs carrying the ␥3 R225Q mutation may inhibit AMPK activation by a feedback mechanism. Consistent with this hypothesis, AMPK activity and phosphorylation of the Thr 172 residue in the ␣-subunit was elevated in Cos cells transiently transfected with plasmids encoding ␣2␤2␥3 R225Q, compared with the cells transfected with plasmids encoding wild-type trimers (17). Interestingly, the introduction of the R225Q mutation at the corresponding site of the AMPK ␥1 (R70Q) or ␥2 (R302Q) leads to increased or decreased kinase activity, respectively, when measured in the presence of AMP (44,45). Our data support a role for the ␥3 R225Q as an activating mutation, as judged by its biological effects, because this substitution rendered the opposing changes in the gene expression profile compared with the changes resulting from genetic ablation of the AMPK ␥3-subunit (Fig. 1).
The wide array of differentially expressed genes reported in this study has not previously been identified to be regulated by AMPK. Moreover, the biological function of a number of the regulated transcripts is poorly described in skeletal muscle and 19 of the differentially expressed genes remain unknown. Therefore, additional molecular and functional studies will be required to understand the full biological significance of the transcriptional changes described here. Importantly, the expression of several genes with known function in fat and carbohydrate metabolism and skeletal muscle ergogenics was coordinately and reciprocally changed in AMPK ␥3 R225Q transgenic and knock-out mice. Thus, the transcriptional regulation by AMPK ␥3-containing complexes could lead to at least some of the physiological differences observed in skeletal muscle from TgPrkag3 225Q and Prkag3 Ϫ/Ϫ mice (Fig. 3). and Prkag3 Ϫ/Ϫ mice (indicated by blue arrows), compared with wild-type littermates, and the predicted physiological response (B). The role of AMPK␥3 in regulation of functions like glycogen deposition, glucose uptake, oxidative metabolism, and muscle fatigue has previously been described (17,18). The current study provides a transcriptional mechanism for the physiological differences between the genotypes.
One of the most distinct physiological differences between skeletal muscle from AMPK ␥3 R225Q transgenic and knock-out mice is the change in glycogen metabolism (17). Glycogen content in skeletal muscle from TgPrkag3 225Q mice is 2-fold higher under either fed or fasted conditions. Furthermore, skeletal muscle glycogen re-synthesis after exercise is markedly enhanced in TgPrkag3 225Q mice. In contrast, glycogenesis after exercise was impaired in Prkag3 Ϫ/Ϫ mice (17). Glycogen synthase catalyzes a rate-determining step in the glycogen biosynthesis pathway. However, at least in situations where glycogen synthase is activated, the reactions mediated by other enzymes in synthesis pathway can become rate-limiting (46). Interestingly, the microarray data and qRT-PCR analysis revealed a significant up-and down-regulation of Ugp2 and Gbe1 mRNAs (encoding two key enzymes in glyocogen synthesis) in skeletal muscle from AMPK ␥3 R225Q transgenic and knockout mice, respectively. In agreement with these results, the protein level and enzyme activity of Ugp2 and Gbe1 is increased in RN pigs carrying an R225Q mutant form of the AMPK ␥3-subunit (14,47). Based on these observations, we hypothesize that the differences in the skeletal muscle glycogen re-synthesis rate displayed by ␥3 R225Q transgenic and knock-out mice may be at least partly explained by coordinated differential transcription of Ugp2 and Gbe1.
Activation of AMPK either by physiological stimulation such as muscle contraction or by the pharmacological activator 5-amino-4-imidazole-carboxamide riboside (AICAR) leads to an increase in skeletal muscle glucose transport by promoting Glut4 translocation to the cell surface (48), as well as increasing Glut4 gene transcription (49). Importantly, the AICAR effect on glucose uptake is completely abolished in AMPK ␥3 knock-out mice, suggesting that the ␥3-subunit is essential for AICAR-induced glucose transport in skeletal muscle (17). Nonetheless, our microarray data did not detect any difference in the expression of the Glut4 gene when comparing skeletal muscle from TgPrkag3 225Q and Prkag3 Ϫ/Ϫ mice. Mapk14 (also known as p38 MAPK) has been implicated as a downstream mediator of AMPK signaling to glucose transport in response to AICAR. In this context, it is interesting to note that mRNAs encoding two proteins involved in regulation of the phosphorylation pattern of Mapk14, Map2k6 and Dusp10, were coordinately and reciprocally regulated in AMPK ␥3 R225Q and knock-out mice. Another mediator implicated in AMPK-regulated glucose transport is nitric-oxide synthase, Nos (50). The microarray and qRT-PCR analyses revealed the mRNA encoding Nos1, the predominant nitric-oxide synthase isoform in skeletal muscle, was significantly up-and down-regulated by ϳ2-fold in TgPrkag3 225Q and Prkag3 Ϫ/Ϫ muscle, respectively. Thus, the present study provides the first evidence of Nos1 being a transcriptional target of AMPK signaling.
Previously, the AMPK ␥3-subunit has been shown to influence muscle ergogenics (18). In response to electrically stimulated muscle contraction, isolated EDL muscle from TgPrkag3 225Q mice is resistant to fatigue. Conversely, skeletal muscle from Prkag3 Ϫ/Ϫ mice is fatigue prone (18). The increase in glycogen content in skeletal muscle of TgPrkag3 225Q mice may have a direct positive effect on muscle performance. Additionally, skeletal muscle is known to adapt to endurance exercise by increasing the oxygen carrying capacity. Our microarray data demonstrate a coordinated up-regulation of genes involved in the storage (Ftl1) and use (Alas1) of cellular iron in TgPrkag3 225Q mice, which may indicate improved iron status and correspondingly, increased oxidative capacity in skeletal muscle. Accordingly, the level of skeletal muscle oxygen carrying protein, myoglobin, was increased in TgPrkag3 225Q mice, compared with the wild-type mice.
Endurance exercise is dependent upon oxidation of fatty acids as a major source of ATP. Two transcription factors implicated in the reg-ulation of fatty acid homeostasis, Ppargc1a and Srebf1, as well as several genes necessary for the transport (Cd36) and utilization (Ces3 and Oxct1) of fatty acids were differentially expressed in skeletal muscle from TgPrkag3 225Q mice, suggesting increased fatty acid availability and oxidation. In agreement with these observations, elevated reliance of fat-derived energy has been described in ␥3 R225Q transgenic mice during exercise, as well as after challenge with high fat diet (17,18).
We have applied global transcriptome analysis by oligonucleotide microarrays to systematically characterize the role of the AMPK ␥3-subunit in the regulation of gene expression. Recently, a number of studies have used qRT-PCR analysis to characterize the role of AMPK complexes in transcription regulation (51,52). However, gene array approaches allow for concurrent analysis of global mRNA profiles and offer an advantage of reducing bias in data collection, compared with the candidate gene-based approaches. Taken together, our data indicate that a number of genes involved in carbohydrate and fat metabolism, as well as skeletal muscle ergogenics are coordinately and reciprocally regulated in the skeletal muscle from AMPK ␥3 R225Q transgenic and knock-out mice and provide a molecular mechanism for the previously described physiological differences between the models. Furthermore, the current study identifies that AMPK ␥3-containing complexes play an important role in regulation of novel downstream targets and pathways in skeletal muscle.
AMPK has been identified as an attractive therapeutic target in the treatment of obesity and type II diabetes (53). However, the ideal therapeutic small molecule should modify the activity of AMPK heterotrimers in metabolic tissues such as skeletal muscle to increase fatty acid oxidation and glucose uptake without exerting any effect on organs including central nervous system, heart, pancreas, or lung, where AMPK activation potentially would lead to harmful consequences (54 -57). Therefore, therapeutic agents that selectively modulate the activity of ␥3-containing AMPK heterotrimers could potentially provide a way to specifically target skeletal muscle and thereby prevent any adverse effects in other organ types. The current study contributes to the understanding of the role of the AMPK ␥3-subunit in the regulation of gene transcription and defines several sets of potential biomarkers to characterize the molecular effects in response to the administration of lead substances ideally mimicking the TgPrkag3 225Q phenotypes in mice and pigs.