To evaluate the effect of AMPK activation on insulin signal transduction, 3T3-F442a adipocytes were treated with AICAR (1 mM, 2 h), followed by insulin (100 nM, 15 min), and IRS1-associated PI3K activity examined. As shown in Figure 1A, while AICAR treatment caused a slight inhibition of IRS1-associated PI3K activity at the basal level, it augmented the activity by almost 2-fold in the cells treated with insulin. Concurrently, pretreatment with AICAR enhanced insulin-stimulated phosphorylation of Akt at S473 by 90% (Figure 1B), which was accompanied by an increase in phosphorylation of GSK3 (data not shown). In contrast, insulin-stimulated phosphorylation of S6K1 was markedly suppressed by AICAR (Figure 1B). Similar results were obtained in differentiated 3T3-L1 adipocytes (data now shown)

To ascertain if the effects of AICAR are mediated by AMPK, we established stable cell lines in 3T3-F442a preadipocytes using a lentiviral system expressing a dominant negative AMPK α1 catalytic subunit (D157A mutation) 23 . We then assessed the effect of this mutant on AICAR-regulated phosphorylation of S473 on Akt in preadipocytes. One reason to use preadipocytes in this and the following experiments is that the ability of the preadipocytes expressing the mutant to differentiate into adipocytes is much greater than that in presence of the empty vector. As a result, it was difficult to compare the effects of insulin and AICAR in adipocytes differentiated from these two cell lines, owing to different expression levels of IRS1, insulin receptors, and AMPK (Data not shown). A second reason is that the expression of the dominant negative mutant somehow decreases the level of endogenous AMPKα subunits (Figure 2B), which seems to work as a true dominant negative interfering mutant. As shown in Figure 2A, the expression of the α1 mutant markedly inhibited insulin-stimulated Akt phosphorylation and also prevented the increase in the cells pretreated with AICAR. In contrast, mTOR/S6K1 signaling changed in an opposite direction, as manifested by the result that the α1 mutant prevented the reduction of S6K1 and S6 phosphorylation (Figure 2A). Thus, we conclude that the effects of AICAR on insulin signaling are mediated by AMPK.

To seek after the factor that mediates AMPK-induced activation of Akt, we treated 3T3-F442a preadipocytes with EGF, which activates Akt independent of IRS1. As shown in Figure 3A, the dominant negative mutant of AMPK displayed a similar inhibitory effect on EGF-stimulated Akt phosphorylation. By comparing the phospho-signal intensity, we noticed that the Akt phosphorylation by EGF was much weaker than that by insulin (Data not shown). Longer exposure of Western blot allowed visualization of enhanced Akt phosphorylation when the cells were incubated with AICAR alone for 2 hours, which was blunted by the dominant negative AMPK mutant. This suggests that AMPK induces Akt activation independent of IRS1.

To confirm the direct effect of AMPK activation on Akt phosphorylation, we chronically treated the preadipocytes with AICAR and assessed the phosphorylation of Akt. As shown in Figure 3B, AICAR progressively increased phosphorylation of S473, reaching the maximum after 4 h. This was abrogated by expression of the dominant negative mutant of AMPK.

Next, we examined if AMPK activation upregulates intracellular levels of PIP3. In doing so, the cells were treated with AICAR, as compared with insulin, and immunofluorescently stained with anti-PIP3 antibody. The data showed that the treatment of the cells with AICAR for 4 hours increased the abundance of PIP3 to a level similar to that induced by insulin (Figure 4A), whereas the effect of AICAR was almost completely suppressed by the expression of the dominant negative AMPK mutant (Figure 4B). Therefore, these results suggest that AMPK regulates Akt phosphorylation via regulating PI3K or PTEN, an event independent of IRS1.

We then asked if the effect of ACIAR could be blunted by Wortmannin, a specific inhibitor of PI3K. Thus, we treated 3T3-F442a preadipocytes with AICAR or insulin together with Wortaminnin. Figure 5 shows that Wortmainnin suppressed AICAR-induced Akt phoshorylation to an extent that is similar to its effect on insulin. To assess if PTEN is involved in AMPK regulation of Akt activation, we made stable cell lines expressing the dominant negative mutant PTEN C124S. As shown in Figure 6, the expression level of PTEN C124S reached approximately 10 times the endogenous protein; however, it did not display a significant effect on either insulin or AICAR-stimulated Akt activation. In a parallel experiment, the endogenous PTEN was knocked down with its shRNA by about 50%, which also did not yield an evident effect on Akt activation (data not shown). These results clearly exclude the possibility that AMPK regulates PTEN to impact on Akt activation. Therefore, they suggest that AMPK activates Akt via regulating PI3K.

Antibodies against phospho- and nonphospho-proteins of Akt, S6K1, S6, AMPK, and ACC were purchased from Cell Signaling Technology (Danvers, MA). Antibodies for IRS1 were from Millipore (Billerica, MA). Mouse monoclonal IgM antibody against PIP3 was purchased from Echelon Bioscience (Salt Lake City, UT). Cyanine-3-conjugated goat anti-mouse antibody was purchased from Jackson Immunoresearch (West Grove, PA). Alexa Fluor 488 donkey anti-rabbit IgG was from Invitrogen (Carlsbad, CA). DAPI and trisacryl protein A beads were purchased from Pierce (Rockford, IL). PI(4,5)P₂ was from Sigmaaldrich (St. Louis, MO). ³²P-γ-ATP was from PerkinElmer (Fremont, CA).

3T3-F442a and 3T3-L1 fibroblasts were grown and differentiated as described previously 33 . Cells were used 8 to 12 days after differentiation 34 . C4-2 prostate cancer cells were purchased from American Type Culture Collection and cultured in RPMI1640 containing 10% fetal bovine serum in 5% CO₂ incubator.

The cDNA for dominant negative mutant of human AMPK α1 catalytic subunit was kindly provided by Dr. Carling 23 and cloned by PCR into a lentiviral vector where a flag epitope was added to the amino terminus of the α1 mutant. Lentivirus was prepared and the cells infected as described previously 35 36 . The cDNA for PTEN (C124S) in pSG5L (Addgene, Cambridge, MA) was digested with BamHI and EcoRI and subcloned into the same Lentiviral vector as used for AMPKα1 (TET-OFF) 35 36 .

Cell extracts were separated by SDS-PAGE and electrophoretically transferred to immobilon, as described previously 34 . The membranes were blocked and sequentially incubated with specific primary antibodies and second antibodies conjugated with horseradish peroxidase. Immunoreactive bands were visualized by ECL.

3T3-F442a cells were cultured on coverslips pretreated with poly-L-lysine. Immunostaining was carried out, as previously described 34 . In brief, the cells were fixed in 4% paraformaldehyde in PBS and then washed with PBS. For PIP3 staining, after blocked with 10% normal goat serum (NGS) for 30 min, samples were incubated with mouse anti-PIP3 monoclonal antibody at 1:100 dilution for 2 h and non-immune mouse IgM was used as a negative control. After washing 3 times with PBS, the samples were incubated with Cyanine-3-conjugated goat anti-mouse antibody at 1:500 dilution for 1 h. All antibodies were diluted in PBS containing 2% NGS. After immunofluorescent staining, the cells were also counterstained with DAPI (0.5 μg/ml for 5 min) to detect the nuclei and the coverslips were mounted in spectrometric grade glycerol and sealed with nail polish. Fluorescent images were taken under Deltavision microscope (Applied Precision, Issaquah, Washington).

The assay was performed according to a protocol previously described 34 . Briefly, cell extracts were immunoprecipitated with antibodies against IRS1 preimmobilzed in trisacryl protein A beads. The immunocomplex was incubated with PI(4,5)P₂ and ³²P-γ-ATP. The labeled products were extracted, separated by thin-layer chromatography and exposed to X-Ray film. The PI(3,4,5)P₃ spots were excised and measured by scintillation counting.