Activation of Protein Kinase Cζ by Peroxynitrite Regulates LKB1-dependent AMP-activated Protein Kinase in Cultured Endothelial Cells*

We previously reported the phosphoinositide 3-kinase-dependent activation of the 5′-AMP-activated kinase (AMPK) by peroxynitrite (ONOO-) and hypoxia-reoxygenation in cultured endothelial cells. Here we show the molecular mechanism of activation of this pathway. Exposure of bovine aortic endothelial cells to ONOO- significantly increased the phosphorylation of both Thr172 of AMPK and Ser1179 of endothelial nitric-oxide synthase, a known downstream enzyme of AMPK. In addition, activation of AMPK by ONOO- was accompanied by increased phosphorylation of protein kinase Cζ (PKCζ) (Thr410/403) and translocation of cytosolic PKCζ into the membrane. Further, inhibition of PKCζ abrogated ONOO--induced AMPK-Thr172 phosphorylation as that of endothelial nitric-oxide synthase. Furthermore, overexpression of a constitutively active PKCζ mutant enhanced the phosphorylation of AMPK-Thr172, suggesting that PKCζ is upstream of AMPK activation. In contrast, ONOO- activated PKCζ in LKB1-deficient HeLa-S3 but affected neither AMPK-Thr172 nor AMPK activity. These data suggest that LKB1 is required for PKCζ-enhanced AMPK activation. In vitro, recombinant PKCζ phosphorylated LKB1 at Ser428, resulting in phosphorylation of AMPK at Thr172. Further, direct mutation of Ser428 of LKB1 into alanine, like the kinase-inactive LKB1 mutant, abolished ONOO--induced AMPK activation. In several cell types originating from human, rat, and mouse, inhibition of PKCζ significantly attenuated the phosphorylation of both LKB1-Ser428 and AMPK-Thr172 that were enhanced by ONOO-. Taken together, we conclude that PKCζ can regulate AMPK activity by increasing the Ser428 phosphorylation of LKB1, resulting in association of LKB1 with AMPK and consequent AMPK Thr172 phosphorylation by LKB1.

activity is stimulated by an increase in intracellular AMP-to-ATP ratio in response to stresses such as exercise (4 -6), hypoxia (7,8), oxidant stress (9,10), and glucose deprivation (11). AMPK activation switches on catabolic pathways that produce ATP and switches off anabolic pathways that consume ATP. The activation of AMPK leads to phosphorylation of a number of proteins that result in increased glucose uptake and metabolism as well as fatty acid oxidation and simultaneously in inhibition of hepatic lipogenesis, cholesterol synthesis, and glucose production (reviewed in Refs. [12][13][14]. AMPK is also responsible for increased fatty acid oxidation in response to the adipocyte-derived hormones leptin (15) and adiponectin (16). Because AMPK activation could have beneficial metabolic consequences for diabetic patients, AMPK has emerged as a potential target for the treatment of obesity and type II diabetes (reviewed in Refs. 3 and 17). It has been demonstrated that two classes of anti-diabetic drugs, metformin (18,19) and thiazolidinediones (20), can act at least in part through activation of AMPK in liver and muscle.
AMPK is an obligatory heterotrimer containing catalytic ␣ subunit and regulatory ␤ and ␥ subunits, each of which occur in at least two isoforms. Activation of AMPK absolutely requires its phosphorylation at Thr 172 in the activation loop of ␣1 and ␣2 subunits by one or more upstream kinases (AMPKKs) (21,22).
The major breakthrough in identifying the first AMPKK came from research on the regulation of the AMPK ortholog Snf-1 in Saccharomyces cerevisiae (23,24). The T-loop residue of Snf-1 was phosphorylated by a group of three related protein kinases bearing homology to mammalian LKB1, which was subsequently identified by several laboratories as being the major upstream kinase for AMPK (25)(26)(27). LKB1 was found to co-purify with liver AMPK and to phosphorylate recombinant AMPK complexes. In addition, AMPK could not be activated in mammalian cells that lacked LKB1 expression or in cells that were treated with Hsp90 inhibitors, which decrease LKB1 expression (28,29). Finally, LKB1 turned out to phosphorylate the T-loop of all the 12 human kinases that are phylogenetically related to AMPK (AMPK subfamily) (30). However, paradoxically, neither the activity of LKB1 itself nor that of AMPK-related kinases was regulated directly by the stimuli known to activate AMPK, like e.g. AMP, AICAR, or muscle contraction (31,32). Thus, the question remained regarding how AMPK stimuli can lead to LKB1-dependent AMPK activation.
We had previously reported that peroxynitrite (ONOO Ϫ ), a potent oxidant formed by the combination of superoxide anions (O 2 . ) and NO at a diffusion-controlled rate, activated AMPK independent of the cellular AMP/ATP ratios (19,33,34). We further demonstrated that short periods of hypoxia-reoxygenation (H/R)-activated AMPK in a ONOO Ϫdependent manner, which was also independent of the AMP/ATP ratios, but was sensitive to PI 3-kinase inhibition with either pharmacological inhibitors or overexpression of PDK1 dominant negative mutants (PDK1-DN) (34). PI 3-kinase-dependent AMPK activation has since been observed by others in insulin-stimulated platelets (35) and adiponectin-stimulated endothelial cells (36). Paradoxically, activation of PI 3-kinase with insulin or growth factors that stimulate the PI 3-kinase/PDK1 pathway either did not affect AMPK in most cell types (37) or rather caused AMPK inhibition in some cases (34,38). Therefore, PI 3-kinase/PDK1 will very likely not serve as a direct upstream kinase for LKB1, and the mechanism for AMPK activation remains to be established.
In the present study, we have established a central role of atypical protein kinase C (PKC), a protein kinase of the AGC family, as a key regulator in LKB1-dependent AMPK activation. We found that inhibition of PKC with pharmacological and genetic inhibitors effectively blocked AMPK activation caused by ONOO Ϫ . This AMPK activation pathway was LKB1-dependent, involved LKB1 phosphorylation at Ser 428 within the C-terminal part of LKB1, and led to the association of LKB1 with AMPK. The central role of PKC in the LKB1-AMPK axis is further supported by the fact that inhibition of PKC with either pharmacological (PKC-PS) or genetic (PKC-dominant negative mutants, PKC-DN) inhibitors blunted AMPK activation caused by (ONOO Ϫ ). Finally, in vitro, recombinant PKC phosphorylated both LKB1 at Ser 428 . We conclude that PKC-dependent and LKB1-mediated AMPK activation might play important roles in regulating not only cellular energy metabolism but also signaling pathways that control cell growth, differentiation, and survival.

Materials
Bovine aortic endothelial cells (BAEC) and cell culture media were obtained from Clonetics Inc. (Walkersville, MD). BAEC were maintained in endothelial basal medium with 2% serum and growth factors before use. HeLa-S3 cells were obtained from ATCC (Manassas, VA) and were grown in Dulbecco's modified Eagle's medium supplemented with 10% serum. All culture media were added with penicillin (100 units/ml) and streptomycin (100 g/ml) (32

Methods
Cell Culture and Adenoviral Infection-To generate the adenovial vector expressing a constitutively active mutant of AMPK ␣1 (AMPK-CA), a rat cDNA encoding residues 1-312 of AMPK␣1 and bearing a mutation of Thr 172 into aspartic acid (T172D) was subcloned into a shuttle vector (p-shuttle CMV). The c-Myc epitope tag was fused in frame to the 5Ј-terminus of the coding sequence. The resulting plasmid was linearized by digesting with PmeI and co-transfected into Escherichia coli BJ5183 with the adenoviral backbone plasmid, pAdEasy-1. Homogenous recombinants were selected with kanamycin. The linearized recombinant plasmid was infected into transformed human embryonic kidney 293 cells. Recombinant adenoviruses were amplified on 293 cells and purified by two ultracentrifugation steps on cesium chloride gradients. The number of viral particles was assessed by measurement of the optical density at 260 nm.
BAEC were infected with adenovirus expressing a constitutively active PKC mutant (PKC-CA), a dominant negative mutant PKC (PKC-DN) (39,40), or adenovirus coding constitutively active AMPK (AMPK-CA) (33,34). A replication-defective adenoviral vector expressing green fluorescence protein (Ad-GFP) was used as control. BAEC were infected with the adenoviruses with a multiplicity of infection of at least 10 in medium deprived of serum overnight. The cells were then washed and incubated in fresh endothelial base medium without serum for an additional 18 -24 h prior to experimentation. Under these conditions, infection efficiency was typically Ͼ80% as determined by GFP expression.
Hypoxia-Reoxygenation of BAEC-H/R was performed as described previously (34). Briefly, BAEC were cultured in 6-well plates. The cells were first infected with indicated adenoviruses for 2 days if required. The cells were placed in a water bath (37°C, total volume of 2 liters) filled with 1 liter of prewarmed Kreb-Ringer's buffer, gassed with 95% O 2 , 5% CO 2 . After 30 min of incubation, the oxygen tension was reduced abruptly from 95% O 2 , 5% CO 2 to 95% N 2 , 5% CO 2 and was maintained for 15 min as indicated. After this phase of hypoxia, 95% O 2 , 5% CO 2 was resumed (reoxygenation) for 15 min. After that the cells were washed with phosphate-buffered saline buffer twice and collected for Western blot and immunoprecipitation assays. The control BAEC was gassed only with 95% O 2 , 5% CO 2 for equivalent periods.
Assay of AMPK Activity-AMPK activity was assayed by using the SAMS peptide, as previously described (19,33,34). The difference between the presence and absence of AMP (200 M) was calculated as the AMPK activity.
LKB1 Activity Assay-LKB1 was immunoprecipitated from untreated (control) or treated cells with an antibody against LKB1 (Santa Cruz, catalogue number Sc-5638, D-19) overnight at 4°C in the presence of protein A/G-agarose. LKB1 activity present in the immunoprecipitates was determined by its ability to activate recombinant AMPK as described previously (20).
In Vitro Kinase Assays-To determine the effects of PKC or PKC␤II on AMPK or LKB1 or both, recombinant LKB1 or recombinant AMPK ␣1␤1r1 (41) were incubated with PKC at concentrations indicated for 15 min at 37°C in the presence of [ 32 P]ATP (1 Ci) with or without AMP (200 M). The SAMS peptides (final concentration, 200 M) were added if needed. AMPK activity was calculated by counting the phosphorylated SAMS peptides in the supernatants (25 l) as described previously (33,34). The beads were added 20 l of 3ϫ sample buffer and boiled for 5 min at 95°C. The proteins were separated with 12% SDS-PAGE, and the dried gels were subjected to radioautography.
Immunoprecipitation and Western Blotting-The proteins were immunoprecipitated with specific antibodies and Western blotted into nitrocellulose membranes, and the proteins were detected by specific antibodies, as described previously (33,34).
Preparation of Membrane Fractions-Cellular cytosolic and membrane fractions were prepared as described previously (42). The membranes were then incubated with appropriate secondary antibodies and analyzed in Western blots.
Site-directed Mutagenesis of Ser 428 and Asp 194 of Human LKB1 and Plasmid Transfection-Human cDNA clone was purchased from Invitrogen (clone 3689780). Wild type LKB1 gene coding region was amplified by PCR. The PCR product was ligated into TA cloning vector pGEM-T easy (Promega). LKB1 gene was released with enzymes of EcoRI/NotI from TA cloning vector and was cloned into pCI-neo mammalian expression vector (Promega; catalogue number E184). Ser 428 of LKB1 was mutated into either alanine or aspartic acid. Aspartic acid 194 of LKB1, which is essential for maintaining LKB1 activity, was mutated into alanine. All of the site-directed mutagenesis was done by using the QuikChange kits from Stratagene according to the manufacturer's instructions. All of the mutation vectors were confirmed by DNA sequencing. Plasmid DNA was extracted in large scale by using Qiagen EndoFree plasmid maxi kit (catalogue number 12362) and were transfected to HeLa-S3 by using Lipofectamine 2000 kit from Invitrogen (catalogue number 11668-019), according to the instruction provided by the supplier. Twenty four hours after transfection, the cells were treated as indicated. In this experiment, both LacZ expression vector and untreated cells are used as control.
Quantification of Western Blot-The intensity (area ϫ density) of the individual bands on Western blots was measured by densitometry (model GS-700, Imaging Densitometer; Bio-Rad). The background was subtracted from the calculated area.
Statistical Analysis-The results were analyzed by using two-way analysis of variance. The values are expressed as the means Ϯ S.E. of the mean for n assays. A p value of Ͻ 0.05 is considered statistically significant.
PDK1 serves as an important link between PI 3-kinase and several other kinases in the so-called AGC family, consisting of protein kinase A (PKA), protein kinase G (PKG), and atypical PKC (43). To identify which of these kinases leads to AMPK activation, various selective protein kinase inhibitors were preincubated with BAEC before exposure to ONOO Ϫ . After treatment, AMPK activation was determined by monitoring both Thr 172 phosphorylation of AMPK using a specific antiphospho-antibody and AMPK activity by using [ 32 P]ATP phosphorylation of the SAMS peptide (19,34). As expected, ONOO Ϫ (50 M) significantly increased AMPK-Thr 172 phosphorylation and AMPK activity ( Fig. 1, a and b). Interestingly, inhibition of PKC with either GO6983 (1 M) or arcyriarubin A (1 M) significantly attenuated ONOO Ϫ -enhanced AMPK Thr 172 phosphorylation and AMPK activity (Fig. 1b). In contrast, inhibition of either PKA with H89 (10 M) or PKG with KT5823 (10 M) had no effect ( Fig. 1, a and b). These results suggest that ONOO Ϫ might activate AMPK via selective PKC activation. eNOS residue Ser 1179 (homologous to 1177 in the human sequence) is a substrate for several kinases including both AMPK and protein kinase B/Akt. We had previously reported that ONOO Ϫ induced AMPK-dependent phosphorylation of eNOS-Ser 1179 in BAEC (33,34). Thus, we examined the effects of protein kinase inhibitors on eNOS-Ser 1179 in BAEC. As expected, ONOO Ϫ (50 M) significantly increased the phosphorylation of eNOS-Ser 1179 . Neither H89 nor KT5823 altered ONOO Ϫ -enhanced eNOS-Ser 1179 phosphorylation, whereas GO6983 and arcyriarubin A, both of which were shown to inhibit AMPK ( Fig. 1,  a and b), significantly attenuated ONOO Ϫ -enhanced eNOS-Ser 1179 phosphorylation. These results again suggest that ONOO Ϫ might upregulate AMPK and its downstream enzyme, eNOS, by activating PKC.
Activation of Protein Kinase C by ONOO Ϫ -Because the employed PKC inhibitors cannot distinguish between these isoforms, we next determined which PKC isoforms are activated by ONOO Ϫ . Specific antibodies revealed that BAEC expressed several major isoforms of PKC such as PKC␣, ␤, and atypical PKC (data not shown). Atypical PKC FIGURE 1. Inhibition of protein kinase C attenuates ONOO ؊ -enhanced AMPK-Thr 172 phosphorylation and AMPK activity. Confluent BAEC were preincubated with vehicles (Me 2 SO) or with protein kinase inhibitors for 15 min before being exposed to ONOO Ϫ (50 M). After treatment, the cells were lysed and extracted. Both phosphorylated AMPK-Thr 172 and eNOS-Ser 1179 were detected in Western blots by using the specific antibodies, and AMPK activity was assayed by [ 32 P]ATP incorporation into the SAMS peptide, as described under "Experimental Procedures. " a, effects of protein kinase inhibitors on ONOO Ϫ -enhanced AMPK Thr 172 phosphorylation. Of note is that PKC inhibitors, but not PKA nor PKG inhibitors, attenuated ONOO Ϫ -enhanced AMPK-Thr 172 phosphorylation. The blot is representative of blots from five individual experiments. b, inhibition of PKC attenuated ONOO Ϫ enhanced AMPK activity (n ϭ 6; ࡔ , p Ͻ 0.05, control versus ONOO Ϫ treated; †, p Ͻ 0.05, ONOO Ϫ versus ONOO Ϫ plus inhibitors). c and d, effects of protein kinase inhibitors on ONOO Ϫ -enhanced eNOS-Ser 1179 phosphorylation. Of note is that PKC inhibitors, but not PKA nor PKG inhibitors, attenuated ONOO Ϫ -enhanced AMPK Thr 172 phosphorylation and eNOS phosphorylation. The blot is representative of five blots from five individual experiments (n ϭ 6; ࡔ , belongs to the PI 3-kinase/PDK1 family, and its activation is linked to phosphorylation at Thr 410/403 that can be monitored by antibodies (43)(44)(45). Exposure of BAEC to either ONOO Ϫ or ONOO Ϫ donor SIN-1 significantly increased Thr 410/403 phosphorylation without altering PKC expression (Fig. 2a). In contrast, ONOO Ϫ decreased phosphorylated PKC␣/␤ (Thr 638/641 ), an activated form of PKC␣/␤ (Fig. 2b), suggesting that ONOO Ϫ selectively activated PKC while inhibiting PKC ␣/␤. In addition, PKC inhibitors Go6983 and arcyriarubin A, but not PKA/PKG inhibitors KT5823 and H89, significantly inhibited ONOO Ϫ -induced phosphorylation of PKC (Fig. 2c), AMPK (Fig. 1, a  and b), and eNOS (Fig. 1, c and d). These results again suggested that PKC might be involved in ONOO Ϫ -triggered activation of AMPK.
Translocation of cytosolic PKC into the membrane is considered an important step for PKC activation (42). We therefore investigated whether ONOO Ϫ altered the distribution of PKC within a cell. As shown in Fig. 2d, exposure of BAEC to either chemically synthesized ONOO Ϫ or SIN-1, a donor of ONOO Ϫ , significantly increased the translocation of PKC from the cytosol into membrane, confirming the activation of PKC by ONOO Ϫ .
Pharmacological Inhibition of Protein Kinase C Attenuates ONOO Ϫ -enhanced Phosphorylation of both AMPK-Thr 172 and eNOS-Ser 1179 -In an independent approach to establish PKC as a mediator of ONOO Ϫ effects on AMPK, we used PKC-PS, a synthetic peptide that selectively inhibits PKC without affecting other PKC isoforms (46,47). As shown in Fig. 3 (a and b), PKC-PS concentrationdependently attenuated ONOO Ϫ -enhanced AMPK-Thr 172 phosphorylation as well as AMPK activity. In contrast, inhibition of PKC␤II with the corresponding pseudosubstrate peptides did not alter AMPK-Thr 172 phosphorylation or AMPK activity (Fig. 3, a and b). In parallel, PKC-PS but not PKC␤II-PS dose-dependently inhibited ONOO Ϫ -enhanced eNOS-Ser 1179 phosphorylation, a downstream target of AMPK in BAEC (Fig. 3c). Taken together, these results imply that selective inhibition of PKC attenuated the effect of ONOO Ϫ on AMPK in BAEC.
Genetic Inhibition or Activation of Protein Kinase C Alters the Phosphorylation of Both AMPK-Thr 172 and eNOS-Ser 1179 -Further direct evidence for PKC-dependent AMPK activation was obtained from genetic inhibition and overexpression of constitutively active PKC mutants. First, we tested whether expression of a dominant negative PKC mutant (PKC-DN) alters ONOO Ϫ -induced AMPK activation. As expected, overexpression of PKC-DN but not GFP prevented ONOO Ϫ -induced translocation of PKC from cytosolic fractions into the membrane (Fig. 4a) and its Thr 410/403 phosphorylation (data not shown). In addition, overexpression of PKC-DN but not GFP blunted ONOO Ϫ -enhanced phosphorylation of both AMPK-Thr 172 and eNOS-Ser 1179 (Fig. 4, b and c). Taken together, these results provide strong evidence that PKC is required for ONOO Ϫ -induced AMPK activation.
Activation of AMPK by Protein Kinase C Is LKB1-dependent-Recent studies (25)(26)(27) suggested that LKB1 (also known as STK11) acts as AMPK kinase in vitro and in cultured cells. Recombinant LKB1 purified from mammalian cells phosphorylates and activates AMPK ␣1␤1␥1 (25)(26)(27). LKB1 phosphorylates catalytically inactive mutants of AMPK on Thr 172 within the ␣ subunit, and phosphorylation of AMPK requires LKB1. To study the dependence on LKB1 of AMPK activation by ONOO Ϫ , phosphorylation of AMPK was examined in HeLa S3 cells that lack LKB1. HeLa S3 cells, which are deficient of LKB1 but express normal amounts of both PKC and AMPK (Fig. 5a) (25-27), were exposed to ONOO Ϫ (100 M). ONOO Ϫ , which significantly increased the phosphorylation of both PKC and AMPK-Thr 172 in BAEC, did not alter the phosphorylation of AMPK-Thr 172 , whereas phospho-PKC was increased in HeLa S3 (Fig. 5a). In addition, ONOO Ϫ failed to alter AMPK activity, as assayed by AMPK activity assessed by formation of [ 32 P]SAMS peptide (Fig. 5b). Taken together, these results suggest that LKB1 is required for PKC-enhanced AMPK activation in BAEC.
We next determined whether ONOO Ϫ activated AMPK by increasing LKB1 activity. LKB1 was first immunoprecipitated from BAEC and then exposed to ONOO Ϫ (50 M). LKB1 activity was assayed by incubating with recombinant AMPK␣1␤1r1 (20 g) as its substrate. As shown in Fig. 5c, exposure of LKB1 directly to ONOO Ϫ (up to 50 M) did not alter its activity (Fig. 5c). In addition, exposure of recombinant AMPK ␣1␤1r1 to ONOO Ϫ (up to 50 M) instead significantly inhibited AMPK activity. 3 These data exclude the possibility that ONOO Ϫ directly up-regulates AMPK activity in BAEC.
Activation of Protein Kinase C Promotes the Association of LKB1 and AMPK-We have shown previously that metformin activates AMPK by increased production of ONOO Ϫ , which increased the association of  LKB1 with AMPK (19). Thus, we investigated whether ONOO Ϫ may activate AMPK by increasing the interaction of AMPK and LKB1. Immunoprecipitated LKB1 was analyzed for AMPK and vice versa. As shown in Fig. 6a, immunoprecipitates from BAEC treated with ONOO Ϫ showed significantly increased co-immunoprecipitation of LKB1 and AMPK-␣.
We next determined whether PKC contributed to the increased association of LKB1 with AMPK. Although ONOO Ϫ induced association of AMPK with LKB1 as above, this response was blunted by PKC-PS (Fig. 6b), suggesting that PKC activity enhances association of LKB1 with AMPK. We next measured the amount of AMPK that was associated with LKB1. Because the SAMS peptide is a substrate for AMPK but not LKB1, the phosphorylation of [ 32 P]SAMS peptides can be used as an index of the amount of AMPK that was co-immunoprecipitated with LKB1. As shown in Fig. 6c, ONOO Ϫ significantly increased the amount of [ 32 P]ATP incorporation into the SAMS peptides, suggesting increased AMPK activity. In contrast, PKC-PS reduced AMPK activity enhanced by ONOO Ϫ (Fig. 6c). These results further indicate that ONOO Ϫ activated PKC, resulting in increased interactions of LKB1 with AMPK.
AMPK is a member of a family consisting of 12 protein kinases called the AMPK family (30). Previous studies showed that stimuli such as AICAR, phenformin, or physical exercise activated AMPK without affecting the activities of LKB1 or other kinases from the AMPK family (31,32). Because PKC activation affected LKB1 in BAEC, we analyzed whether, in addition to AMPK, other kinases from this family were activated. We analyzed MARK-3 as one family member, because commercial antibodies against this kinase are available. Co-immunoprecipitation of LKB1 with MARK-3 was slightly increased in ONOO Ϫ -treated BAEC as compared with controls (Fig. 6d). However, the association of LKB1 and MARK-3 was unchanged when PKC was inhibited with FIGURE 3. Inhibition of PKC with selective PKC pseudosubstrate abolishes ONOO ؊ -enhanced PKC activation. a and b, PKC-PS, but not PKC␤II-PS cause concentration-dependent inhibition on ONOO Ϫ -enhanced AMPK in BAEC. Control cells were exposed to 30 M of PKC-PS. Confluent BAEC were exposed to ONOO Ϫ and phosphorylated AMPK-Thr 172 was detected by the phosphospecific antibody in Western blots. The blot is representative of five blots from five independent experiments (n ϭ 5; ࡔ , p Ͻ 0.05, control versus ONOO Ϫ ; †, p Ͻ 0.05, ONOO Ϫ versus ONOO Ϫ plus PKC-PS). c, PKC-PS, but not PKC␤II-PS, concentration dependently inhibits ONOO Ϫ -enhanced eNOS phosphorylation. PKC-PS (Fig. 6d). Thus, unlike AMPK, PKC may not be involved in ONOO Ϫ -enhanced association of LKB1 with MARK-3.
Protein Kinase C Promotes LKB1 Phosphorylation at Ser 428 -PKC increased the association of AMPK and LKB1 (Fig. 6). In addition, ONOO Ϫ activated PKC in LKB1-deficient HeLa-S3 cells but without activating AMPK (Fig. 5a). These data suggested that PKC is an upstream AMPK kinase and prompt speculation that PKC might phosphorylate LKB1, resulting in increased association of LKB1 with AMPK. We next determined whether ONOO Ϫ enhanced LKB1 phosphorylation in BAEC. Immunoprecipitated LKB1 was analyzed with an antibody recognizing all phosphorylated serine or threonine residues. As shown in Fig. 7a, ONOO Ϫ significantly increased the Ser/Thr phosphorylation of LKB1. In addition, inhibition of PKC with PKC-PS significantly attenuated the effects of ONOO Ϫ on LKB1 phosphorylation, suggesting a PKC-dependent process. To further identify the site(s) of LKB1 phosphorylation, we employed several commercially available antibodies against LKB1-Ser 428 , -Ser 334 , and -Thr 189 . As depicted in Fig.  7b, ONOO Ϫ increases phosphorylation of Ser 428 in LKB1, which was significantly attenuated by PKC-PS. In contrast, ONOO Ϫ did not alter the signal of phospho-Thr 189 and might have decreased phospho-Ser 334 (Fig. 7c). Most importantly, incubation of purified recombinant LKB1 with recombinant PKC dose-dependently increased the LKB1 Ser 428 FIGURE 5. LKB1 is required for ONOO ؊ -dependent AMPK activation. a, ONOO Ϫ activates AMPK in BAEC, but not LKB1-deficient HeLa S3 cells. The blot is representative of at least three blots from at least three independent assays. b, ONOO Ϫ did not activate AMPK in LKB1-deficient HeLa-S3. Confluent HeLa-S3 cells were exposed to ONOO Ϫ , and AMPK activity was assayed as described under "Experimental Procedures" (n ϭ 5). c, ONOO Ϫ does not alter LKB1 activity. LKB1 was first immunoprecipitated from BAEC and then exposed to ONOO Ϫ . LKB1 activity was assayed by including recombinant AMPK as described under "Experimental Procedures" (n ϭ 5). FIGURE 6. ONOO ؊ via PKC enhances the co-immunoprecipitation of LKB1 and AMPK. a, ONOO Ϫ increases the association of LKB1 and AMPK. LKB1 or AMPK were immunoprecipitated (IP) from BAEC and AMPK or LKB1 detected in Western blots (WB). The blot is representative of five blots from five independent experiments. b, inhibition of PKC attenuates ONOO Ϫ -enhanced association of LKB1 with AMPK. LKB1 was immunoprecipitated and stained for AMPK in Western blots. Of note is that ONOO Ϫ increased LKB1 association with AMPK in BAEC, which was sensitive to PKC-PS. The blot is representative of five blots from five independent experiments. c, ONOO Ϫ increases LKB1-associated AMPK activity. LKB1 were immunoprecipitated as described above. AMPK activity was assayed by [ 32 P]ATP phosphorylation of the SAMS peptide, as described under "Experimental Procedures" (n ϭ 5; ࡔ , p Ͻ 0.05 control versus ONOO Ϫ ; ϩ, p Ͻ 0.05 ONOO Ϫ versus ONOO Ϫ plus PKC-PS. d, ONOO Ϫ increased the co-immunoprecipitation of LKB1 with MARK-3. The lower panel is the results of MARK-3 staining in LKB1 immunoprecipitates (n ϭ 3; ࡔ , p Ͻ 0.05). Of note is that ONOO Ϫ slightly increased the co-immunoprecipitation of LKB1 with MARK-3. However, PKC does not alter ONOO Ϫ -enhanced co-immunoprecipitation of LKB1 with MARK-3. The blot is representative of three blots from three individual experiments. phosphorylation (Fig. 7d), indicating a direct interaction of LKB1 with PKC. Collectively, these results suggest that PKC activated LKB1, likely via promoting LKB1-Ser 428 phosphorylation.

Inhibition of Protein Kinase C Blunts ONOO Ϫ -induced Phosphorylation of AMPK-Thr 172 and LKB1-Ser 428 in Nonendothelial Cells
Derived from Human, Rat, and Mouse-We next investigated whether ONOO Ϫ activated AMPK via LKB1-dependent mechanism in cells types other than BAEC. To this end, human retinal pericytes, cultured rat vascular smooth muscle cells, and mouse 3T3-L1 pre-adipocytes cells were exposed to ONOO Ϫ in the presence or absence of PKC-PS (10 M). As shown in Fig. 8a, ONOO Ϫ significantly increased the phosphorylation of both AMPK-Thr 172 and LKB1-Ser 428 in all of the cell types tested. Mostly important, inhibition of PKC with PKC-PS attenuated the effects of ONOO Ϫ on both AMPK and LKB1 in these cells (Fig. 8a), suggesting that ONOO Ϫ -activated PKC might represent a common pathway for LKB1-AMPK activation in tissues other than endothelium.
Protein Kinase C Is Implicated in AMPK Activation Caused by Hypoxia-Reoxygenation and Metformin-AMPK is activated by various stimuli including hypoxia and the antidiabetic drug metformin (12)(13)(14). We next investigated whether PKC is implicated in the AMPK activation caused by stimuli other than ONOO Ϫ . As shown in Fig. 8b, metformin (1 mM) significantly increased AMPK activity in BAEC. Interestingly, inhibition of PKC with PKC-PS concentration-dependently suppressed metformin-enhanced AMPK activation. Further, overexpression of PKC-DN but not GFP significantly blunted hypoxia-enhanced AMPK activity, whereas overexpression of PKC-CA enhanced hypoxia-induced AMPK activation (Fig. 8c). These data strongly suggest that PKC is involved in hypoxia-induced AMPK activation. Taken together, inhibition of PKC attenuates AMPK activation caused by known AMPK stimuli such as metformin and H/R. These data imply that PKC might play an important role in both AMP-dependent and AMP-independent AMPK activation.
Treatment of Phorbol 12-Myristate 13-Acetate Activates AMPK via Protein Kinase C-We next determined whether activation of PKC with pharmacological reagents such as phorbol 12-myristate 13-acetate (PMA) alters AMPK activation. Similar to ONOO Ϫ , PMA (5-10 nM) for 5 min significantly increased the phosphorylation of LKB1-Ser 428 (Fig.  8d). In parallel, PMA significantly increased the phosphorylation of both AMPK-Thr 172 and ACC-Ser 79 , which were abolished by PKC-PS ( Fig. 8e). Taken together, these data strongly support the possibility that PKC-LKB1-AMPK is not unique to ONOO Ϫ and might be a common pathway shared by other stimuli.
Mutation of Ser 428 of LKB1 with Alanine Abolishes ONOO Ϫ -induced AMPK Activation in HeLa-S3 Cells-We next investigated whether mutation of Ser 428 of LKB1 altered ONOO Ϫ -induced AMPK activation. Using direct mutagenesis techniques, Ser 428 of LKB1 was mutated into alanine (loss of function). In addition, aspartic acid 194 of LKB1, which is essential for LKB1 activity, was also mutated with alanine. These plasmids were transfected into LKB1-deficient HeLa-S3 cells. As shown in Fig. 9, ONOO Ϫ (100 M) significantly increased the phosphorylation of both AMPK-Thr 172 and ACC-Ser 79 , a downstream enzyme of AMPK in HeLa-S3 cells transfected with LKB1 wild type but not in nontransfected HeLa-S3, implying the essential role of LKB1 in the activation of AMPK by ONOO Ϫ . In addition, cells transfected with kinase-inactive mutants (LKB1 aspartic acid 194 was replaced with alanine, LKB1-D194A) also blocked the effects of ONOO Ϫ on AMPK, supporting the essential role of LKB1. Interestingly, mutation of LKB1 Ser 428 with alanine (LKB1-S428A), like the kinase-inactive mutant D194A, abolished ONOO Ϫ -enhanced phosphorylation of both AMPK-Thr 172 and ACC-Ser 79 , implying the essential of LKB1-Ser 428 in the activation of AMPK by LKB1. Similar results were also obtained from A549 and HeLa cells (data not shown).

DISCUSSION
AMPK is an obligatory heterotrimeric enzyme consisting of three subunits: ␣, ␤, and ␥. The ␣ subunit contains the catalytic site, and phosphorylation of Thr 172 in its activation loop by one or more upstream kinase (AMPKK) is absolutely required for activation (1)(2)(3). Recent works from several laboratories have demonstrated that LKB1 is one of the major upstream kinases for AMPK in both cell-free systems and mammalian cells (25)(26)(27). LKB1 is a tumor suppressor kinase and can phosphorylate the T-loop of all the members of the human AMPK family, which consists of 12 protein kinases (30). Paradoxically, neither the activity of LKB1 itself nor that of AMPK-related kinases is influenced directly by stimuli known to activate AMPK (H/R, metformin, etc.) (31)(32)(33)(34). Thus, how these stimuli lead to LKB1-dependent AMPK activation has remained unclear so far. With the evidence presented here, we propose that PKC might act as a key factor in controlling FIGURE 7. PKC increases the co-immunoprecipitation of LKB1 and AMPK by enhancing the Ser 428 phosphorylation of LKB1. a, inhibition of PKC attenuates ONOO Ϫ -enhanced LKB1 phosphorylation detected by the antibody against phospho-Ser/Thr. LKB1 was first immunoprecipitated (IP) from BAEC but Western blotted (WB) against the antibody against phosphorylated Ser/ Thr. Of note is that ONOO Ϫ enhanced the Ser/Thr phosphorylation, which was attenuated by PKC-PS. The blot is representative of three blots from three independent experiments. b, PKC-dependent LKB1-Ser 428 phosphorylation. LKB1 was first immunoprecipitated and Western blotted with antibody against phosphorylated LKB1-Ser 428 . Of note is that PKC-PS blocked ONOO Ϫ enhanced detection of LKB1-Ser 428 . The blot is representative of three blots from three independent experiments. c, ONOO Ϫ does not alter the phosphorylation of LKB1-Ser 189 or -Ser 334 . The blot is representative of five blots from five independent experiments. d, recombinant PKC increases LKB1 phosphorylation at Ser 428 in vitro. The blot is representative of three blots from three independent experiments.
LKB1-dependent AMPK activation. The key evidence may be summarized as follows. First, ONOO Ϫ increased PKC activation in cultured endothelial cells. Second, ONOO Ϫ activated AMPK in a variety of cells including human retinal pericytes, BAEC, vascular smooth muscle cell, 3T3-L1 preadipocytes. Importantly, inhibition of PKC attenuated ONOO Ϫ -enhanced AMPK activation in these cell types tested. Third, either pharmacological or genetic inhibition of PKC inhibits AMPK activation caused by known AMPK stimuli including metformin, ONOO Ϫ , and H/R. Fourth, activation of AMPK by ONOO Ϫ was dependent on LKB1. Fifth, ONOO Ϫ increased the co-immunoprecipitation of LKB1 with AMPK, which was in parallel with increased phosphorylation of LKB1-Ser 428 and which was sensitive to PKC inhibition. Sixth, in the presence of LKB1, PKC increased the phosphorylation of recombinant AMP-T172K in parallel with the phosphorylation of LKB1 at Ser 428 . Seventh, activation of PKC with PMA significantly increased the phosphorylation of AMPK-Thr 172 . Finally, mutation of Ser 428 into alanine, like the kinase-inactive mutant of LKB1, abolished ONOO Ϫenhanced AMPK activation, implying the essential role of Ser 428 phosphorylation by PKC in the regulation of AMPK. Taken together, PKC might function as one of the major kinases in regulating AMPK activation, in particular in the so-called AMP-independent pathways.
The most important finding in the present study is that PKC regulates the tumor suppressor, LKB1. The human LKB1 is a serine-threonine kinase of 433 amino acids, which contains both a kinase domain and a nuclear localization signal in its N-terminal region (28) Germline mutations of the LKB1 gene are responsible for the cancer-prone Peutz-Jeghers syndrome (28). The majority of Peutz-Jeghers syndrome missense mutations are located in the region coding for the kinase domain and result in the abolition of the enzymatic activity, thus disrupting all   HeLa-S3 cells transfected with LKB1 wild type (WT) or LKB1 were exposed to ONOO Ϫ (100 M), and AMPK activation was monitored in Western blots using the specific antibodies. The blot is representative of three to four blots obtained from three or four independent experiments.
functions attributed to LKB1. The C-terminal region of LKB1 consists of 124 residues and contains several post-translational modifications. Five phosphorylation sites have been identified; two residues are autophosphorylation sites (Thr 336 and Thr 402 ), and three others (Ser 325 , Thr 363 , and Ser 428 ) are phosphorylated by upstream kinases (28). In addition, LKB1 has been shown to undergo farnesylation at a cysteine residue located in the C-terminal region (Cys 430 in human LKB1). Possibly, the LKB1 C-terminal region serves as a regulatory domain mediating dynamic interactions with several classes of proteins and promotes subcellular targeting (44). In the present study, we have provided evidence that PKC regulates AMPK activation by phosphorylating LKB1, which results in increased association of LKB1 with AMPK. The key findings can be summarized as follows. First, AMPK is not activated by ONOO Ϫ in cells lacking LKB1 (HeLa-S3 cells) but expressing normal levels of calcium calmodulin-dependent kinase kinase (CaMKK) (48 -50). Second, ONOO Ϫ increased the co-immunoprecipitation of LKB1 with AMPK, which was sensitive to inhibition of PKC. Third, inhibition of PKC attenuated ONOO Ϫ -enhanced LKB1-Ser 428 phosphorylation as well as the association of LKB1 with AMPK. Fourth, recombinant PKC in a concentration-dependent manner caused the phosphorylation of LKB1-Ser 428 in vitro. Fifth, stimulation of BAEC with PMA activated AMPK in parallel with increased LKB1 phosphorylation at Ser 428 . Finally, mutation of Ser 428 into alanine abolished ONOO Ϫ -enhanced AMPK activation. These results suggest that Ser 428 located in the C terminus of LKB1 might play a crucial role in regulating AMPK activation. Indeed, a recent study carried out by Forcet et al. (51) suggests that naturally occurring C-terminal mutations, which neither disrupt LKB1 kinase activity nor interfere with LKB1-induced growth arrest, reduce LKB1-mediated activation of AMPK and impair downstream signaling. Inhibition of PKC abolished ONOO Ϫ -enhanced association of LKB1 with AMPK but not with MARK-3, another member of the AMPK family, suggesting that PKC-LKB1 signaling is exclusively active toward the AMPK pathway. These results are in line with the previous findings (31)(32)(33) that AMPK stimuli such as AICAR or exercise activate AMPK without altering the activities of LKB1 and other members of the AMPK family. This selective effect of PKC-LKB1 on AMPK may involve the PKC-dependent phosphorylation of LKB1-Ser 428 within the C terminus of LKB1. Because phosphorylation of none of the other known regulatory sites on LKB1 was affected by PKC and a low level of LKB1/MARK3 interaction was independent of PKC inhibition, LKB1-Ser 428 phosphorylation by PKC may be the specific step leading to LKB1-dependent AMPK activation. However, we cannot exclude the possibility that PKC might regulate the interaction of LKB1-AMPK by post-translational modifications on other sites. In addition, several other kinases such as PKA and p90RSK are reported to phosphorylate LKB1-Ser 428 (52,53). If this site turns out to be indeed essential for LKB1-AMPK activation, one may speculate that these kinases could similarly regulate LKB1-dependent AMPK activation. Further studies are warranted to improve our understanding of these signaling events.
We have used endothelial cells to establish a novel PKC-LKB1-AMPK signaling axis. In addition, we have also replicated these findings in a variety of tissues including human retinal pericytes, rat smooth muscle cells, and mouse 3T3-L1 pre-adipocytes. Further, PKC inhibition attenuates AMPK activation by a variety of known AMPK stimuli including oxidant (ONOO Ϫ ), pharmacological reagent (metformin), and H/R, implying that the PKC-LKB1-AMPK pathway is not unique to endothelium and might be a common pathway shared by other physiological stimuli in a variety of tissues. Thus, we can expect broader implications of this signaling pathway in cell biology. A number of reports have shown that PKC plays critical roles in signaling pathways that control cell growth, differentiation, and survival (28,43). Thus, the interactions of PKC with LKB1 and AMPK that we have described in this study might affect our understanding of not only vascular biology but also cancer development. LKB1 is an upstream activator of AMPK, and regulation of the AMPK pathway is believed to be directly involved in LKB1 tumor suppressor function (54,55). Consistent with this model, LKB1 has recently been identified as a negative regulator of mammalian targets of rapamycin (mTOR) signaling through the sequential stimulation of AMPK and of the TSC1/TSC2 tumor suppressor complex (56,57). Furthermore, ACC, a substrate of AMPK, controls fatty acid synthetic metabolism, which is frequently dysregulated in tumors (54). We found that phosphorylation of LKB1-Ser 428 in its C-terminal part increased association of LKB1 with AMPK as well as activation of its downstream pathways, thus reinforcing the idea that AMPK plays a role in the control of cell transformation. Indeed, C-terminal mutations in LKB1, which decrease LKB1-mediated AMPK activation, compromise the ability of LKB1 to establish and maintain polarity of both intestinal epithelial cells and migrating astrocytes. Our findings further support the notion that LKB1 tumor suppressor activity depends, at least in part, on the regulation of AMPK signaling and downstream effects on cell polarization.
To summarize, the major finding of the present study is that PKC promotes AMPK activation by increasing phosphorylation of and association with the AMPK kinase LKB1. This is an important finding and might explain earlier paradoxical results where several AMPK-activating stimuli like hypoxia, muscle contraction, phenformin, or AICAR failed to activate LKB1 or a group of AMPK-related kinases that are downstream of LKB1. Our results might help explain why contraction, phenformin, and AICAR are not directly stimulating LKB1 activity, although they activate AMPK. We have also provided evidence that PKC-dependent regulation via LKB1/AMPK association depends on the phosphorylation of LKB1-Ser 428 by PKC. We conclude that PKCdependent LKB1-mediated AMPK activation might play important roles in regulating not only cell energy metabolism but also signaling pathways that control cell growth, differentiation, and survival.