Adiponectin Stimulates Production of Nitric Oxide in Vascular Endothelial Cells*

Adiponectin is secreted by adipose cells and mimics many metabolic actions of insulin. However, mechanisms by which adiponectin acts are poorly understood. The vascular action of insulin to stimulate endothelial production of nitric oxide (NO), leading to vasodilation and increased blood flow is an important component of insulin-stimulated whole body glucose utilization. Therefore, we hypothesized that adiponectin may also stimulate production of NO in endothelium. Bovine aortic endothelial cells in primary culture loaded with the NO-specific fluorescent dye 4,5-diaminofluorescein diacetate (DAF-2 DA) were treated with lysophosphatidic acid (LPA) (a calcium-releasing agonist) or adiponectin (10 μg/ml bacterially produced full-length adiponectin). LPA treatment increased production of NO by ∼4-fold. Interestingly, adiponectin treatment significantly increased production of NO by ∼3-fold. Preincubation of cells with wortmannin (phosphatidylinositol 3-kinase inhibitor) blocked only adiponectin- but not LPA-mediated production of NO. Using phospho-specific antibodies, we observed that either adiponectin or insulin treatment (but not LPA treatment) caused phosphorylation of both Akt at Ser473 and endothelial nitric-oxide synthase (eNOS) at Ser1179 that was inhibitable by wortmannin. We next transfected bovine aortic endothelial cells with dominant-inhibitory mutants of Akt (Akt-AAA) or AMP-activated protein kinase (AMPK) (AMPKK45R). Neither mutant affected production of NO in response to LPA treatment. Importantly, only AMPKK45R, but not Akt-AAA, caused a significant partial inhibition of NO production in response to adiponectin. Moreover, AMPK-K45R inhibited phosphorylation of eNOS at Ser1179 in response to adiponectin but not in response to insulin. We conclude that adiponectin has novel vascular actions to directly stimulate production of NO in endothelial cells using phosphatidylinositol 3-kinase-dependent pathways involving phosphorylation of eNOS at Ser1179 by AMPK. Thus, the effects of adiponectin to augment metabolic actions of insulin in vivo may be due, in part, to vasodilator actions of adiponectin.

Adiponectin is one of a number of proteins secreted by adipose cells (e.g. tumor necrosis factor-␣, IL-6, leptin, resistin) that may couple regulation of insulin sensitivity with energy metabolism and serve to link obesity with insulin resistance (1)(2)(3)(4)(5). In humans, plasma levels of adiponectin are negatively correlated with adiposity (6). In addition, decreased plasma adiponectin levels are observed in patients with diabetes and their relatives (7,8), as well as in subjects with coronary artery disease and macroangiopathy (9,10). Adiponectin may augment and mimic metabolic actions of insulin by increasing fatty acid oxidation (11) and insulin-mediated glucose disposal in skeletal muscle (12,13), as well as decreasing hepatic glucose output (14). In adiponectin knock-out mice, diet-induced insulin resistance is associated with increased plasma levels of tumor necrosis factor-␣, increased ␤-oxidation of glucose in muscle and liver, and delayed clearance of free fatty acids in plasma (12,(15)(16)(17). Moreover, human mutations in the adiponectin gene resulting in abnormal adiponectin secretion have been associated with the metabolic syndrome and diabetes (18 -20). Thus, decreased levels of adiponectin may play a key role in the development of insulin resistance. In addition to its metabolic actions, adiponectin also possesses anti-atherogenic properties. For example, adiponectin treatment reduces tumor necrosis factor-␣-mediated expression of adhesion molecules in endothelial cells and decreases cytokine production from macrophages (21,22). Importantly, adiponectin knock-out mice have a significant increase in vascular neointimal formation, suggesting that adiponectin may exert a protective role in vascular homeostasis (16,23). Specific receptors for adiponectin have just recently been identified (24), and at least some of the biological actions of adiponectin are mediated through activation of AMPK 1 (13,25).
Insulin has important vascular actions to stimulate production of nitric oxide (NO) in endothelium, leading to increased blood flow that contributes significantly to insulin-mediated glucose uptake (26,27). Insulin signaling pathways in vascular endothelium regulating production of NO share striking similarities with metabolic insulin signaling pathways in skeletal muscle and adipose tissue (27)(28)(29)(30)(31). Therefore, we hypothesized that adiponectin may exert some of its insulinomimetic actions by stimulating phosphorylation and activation of eNOS in vascular endothelium, resulting in increased production of NO. Demonstrating a novel role for adiponectin in eNOS activation may be helpful for explaining both metabolic and anti-athero-genic properties of adiponectin. This may also give insight into the molecular basis of the relationships among insulin resistance, obesity, atherosclerosis, and other vascular complications of diabetes.

MATERIALS AND METHODS
Purification of Recombinant Adiponectin-Recombinant full-length human adiponectin protein was produced in bacteria and purified as described previously (6).
Cell Culture and Transfection-Bovine aortic endothelial cells (BAEC) in primary culture (Cell Applications; San Diego, CA) were grown in EGM-2 as described (29) and used between passages 3 and 5. Transient transfections were performed using LipofectAMINE Plus (Invitrogen) according to the manufacturer's protocol. For immunoblotting experiments, BAEC were serum-starved overnight with EBM-A (red phenol-free endothelial basal medium from Clonetics Corp. supplemented with 1% platelet-deprived horse serum (Sigma)) prior to initiation of experiments. For measurement of NO production, BAEC were serum-starved for 2 h in EBM-A medium supplemented with 1% platelet-deprived horse serum. NIH-3T3 IR cells (NIH-3T3 fibroblasts stably transfected with human insulin receptors) were cultured as described (32).
Measurement of NO Production in Living Cells-Production of NO was assessed using the NO-specific fluorescent dye 4,5-diaminofluorescein diacetate (DAF-2 DA, Calbiochem) as described (29). Briefly, BAEC grown at 95% confluence were serum-starved for 2 h in EBM-A. L-Arginine (100 M) was added 1 h prior to each study. Cells were loaded with DAF-2 DA (final concentration 5 M, 20 min, 37°C) and then rinsed three times, kept in the dark, and maintained at 37°C with a warming stage (Bioptechs, Inc.) on a Zeiss Axiovert S100 TV inverted microscope (Carl Zeiss Inc.; Thornwood, NY). Cells were then treated sequentially with lysophosphatidic acid (LPA, 5 M) or insulin (250 nM) and adiponectin (10 g/ml). In some experiments, wortmannin (100 nM) was added 30 min before loading with DAF-2 DA. In other experiments, BAEC were co-transfected first with RFP and either Akt-AAA or AMPK-K45R. Green fluorescence intensity was quantified using IP Labs software (Scanalytics Inc.; Fairfax, VA). Data for each experiment were normalized to a reference image of the basal state.
Statistics-Paired t tests were used where appropriate. For comparison between various time courses of NO production, multiple analysis of variance (MANOVA) was employed. p values less than 0.05 were considered to represent statistical significance.

Adiponectin-stimulated Production of NO in BAEC Requires
PI 3-Kinase-To determine whether adiponectin can stimulate production of NO in vascular endothelial cells, we employed our previously established method using the NO-specific fluorescent dye DAF-2 DA to assess NO production in BAEC in pri-mary culture (29,30). The classical mechanism for activation of eNOS involves increased levels of intracellular calcium. Therefore, we used LPA (a phospholipid growth factor that stimulates release of intracellular Ca 2ϩ ) as a positive control for the production of NO in BAEC. As reported previously (29,30), LPA treatment of BAEC caused a rapid, ϳ4-fold increase in NO production ( Fig. 1, A and B, closed circles). Interestingly, when these same cells were subsequently treated with adiponectin (10 g/ml bacterially produced adiponectin), we observed a significant ϳ3-fold increase in production of NO with a distinct time course (Fig. 1, A and B, closed triangles). Similar results were obtained when the order of LPA and adiponectin treatment was reversed (data not shown). Thus, adiponectin has novel vascular actions to acutely stimulate production of NO in vascular endothelium. Moreover, when BAEC were preincubated with the PI 3-kinase inhibitor wortmannin, the production of NO in response to LPA was unaffected, but the action of adiponectin to stimulate NO was completely blocked (Fig. 1C). Therefore, similar to insulin (27-29), the ability of adiponectin FIG. 1. Adiponectin-stimulated production of NO in BAEC is dependent on PI 3-kinase activity. BAEC were serum-starved and loaded with DAF-2 DA as described under "Materials and Methods" followed by sequential treatment with LPA (5 M, t ϭ 0 min) and adiponectin (10 g/ml, t ϭ 5 min). In A, images of phase contrast view and fluorescent view of cells emitting green light (515 nm) upon excitation at 489 nm from a representative experiment are shown for time ϭ 0, 30 s, 6 min, 7, min, and 8 min. In B, relative changes in green fluorescence intensity indicative of NO production in response to LPA and adiponectin were quantified for multiple cells and plotted as a function of time (mean Ϯ S.E. of four independent experiments is shown). In C, BAEC preincubated with wortmannin (100 nM, 30 min) were loaded with DAF-2 DA and then treated with LPA and adiponectin as in panel B. Relative changes in green fluorescence intensity in response to LPA and adiponectin were quantified and plotted as a function of time (mean Ϯ S.E. of four independent experiments is shown).
to stimulate production of NO in endothelium requires PI 3-kinase activity.
Adiponectin Phosphorylates Akt and eNOS in a PI 3-Kinasedependent Manner-The activation of eNOS in response to insulin involves a calcium-independent, phosphorylation-dependent mechanism requiring phosphorylation and activation of Akt that then directly phosphorylates eNOS at Ser 1179 , leading to activation of eNOS (29,30). Since the production of NO in response to adiponectin depends on PI 3-kinase (Fig. 1), we next inquired whether adiponectin treatment of endothelial cells results in phosphorylation of Akt and eNOS. As expected, insulin stimulated a significant increase in phosphorylation of Akt at Ser 473 and eNOS at Ser 1179 in BAEC that was blocked by pretreatment with wortmannin (Fig. 2). Interestingly, adiponectin treatment of BAEC also resulted in phosphorylation of Akt at Ser 473 and eNOS at Ser 1179 at levels that were similar to those elicited by insulin. Moreover, both Akt and eNOS phosphorylation in response to adiponectin was blocked by wortmannin pretreatment. Thus, similar to insulin (29), adi-ponectin stimulates phosphorylation of both Akt and eNOS in a PI 3-kinase-dependent manner.
Role for AMPK but Not Akt in Adiponectin-stimulated Production of NO-Both Akt and AMPK are capable of phosphorylating eNOS at Ser 1179 (33)(34)(35). Since adiponectin can stimulate both Akt (cf. Fig. 2) and AMPK (13,25,36), we used dominant-inhibitory mutants of Akt and AMPK to explore the roles of these serine kinases in production of NO in response to adiponectin in endothelial cells. We transiently co-transfected BAEC with Akt-AAA and pCIS2-RFP, loaded the cells with DAF-2 DA, and stimulated the cells with LPA and adiponectin (Fig. 3A). Transfected cells were distinguished from non-transfected cells in the same field by their expression of RFP. As demonstrated previously (29), LPA-stimulated production of NO was not affected by expression of Akt-AAA. That is, the time courses for production of NO in response to LPA in cells transfected with Akt-AAA (Fig. 3A, open circles) and untransfected cells (Fig. 3A, closed circles) from the same experimental preparation were comparable (p Ͼ 0.69). Similarly, expression of Akt-AAA in BAEC did not affect adiponectin-stimulated production of NO so that the time course for production of NO in response to adiponectin in cells transfected with Akt-AAA (Fig. 3A, open triangles) and untransfected cells (Fig. 3A, closed triangles) were comparable (p Ͼ 0.56). Thus, although Akt is phosphorylated in response to adiponectin in BAEC (Fig. 2, A and B), Akt does not appear to play a role in adiponectinstimulated production of NO in endothelial cells.
We next transiently co-transfected BAEC with AMPK-K45R and pCIS2-RFP, loaded the cells with DAF-2 DA, and stimulated the cells with insulin and adiponectin (Fig. 3B). Insulinstimulated production of NO in cells transfected with AMPK-K45R (Fig. 3B, open circles) was comparable with that in untransfected cells (Fig. 3A, closed circles; p Ͼ 0.37). Stimulating transfected and untransfected cells with LPA gave similar results (data not shown). By contrast, expression of AMPK-K45R in BAEC partially, but significantly, inhibited adiponectin-stimulated production of NO when compared with untransfected cells in the same dish (Fig. 3B, open and closed triangles, respectively; p Ͻ 0.02). Taken together, these results suggest that adiponectin-stimulated production of NO does not require Akt but depends, in part, on activation of AMPK.
Adiponectin-stimulated Phosphorylation of eNOS Is Mediated by AMPK-Adiponectin stimulates phosphorylation of eNOS at Ser 1179 (an AMPK phosphorylation site) (Fig. 2, C and  D), and production of NO in response to adiponectin depends, in part, on AMPK (Fig. 3B). Therefore, we next tested whether AMPK is necessary for the ability of adiponectin to stimulate phosphorylation of eNOS. NIH-3T3 IR cells transiently co-transfected with expression vectors for eNOS and either wild-type AMPK or AMPK-K45R were treated with adiponectin or insulin. Cell lysates from each group were immunoblotted with antibodies against phospho-eNOS S1179 , eNOS, and AMPK (Fig.  4A). Control cells transfected with an empty expression vector did not have detectable levels of endogenous eNOS but showed low levels of endogenous AMPK (Fig. 4A, lane 1). As expected, both adiponectin and insulin stimulation significantly increased phosphorylation of eNOS at Ser 1179 in cells co-transfected with eNOS and wild-type AMPK (Fig. 4, A and B, lanes  3 and 4). Interestingly, in cells co-transfected with eNOS and the dominant-inhibitory mutant AMPK-K45R, phosphorylation of eNOS in response to adiponectin was significantly inhibited (Fig. 4, lane 3 versus lane 5; p Ͻ 0.03), whereas the response to insulin was unaffected (Fig. 4, A and B were serum-starved overnight and then treated with either insulin (Ins, 100 nM, 5 min) or adiponectin (10 g/ml, 5 min). Some groups of cells were pretreated with wortmannin (Wort, 100 nM, 1 h) before stimulating with insulin or adiponectin. A, representative immunoblots obtained using anti-phospho-specific Akt S473 antibody (P-Akt S473 ) and anti-Akt antibody. In B, results of four independent experiments were quantified by scanning densitometry. Phospho-Akt results were then normalized for total Akt (mean Ϯ S.E. shown). C, representative immunoblots obtained using anti-phospho-specific eNOS S1179 antibody (P-eNOS S1179 ) and anti-eNOS antibody. In D, results of six independent experiments were quantified by scanning densitometry. Phospho-eNOS results were then normalized for total eNOS (mean Ϯ S.E. shown). eNOS and wild-type AMPK (Fig. 4, A and B, lane 7). In related experiments, we co-transfected NIH-3T3 IR cells with eNOS and wild-type AMPK and treated cells with adiponectin or insulin in the absence and presence of wortmannin (Fig. 4C). When cell lysates were immunoblotted with a phospho-specific antibody against AMPK T172 , we observed that adiponectin, but not insulin, stimulated phosphorylation of AMPK. Moreover, the phosphorylation of AMPK in response to adiponectin was in-hibited by wortmannin pretreatment. Taken together, these results provide additional support for the role of AMPK in phosphorylation and activation of eNOS in response to adiponectin in a PI 3-kinase-dependent manner.

FIG. 4. Dominant-inhibitory mutant of AMPK inhibits phosphorylation of eNOS at Ser 1179 in response to adiponectin (Adn) but not insulin (Ins). NIH-3T3 IR cells transiently co-transfected (Tx)
with eNOS and either AMPK-WT or AMPK-K45R were treated with vehicle, adiponectin (10 g/ml, 5 min), insulin (100 nM, 5 min), or 5-aminoimidazole-4-carboxamide-1-␤-D-riboside (AICAR, 2 mM, 1 h). Cell lysates were subjected to immunoblotting with antibodies against phospho-eNOS S1179 (P-eNOS S1179 ), eNOS, and AMPK. A, representative immunoblots. In B, the results of six independent experiments were quantified by scanning densitometry. Phospho-eNOS results were then normalized for total eNOS (mean Ϯ S.E. shown). Expression of AMPK-K45R significantly inhibited phosphorylation of eNOS in response to adiponectin (lane 3 versus lane 5, p Ͻ 0.03) but did not significantly affect phosphorylation of eNOS in response to insulin (lane 4 versus lane 6, p Ͼ 0.50). In C, NIH-3T3 IR cells transiently co-transfected with eNOS and AMPK-WT were treated with vehicle, adiponectin (10 g/ml, 5 min), or insulin (100 nM, 5 min) in the absence or presence of wortmannin pretreatment (100 nM for 1 h). Cell lysates were subjected to immunoblotting with antibodies against phospho-AMPK T172 (P-AMPK T172 ) or AMPK. Representative immunoblots from experiments that were repeated independently three times are shown. from adipose cells, dysregulation of adiponectin action may provide a link among insulin resistance, diabetes, obesity, atherosclerosis, and vascular complications of diabetes. Adiponectin consists of an N-terminal collagenous domain and a Cterminal globular domain. Post-translational changes including multimeric assembly may be important for its biological activity (5,(37)(38)(39). Indeed, some human mutations in adiponectin affect both multimerization and bioactivity of adiponectin (40). There is also evidence that the source of purified adiponectin may affect its bioactivity. For example, adiponectin produced in mammalian cells may multimerize differently than that produced in bacterial cells (40). Moreover, some studies have reported that the globular domain of adiponectin has greater bioactivity than the full-length protein (see Ref. 36 and references therein). The differences in bioactivity between globular and full-length adiponectin may have to do with both multimerization as well as differential binding affinities for various adiponectin receptor isoforms and the tissue-specific distribution of these receptors (24). Thus, when comparing results from different studies, it is important to note the source and properties of the adiponectin preparations used.
We have recently elucidated a complete biochemical insulin signaling pathway in vascular endothelium responsible for insulin-stimulated production of NO. This involves activation of the insulin receptor tyrosine kinase, phosphorylation of IRS-1, and subsequent activation of PI 3-kinase and PDK-1, leading to phosphorylation and activation of Akt, which then directly phosphorylates and activates eNOS, resulting in increased production of NO (27)(28)(29)(30). This phosphorylation-dependent signaling pathway for activation of eNOS is completely separable and independent from the classical calcium-dependent pathway for activation of eNOS (29). Production of NO in response to insulin leads to vasodilation and increased blood flow, which contributes significantly to insulin-mediated glucose disposal in vivo and may help to couple regulation of metabolic and hemodynamic homeostasis (for a review, see Ref. 41). Therefore, we reasoned that adiponectin may have vascular actions similar to insulin to stimulate production of NO in endothelium.
In this study, we demonstrate for the first time a novel biological action of full-length adiponectin (produced and purified from bacteria) at physiological concentrations to stimulate production of NO from vascular endothelial cells. Since we use bacterially produced full-length adiponectin, it is possible that our preparation may potentially be contaminated with endotoxin, a known inducer of iNOS. However, iNOS is not expressed in endothelial cells. Moreover, the time course of NO production in response to endotoxin or other inducers of iNOS is transcriptionally regulated and takes many hours. Finally, the level of NO produced by iNOS is several orders of magnitude larger than that produced by eNOS. Thus, it is extremely unlikely that the rapid, low level production of NO in endothelium in response to adiponectin can be explained by endotoxin contamination.
Similar to insulin (27,28), the ability of adiponectin to stimulate NO production requires PI 3-kinase activity since wortmannin pretreatment of cells completely inhibited NO production in response to adiponectin. In insulin signaling pathways in endothelium, Akt is downstream from PI 3-kinase and is capable of directly phosphorylating eNOS at Ser 1179 , resulting in its activation (29). Interestingly, adiponectin also stimulated phosphorylation of both Akt at Ser 473 and eNOS at Ser 1179 in a PI 3-kinase-dependent manner. Although adiponectin has been reported to enhance Akt phosphorylation and activity in response to insulin (12), to our knowledge, results from the present study represent the first report of adiponectin directly stimulating phosphorylation of Akt. This may be a phenome-non specific to endothelial cells as a recent study reported that treatment of rat adipose cells with the globular domain of adiponectin was unable to stimulate phosphorylation of Akt at Ser 473 (36). Consistent with a PI 3-kinase-and phosphorylationdependent mechanism for activation of eNOS, the ability of adiponectin to increase phosphorylation of eNOS at Ser 1179 was blocked by wortmannin pretreatment.
Despite the facts that adiponectin can stimulate phosphorylation of Akt in endothelial cells and that insulin requires Akt to stimulate activation of eNOS, Akt activity does not appear to be required for activation of eNOS in response to adiponectin. Expression of a dominant-inhibitory mutant of Akt (Akt-AAA) had no significant effect on adiponectin-stimulated production of NO. The conditions used for experiments in the present study were similar to those we have used in a previous study showing that expression of either Akt-K179A or Akt-AAA was sufficient to block insulin-stimulated phosphorylation and activation of eNOS in endothelial cells (29). Since AMPK has been implicated in the biological actions of adiponectin (13,25,36), and AMPK is capable of phosphorylating eNOS at Ser 1179 (35), we used a dominant-inhibitory mutant of AMPK (AMPK-K45R) to test whether AMPK mediates the effects of adiponectin on phosphorylation and activation of eNOS. Expression of this mutant inhibited adiponectin-stimulated phosphorylation of eNOS and partially, but significantly, inhibited NO production in response to adiponectin. Expression of AMPK-K45R had no effect on insulin-stimulated phosphorylation of eNOS at Ser 1179 or production of NO in response to insulin or LPA. Thus, whereas both insulin and adiponectin stimulate production of NOfromendothelialcellsbyaPI3-kinase-andphosphorylationdependent mechanism, the downstream kinases responsible for phosphorylation and activation of eNOS by these two hormones appear to be distinct. The effect of insulin is completely dependent on Akt (29) and independent of AMPK (Fig.  3B). By contrast, the effect of adiponectin is independent of Akt and partially dependent on AMPK. Two recent studies have implicated AMPK in activation of eNOS and NO production in aortic endothelium (42,43). However, AMPK is not typically thought to be regulated by PI 3-kinase. Nevertheless, in our experiments, wortmannin inhibited the phosphorylation of AMPK in response to adiponectin. Moreover, one very recent study has demonstrated that AMPK activity can be regulated in a PI 3-kinase-dependent manner and that wortmannin inhibits AMPK phosphorylation and activity in endothelial cells (43). Taken together, these data are consistent with and support our current study implicating PI 3-kinase in the activation of AMPK in endothelium. Although wortmannin completely inhibited adiponectin-stimulated phosphorylation and activation of eNOS, expression of AMPK-K45R only partially inhibited phosphorylation and activation of eNOS in response to adiponectin. Thus, it seems likely that there are other PI 3-kinase-dependent kinases, in addition to AMPK, that are responsible for phosphorylating and activating eNOS in response to adiponectin. This has important therapeutic implications as it may be possible to target regulation of multiple downstream kinases to enhance metabolic actions of insulin and retard development of atherosclerosis in insulin-resistant states such as diabetes.
Additional experiments in intact animals may provide important information related to the physiological implications of our current findings. For example, it is of interest to characterize the hemodynamic phenotype and vascular endothelial function of adiponectin knockout mice. Similarly, assessing the NO-dependent hemodynamic and vascular response to adiponectin in normal mice may reveal interesting insights. Al-though these proposed physiological studies are well beyond the scope of the present study, there are a variety of observations that suggest our present findings have physiological relevance. First, the time course and magnitude of adiponectinstimulated production of NO that we observe in endothelium is similar to that seen previously with insulin stimulation (29,30). Second, in the case of insulin, both animal and human studies have convincingly shown that insulin-stimulated production of NO in endothelium contributes significantly to both increased blood flow and insulin-mediated glucose disposal (26, 44 -46). Finally, low adiponectin levels are linked with endothelial dysfunction and decreased limb blood flow in humans (47). Thus, it seems reasonable to propose that adiponectin may also be augmenting both blood flow and glucose disposal in an NO-dependent manner similar to that of insulin.
In summary, we describe a novel vascular action of adiponectin to stimulate production of NO from endothelial cells. Thus, adiponectin mimics vascular as well as metabolic actions of insulin. The ability of adiponectin to stimulate production of NO in the vasculature may lead to vasodilation and increased blood flow that contributes to the effects of adiponectin to enhance glucose disposal. Moreover, the production of NO in the vasculature in response to adiponectin may also contribute to its anti-atherogenic properties. The fact that insulin and adiponectin regulate activation of eNOS by slightly different mechanisms suggests that therapies designed to increase adiponectin levels may be beneficial in the treatment of insulin resistance, diabetes, and its vascular complications.