In this study, we have explored the signaling pathways by which insulin suppresses lipolysis in adipocytes, a process critical to the metabolic transition from the fasting to the fed state. There are considerable data implicating a defect in antilipolysis as a critical etiological abnormality initiating the positive amplifying circuit that characterizes insulin resistance (18
). Thus, according to this prevailing model, resistance to the suppression of lipolysis by insulin increases extracellular fatty acids and indirectly increases triglycerides, which deposit in tissue, exacerbating the insulin resistance (15
). In spite of its importance, the mechanism by which insulin antagonizes adipocyte lipid mobilization has not been established unequivocally, though an attractive model has emerged. There is experimental support for the idea that insulin activates Akt, which phosphorylates PDE3b, thus stimulating the enzyme responsible for the degradation of cAMP (14
). The data presented in this report refine and, to some degree, contradict this model, presenting two important conclusions regarding the regulation of lipolysis by insulin. First, under conditions of the submaximal stimulation of lipolysis, insulin antagonizes triglyceride hydrolysis by utilizing a mechanism independent of Akt and thus different from the generally accepted pathway referred to above. This contrasts with the requirement of Akt as an obligate intermediate in the control of most metabolic processes regulated by insulin, most notably glucose transport (16
). Second, the insulin-dependent suppression of adipocyte lipolysis occurs independently of the regulation of whole-cell PKA activity while preferentially affecting perilipin phosphorylation, probably through the spatial compartmentalization of signaling pathways. Spatial compartmentalization is a widely used strategy for conferring biological specificity, and the assembly of regulatory complexes by anchoring proteins has been characterized in regard to signaling by cyclic nucleotides (3
). However, this is the first indication of such a system for the control of lipolysis and is particularly intriguing as a novel target of insulin action.
Though insulin inhibited lipolysis at all concentrations of isoproterenol tested, the requirement for Akt depended on the degree of beta-adrenergic activation. Submaximal stimulation may more closely approximate conditions that occur within an organism during fasting and feeding. The circulating concentration of norepinephrine is approximately 2 to 10 nM during fasting (25
). In rat adipocytes, glycerol release at ~1 nM isoproterenol is equivalent to that at ~5 nM norepinephrine (42
). Therefore, assuming similar conditions in 3T3-L1 adipocytes, the concentration we used in our analyses (2 nM isoproterenol) would be a close approximation to physiological levels of catecholamine during the fasting state, though admittedly the local concentrations might be considerably higher. Nonetheless, we propose that this Akt-independent pathway is predominant under typical fasting conditions. It is likely that the difference in insulin inhibition at low versus high doses of isoproterenol derives from the nature of the intracellular sequestration of signaling proteins. For example, at higher doses of isoproterenol, the response to insulin appears to be completely Akt dependent, suggesting that a shift from compartmentalized to total cellular signaling pathways confers dependence on the control of cytosolic cAMP by PDE3b. It is likely that at high concentrations of catecholamines, cytosolic cAMP rises to sufficient levels to overwhelm signaling normally restricted to the lipid droplet, and the Akt phosphorylation of PDE3b becomes the dominant means of insulin action. Possibly this pathway comes into play during conditions of extreme stress, when adrenergic stimulation rises to higher levels than typically occur during the fast. Nevertheless, at lower, submaximal levels of stimulation, our data suggest that insulin acts mainly through an Akt-independent pathway.
Our data do not exclude a role for PDE3b in this pathway. Previous studies suggest that PDE3b is required for insulin action under certain conditions. Insulin was not able to inhibit beta-adrenergic-stimulated glycerol release in PDE3b
null mice or adipocytes isolated from these animals (12
). However, PDE3b-deficient adipocytes exhibit increased levels of glycerol release in response to beta-adrenergic stimulation, and it is likely that supraphysiological levels of cAMP can overwhelm any insulin response dependent on the reduction of PKA-catalyzed phosphorylation, whether or not PDE3b functions as a downstream signaling target. Thus, the phenotype of the PDE3b
knockout mice does not exclude a PDE3b-independent pathway in the regulation of adipocyte antilipolysis, nor do our data rule out an Akt-independent modulation of PDE3b. Potential downstream effectors of insulin other than Akt that also may depend on PI3K include atypical protein kinase C (PKC) and serum glucocorticoid kinase (SGK). PKCs have been implicated in insulin-stimulated glucose transport in adipocytes, and perhaps they have additional functional roles in lipolysis regulation (19
). The SGK family of kinases is similar in structure to Akt, is also activated by phosphoinositide-dependent kinase-1 (PDK1), and shares common substrates, such as B-raf and FKHR (10
). However, the role of SGKs in adipocyte metabolism has not been thoroughly studied.
Another pathway by which PI3K could suppress lipolysis independently of Akt is through the regulation of lipid droplet trafficking by Rab proteins. PI3 kinases have been proposed to interact with Rab proteins and have been implicated in membrane trafficking (31
). The proteomic analysis of lipid droplets has identified associated small GTP-binding proteins such as Rab5 and Rab18 (6
). In particular, Rab18 is recruited to a subset of lipid droplets in response to beta-adrenergic stimulation, although its role in regulating lipolysis currently is undetermined (38
). One possibility is that Rab proteins mediate the interaction between the lipid droplet and other membranes and thus potentially regulates lipid trafficking in the cell (35
). Thus, PI3 kinases might also act downstream of the insulin receptor to regulate lipolysis via changes in lipid droplet trafficking.
The stimulation of lipolysis is associated with the PKA-dependent phosphorylation of two critical substrates, HSL and perilipin. HSL phosphorylation in the cytosol leads to its translocation from the cytosol to the lipid droplet, where it acts mainly as a diglyceride lipase. Our data support the notion that HSL phosphorylation is not the sole determinant of lipolysis, as insulin inhibited glycerol release under conditions in which HSL remained phosphorylated at Ser660. A second lipase, ATGL, is responsible for most of the triglyceride lipase activity in adipocytes and is a rate-determining enzyme for lipolysis (24
). Although ATGL is not regulated directly by PKA phosphorylation, its activity depends on the phosphorylation state of perilipin at Ser517 (46
). The precise mechanism by which phosphorylation triggers ATGL activity is unknown, though it probably involves CGI-58, which can increase ATGL activity by 20-fold (33
). CGI-58 binds to perilipin in the basal state and is released upon beta-adrenergic stimulation, presumably allowing it to activate ATGL (21
). The PKA phosphorylation of perilipin Ser492 also is critical for lipid droplet dispersion following beta-adrenergic stimulation (37
). Other phosphorylation sites of perilipin also may be necessary for achieving maximal lipolysis (47
). The data presented herein support an essential role for perilipin phosphorylation in regulating lipolysis, as in all of the experimental manipulations it remains the best correlate of glycerol release. Taken together, these data support a model in which perilipin is the central regulatory hub for lipolytic events in the fat cell.
In conclusion, our data demonstrate a novel, noncanonical insulin signaling pathway that inhibits adipocyte lipolysis. An important implication of this work is that distinct signaling pathways downstream of insulin mediate the control of different metabolic processes, e.g., antilipolysis versus glucose transport. This makes possible in adipose tissue the development of selective insulin resistance during pathological states in which some insulin actions are preserved. Recently, evidence has accumulated for such a phenomenon in the insulin-resistant liver, where function is blunted toward glucose metabolism but preserved toward lipid metabolism (9
). Perhaps a similar state occurs within adipose tissue as well during type 2 diabetes mellitus or the metabolic syndrome. The existence of these distinct pathways will undoubtedly influence the approach to the development of treatments that target specific components of the insulin signaling pathway.