Endocytic Cycling of Glucose Transporters and Insulin Resistance Due to Immunosuppressive Agents

For many investigators studying type 2 diabetes, the term “insulin resistance” has become synonymous with impaired signaling immediately downstream of the insulin receptor (IR). There is good reason for this. Early studies of impaired insulin action in diabetes identified “receptor” and “postreceptor” defects; most impairment was in the postreceptor category. Subsequent studies of normal insulin action proceeded stepwise from the IR, often using biochemical methods to elucidate signaling molecules. This body of work led to a linear conception of pathways leading from the IR to various actions, including the stimulation of glucose uptake in fat and muscle. Yet, occasional reports suggested that this notion of a stepwise linear pathway might be overly simplistic. As with metabolism itself, signal transduction involves a network of interconnected pathways (1). Moreover, the response to insulin stimulation depends, to some degree, upon the state of the responding cell. It follows that insulin resistance may involve a range of alterations, not limited to reduced signaling immediately downstream of the IR. In this issue of the JCEM, Pereira et al (2) show that immunosuppressive agents cause insulin resistance not by affecting major signaling molecules, but by accelerating the removal of GLUT4 glucose transporters from the cell surface. This unexpected finding suggests a more nuanced view of mechanisms that cause insulin resistance—including those associated with obesity and type 2 diabetes.

The clinical observation that prompted the work of Pereira et al (2) is that new-onset diabetes occurs with increased frequency after solid organ transplantation and is associated with substantially increased risks of cardiovascular disease, transplant failure, and death. The development of diabetes in this setting is attributed in part to immunosuppressive agents used to prevent transplant rejection. The authors previously showed that sirolimus (rapamycin) impairs insulin signaling, which occurs as a result of its action to inhibit the mechanistic Target Of Rapamycin kinase, mTOR (3). The present work examines cyclosporine A and tacrolimus. These drugs inhibit calcineurin, a phosphatase that links calcium signaling to the nuclear entry of transcription factors controlling T-cell activation. The authors note that although these drugs reduce insulin secretion, studies in mice and humans have also demonstrated that they reduce insulin sensitivity. Precisely how this occurs remained unknown but is addressed by the present study.

Insulin stimulates glucose uptake into adipose and muscle by causing the movement, or translocation, of GLUT4 glucose transporters from internal membranes to the cell surface (4). The most widely studied insulin signaling pathway for this action can be summarized in seven steps: IR→IRS→PI3K→Akt→AS160→Rab→GLUT4. Specifically, insulin activates the IR to cause tyrosine phosphorylation of IR substrate (IRS) proteins. This creates docking sites for phosphatidyl-3-kinase (PI3K), which produces membrane phosphoinositides to recruit the serine/threonine kinase Akt (also called PKB). At the plasma membrane, Akt is phosphorylated by PDK1 and mTORC2, which activates its kinase activity toward targets including AS160 (also called TBC1D4) and, in muscle, TBC1D1. Phosphorylation of AS160 causes the activation of specific Rab proteins, which are GTPases that direct vesicle traffic within cells. The activated Rab isoforms target GLUT4-containing vesicles to move to the cell surface. When it was identified, AS160 was hailed as the final link in a chain connecting insulin signaling with vesicle trafficking. Indeed, a genetic variant of this protein causes insulin resistance and is estimated to account for more than 10% of all type 2 diabetes in Greenland (5).

Pereira et al (2) isolated subcutaneous and omental adipose from nondiabetic donors, treated the isolated adipocytes with cyclosporine A or tacrolimus, and measured effects on glucose uptake. The drugs reduced both basal and insulin-stimulated glucose uptake; this occurred in a dose-dependent manner and was observed at drug concentrations even lower than those used therapeutically. The authors next turned their attention to insulin signaling. Although tacrolimus slightly reduced IR phosphorylation, no effect of either drug was observed on IRS→PI3K→Akt→AS160 phosphorylation or abundance, on GLUT4 abundance, or on other proteins. Yet, both drugs reduced the amount of GLUT4 present at the cell surface. To study this further, the authors used a tagged protein to detect GLUT4 specifically at the cell surface in cultured, muscle-like L6 cells. Whereas the immunosuppressant drugs reduced glucose uptake in both basal and insulin-stimulated adipocytes, only insulin-stimulated uptake was reduced in L6 cells. This may reflect fundamental differences between fat and muscle, or those between primary and cultured cells. Of note, the GLUT1 transporter is abundant in cultured cell lines and does not respond markedly to insulin, which may have masked an effect on basal GLUT4 targeting (6). Both drugs reduced cell-surface targeting of GLUT4 in the presence of insulin, which accounts for the reduction in glucose uptake. But why was cell-surface GLUT4 reduced, absent any effect on insulin signaling?

GLUT4 moves continuously to and from the cell surface. At any moment, the balance of these fluxes determines the amount of GLUT4 present in the plasma membrane and, thus, the rate of glucose uptake. In unstimulated cells, GLUT4 exocytosis to the cell surface is slow relative to endocytosis, so that only a small fraction of total GLUT4 resides in the plasma membrane. Insulin stimulates exocytosis, with little or no effect on endocytosis, to shift the balance and translocate GLUT4 (7, 8). Consistent with this literature, Pereira et al (2) observed no effect of insulin on GLUT4 endocytosis. Both cyclosporine A and tacrolimus accelerated the net removal of GLUT4 from the cell surface, suggesting increased endocytosis. This effect was observed only in insulin-stimulated cells, and it is possible that exocytosis contributed to effects the authors observed at later (10- to 20-min) timepoints in their assay. The drugs had no detectable effect on exocytosis of prelabeled transporters, yet this assay also may be confounded by ongoing endocytosis (8). Thus, it remains uncertain whether the drugs affect endocytosis itself, that is, the inward budding of vesicles from the plasma membrane. They may act on a subsequent step controlling the targeting of endocytosed GLUT4. Regardless of the precise site of action, the net internalization of GLUT4 is clearly increased. The authors' observation that the drugs affect GLUT4 targeting, without altering canonical insulin signaling, highlights the importance of understanding GLUT4's intracellular itinerary.

After endocytosis, the pathway that is taken by GLUT4 bifurcates (4). In unstimulated cells, GLUT4 is targeted to an intracellular storage compartment, in which it is sequestered while awaiting an insulin signal. Because it is efficiently trapped, GLUT4 in this compartment does not cycle to and from the plasma membrane. To release these transporters, insulin stimulates proteolytic cleavage of TUG proteins, which mobilizes sequestered GLUT4 to the plasma membrane to mediate glucose uptake (9). This response does not require PI3K→Akt, but occurs in response to insulin signaling through the TC10α GTPase. The activation of this pathway may be transient, and it is less well studied than PI3K→Akt signaling. In cells with ongoing insulin stimulation, endocytosed GLUT4 returns directly to the plasma membrane and bypasses the sequestration compartment. Which route is taken after endocytosis is determined in part by the phosphorylation state of AS160. In cells exposed to a range of insulin concentrations, as occurs physiologically, endocytosed GLUT4 may be distributed in varying proportions to the two pathways. Importantly, insulin controls not only the exocytic rate, but also the number of GLUT4 proteins that are released from the sequestration compartment and that are thus available to cycle at the plasma membrane.

The results of Pereira et al (2) are consistent with an effect on endocytosis itself, but they also raise the possibility that calcineurin inhibitors affect the intracellular targeting of endocytosed GLUT4. Such an effect could alter the number of GLUT4 proteins that are available to cycle at the cell surface. A precedent for this type of mechanism has been suggested by previous data on obesity-associated insulin resistance. In individuals with type 2 diabetes, GLUT4 accumulates in intracellular membranes that are distinct from those in which it resides in nondiabetic subjects; this occurs in the fasted state and is thus independent of insulin signaling (10–12). Insulin does not mobilize this aberrantly distributed GLUT4; in effect, the target of the insulin signal is “missing in action.” Mislocalization of GLUT4 within unstimulated cells may also account for some models of insulin resistance in cells and in mice, in which no defect is observed in signaling through IRS and Akt proteins (13).

It is well known that excess accumulation of diacylglycerols and ceramides impairs insulin signaling (14). The effect of diacylglycerols is due, at least in part, to activation of novel protein kinase C isoforms, which act on IRS proteins to inhibit insulin-stimulated phosphorylation. Ceramides also impair signaling by blocking Akt activation (15). These lipids may have other effects as well. Cellular membranes are heterogeneous and are laterally compartmentalized into lipid raft and nonraft domains. Lipid rafts are enriched in ceramides, and these membrane domains may be altered in the setting of obesity. GLUT4, AS160, and associated proteins are palmitoylated (16). This modification facilitates the partitioning of membrane proteins to lipid raft membrane domains; indeed, GLUT4 has been observed in such membranes to a limited extent (17, 18). Thus, a hypothesis is that the movement of GLUT4 through raft and nonraft membrane domains regulates its targeting within unstimulated cells and affects its availability for insulin-responsive translocation and glucose uptake.

In contrast to the insulin resistance associated with obesity, which is likely multifactorial, the insulin resistance induced by cyclosporine A and tacrolimus has a defined molecular target, calcineurin. Yet, these immunosuppressant agents likely inhibit calcineurin-mediated dephosphorylation of many substrates. Which of these may impact GLUT4 targeting is unknown. Although the effect may be indirect, the observation that cyclosporine A and tacrolimus reduced glucose uptake in primary adipocytes after only 30 minutes of treatment suggests a more direct mechanism. Further research to understand how this occurs may well shed light not only on immunosuppressant-induced insulin resistance, but possibly also on the more common form of insulin resistance that is associated with obesity.

More broadly, the work of Pereira et al (2) should stimulate anew the idea that insulin stimulates glucose uptake not by a simple linear pathway, but by acting on a complex network of cellular mechanisms that include multiple signaling and trafficking pathways. These control the arrangement of proteins and membranes within cells and ultimately determine how fat and muscle cells respond to insulin by increasing glucose uptake. The idea that insulin resistance results from a defect in insulin signaling is certainly correct, but there is an issue of semantics: impaired signaling may result from a range of alterations that occur in the absence of insulin signaling. Some of these impair early steps in the IRS-Akt pathway. Others may displace GLUT4. If GLUT4 is mistargeted, so that it cannot be mobilized by insulin, is that not a form of impaired signaling? If so, then at the very least, the report by Pereira et al (2) will broaden our perspectives on insulin action, insulin resistance, and perhaps even the pathogenesis of type 2 diabetes.