Analysis of insulin signalling by RNAi-based gene silencing

Q.L. Zhou, J.G. Park, Z.Y. Jiang, J.J. Holik, P. Mitra, S. Semiz, A. Guilherme, A.M. Powelka, X. Tang, J. Virbasius, M.P. Czech


Using siRNA-mediated gene silencing in cultured adipocytes, we have dissected the insulin-signalling pathway leading to translocation of GLUT4 glucose transporters to the plasma membrane. RNAi (RNA interference)-based depletion of components in the putative TC10 pathway (CAP, CrkII and c-Cbl plus Cbl-b) or the phospholipase Cγ pathway failed to diminish insulin signalling to GLUT4. Within the phosphoinositide 3-kinase pathway, loss of the 5′-phosphatidylinositol 3,4,5-trisphosphate phosphatase SHIP2 was also without effect, whereas depletion of the 3′-phosphatase PTEN significantly enhanced insulin action. Downstream of phosphatidylinositol 3,4,5-trisphosphate and PDK1, silencing the genes encoding the protein kinases Akt1/PKBα, or CISK(SGK3) or protein kinases Cλ/ζ had little or no effect, but loss of Akt2/PKBβ significantly attenuated GLUT4 regulation by insulin. These results show that Akt2/PKBβ is the key downstream intermediate within the phosphoinositide 3-kinase pathway linked to insulin action on GLUT4 in cultured adipocytes, whereas PTEN is a potent negative regulator of this pathway.

  • gene silencing
  • glucose
  • insulin signalling
  • phosphoinositide 3-kinase (PI3K)
  • phospholipase Cγ
  • protein kinase Cλ/ζ siRNA

Insulin signalling is a required mechanism for normal glucose homoeostasis in humans, and loss of either insulin or its receptor tyrosine kinase is lethal. The major metabolic pathways regulated by insulin include inhibition of gluconeogenesis in liver and stimulation of glucose transport in muscle and adipose tissue. Glucose disposal is mediated by the insulin-regulated glucose transporter GLUT4, which is expressed almost exclusively in these latter two tissue types (see [1,2] for reviews). Insulin stimulates the translocation of intracellular GLUT4 glucose transporters to the plasma membrane where they can catalyse glucose uptake [1,2]. Gene ablation of GLUT4 in either muscles or adipose tissue causes glucose intolerance in mice and insulin resistance in skeletal muscles [3]. These results indicate that GLUT4 function in adipose tissue is linked to a mechanism by which adipocytes communicate with muscles to regulate glucose transport and regulate whole-body glucose homoeostasis. Thus understanding the underlying mechanisms of GLUT4 regulation in adipocytes by insulin is an important objective.

In unstimulated adipocytes, GLUT4 is largely present in intracellular membranes that function in specialized recycling with slow exit kinetics. GLUT4 at the cell surface undergoes rapid endocytosis in the basal state, but is very slowly recycled back to the plasma membrane. Insulin significantly stimulates exocytosis of GLUT4-containing membranes [4], and partially inhibits GLUT4 endocytosis [5]. These dynamics enhance the steady-state levels of GLUT4 in the plasma membrane of insulin-treated adipocytes by 10–20-fold. The increase in glucose transport activity due to insulin is in the same range, indicating that GLUT4 translocation mediates most or all of the effect of insulin on this process. There may also be regulation of GLUT4 catalytic activity [6], but direct data are lacking on potential mechanisms involved. Membrane trafficking of GLUT4 involves proteins that promote membrane sorting and move GLUT4-containing vesicles along cytoskeletal tracks [7,8] as well as proteins that facilitate vesicle fusion at the plasma membrane [9]. Although some of these proteins have been discovered, our understanding of these processes is limited and it is probable that many components of the system are not yet identified.

The insulin signalling pathways required for GLUT4 regulation are also incompletely understood. Figure 1 summarizes three signalling circuits that have been implicated in this process. The PI3K (phosphoinositide 3-kinase) pathway has been by far the most studied, and is the most clearly established to be necessary for insulin signalling to GLUT4 (see [10] for a review). The insulin receptor tyrosine kinase catalyses phosphorylation of insulin receptor substrate proteins, which recruit and activate p85/p110-type PI3Ks, leading to conversion of PtdIns(4,5)P2 into PtdIns(3,4,5)P3. This 3′-polyphosphoinositide drives the activation of multiple protein kinases (e.g. atypical protein kinase Cλ/ζ isoforms, Akt/PKB isoforms, SGK1/2, SGK3/CISK and S6K) through the action of the protein kinase PDK1 (see Figure 1) and another unknown protein kinase (results not shown; see [11]). Inhibition of PI3K by wortmannin or other inhibitors blocks GLUT4 regulation by insulin [12], and gene ablation of insulin receptor substrate proteins in mice causes glucose intolerance [13]. These and other experiments have strongly supported the requirement of components in this pathway from the insulin receptor through the PDK1 step for insulin stimulation of glucose transport into fat and muscle cells.

Figure 1 Signalling pathways previously implicated in the regulation of glucose transport by insulin

In the PI3K pathway insulin receptor tyrosine kinase catalyses phosphorylation of insulin receptor substrate proteins (IRS), which recruit and activate PI3K to form PtdIns(3,4,5)P3. The 5′-phosphatase SHIP2 and the 3′-phosphatase PTEN can degrade PtdIns(3,4,5)P3 to PtdIns(3,4)P2 and PtdIns(4,5)P2 respectively. The protein kinase PDK1 can phosphorylate multiple downstream protein kinases as shown. In a second proposed pathway, insulin receptor phosphorylates c-Cbl and Cbl-b proteins, which leads to recruitment of CrkII/C3G and GTP binding to TC10. A third proposed pathway involves the activation of phospholipase Cγ leading to release of IP3 (inositol trisphosphate) and diacylglycerol and activation of protein kinase C. Results from our studies using siRNA to deplete selectively the intermediate proteins in these pathways indicate that the boxed pathway is required for GLUT4 regulation by insulin.

In contrast, there are conflicting data related to the connection between protein kinases downstream of PDK1 and GLUT4 regulation in adipocytes. The protein kinase S6K is apparently not required because its activation through mTOR (mammalian target of rapamycin) is blocked by rapamycin, which has no effect on insulin-stimulated GLUT4 translocation. Also, data arguing against the involvement of SGK1/2 are available [14]. However, the role of CISK/SGK3, an Akt-like protein kinase with a PX domain rather than a PH domain at its N-terminus [15], has not been evaluated, and available data related to atypical protein kinase C isoforms are contradictory. Using dominant inhibitory constructs of protein kinase Cλ/ζ and Akt/PKB in cultured adipocytes, evidence both for and against the role of these protein kinases was obtained [1619]. Gene ablation of Akt1/PKBα in mice is without effect on insulin signalling to glucose transport [20], whereas muscles and fat cells from Akt2/PKBβ knockout mice show attenuated insulin-mediated hexose transport only at a low concentration of insulin [21]. The responsiveness to maximal doses of insulin was normal in muscles from Akt2(−/−) mice [21].

On the basis of these equivocal data, we developed a high efficiency method to introduce siRNA into cultures of fully differentiated 3T3-L1 adipocytes, which can deplete selected proteins by 90–95% within the entire cell culture [22]. Using this method, we have specifically depleted many of the adipocyte proteins within the putative insulin signalling pathways shown in Figure 1. Adipocytes are electroporated with either a single siRNA species directed against an mRNA of interest [22] or a pool of four siRNA species directed against different regions of the same mRNA. The cells are then cultured for 24–72 h after the electroporation before assaying for protein levels and biological function (e.g. insulin signalling to hexose transport). Depletion of CISK/SGK3 for either 24 or 48 h by this method led to almost total loss of this protein kinase from the cultured adipocytes, with little decrease (<20%) in the sensitivity of these cells to insulin (Table 1).

View this table:
Table 1 Effects of siRNA-mediated gene silencing within the PI3K, TC10 and phospholipase Cγ pathways on insulin signalling to hexose transport

All data shown relate to experiments performed with 3T3-L1 adipocytes. IP3, inositol trisphosphate.

Similarly, siRNA-mediated loss of protein kinase Cλ or protein kinase Cζ, or both simultaneously, had no detectable effect on the ability of insulin to stimulate deoxyglucose transport into 3T3-L1 adipocytes (Figure 2 and Table 1). Figure 2 shows that treatment of cultured adipocytes with a pool of four siRNA species directed against protein kinase Cλ causes significant depletion of this protein kinase, as detected by an antibody that recognizes both protein kinase Cλ and protein kinase Cζ. A small amount of protein kinase Cζ is present in mouse tissues, consistent with Figure 2 (B, inset). Despite the 80% loss of protein kinase Cλ, no significant inhibition of insulin-stimulated hexose transport was observed. Knockdown of protein kinase Cζ plus protein kinase Cλ also failed to attenuate insulin action (results not shown). These results indicate that protein kinase Cλ/ζ is not required for the GLUT4 responsiveness to insulin.

Figure 2 Depletion of protein kinase Cλ by siRNA fails to attenuate insulin action on deoxyglucose transport

(A) Protein kinase Cλ domain structure and sequences in the protein kinase Cλ mRNA against which four siRNA species were prepared. (B) 3T3-L1 adipocytes were transfected with a pool of four siRNA species shown in (A) or with siRNA directed against both Akt1/PKBα and Akt2/PKBβ by methods described previously [22]. 3H-deoxyglucose transport was then assayed, and the values of three experiments shown. Only the Akt-directed siRNA inhibited insulin action with statistical significance (*P<0.05). Inset: Western blots using anti-Akt antibody that recognized both Akt1 and Akt2 or antibody that recognized both protein kinases Cλ and Cζ.

In contrast, depletion of Akt2/PKBβ protein expression in our experiments by approx. 70% caused a 60% decrease in insulin responsiveness (Table 1 and [22]). This effect was quite selective because almost complete loss of the Akt1/PKBα isoform resulted in only a small (10–20%) inhibition of insulin action on deoxyglucose transport (Table 1 and [22]). Knockdown of both Akt2/PKBβ and Akt1/PKBα simultaneously led to an 80% loss of insulin responsiveness, indicating only a small contribution of Akt1/PKBα in these cells (Figure 2 and [22]). Recent results published on cells from Akt1(−/−), Akt2(−/−) double knockout mice are consistent with our results [23]. On the basis of the above results, we conclude that Akt2/PKBβ is absolutely required for insulin regulation of hexose transport and GLUT4 translocation in response to insulin [22].

We also tested hypotheses related to the roles of SHIP2 and PTEN in insulin signalling to GLUT4. The phosphatases SHIP2 and PTEN convert PtdIns(3,4,5)P3 into PtdIns(3,4)P2 and PtdIns(4,5)P2 respectively, and have been suggested to regulate negatively insulin action on glucose transport [24,25]. SHIP2 knockout mice have also been reported to exhibit enhanced glucose tolerance and insulin sensitivity [26]. Surprisingly, RNAi (RNA interference)-mediated loss of SHIP2 in 3T3-L1 adipocytes failed to modify insulin signalling to glucose transport (Table 1) or to Akt (results not shown). In contrast, even modest depletion of PTEN expression by approx. 50% caused a greatly increased sensitivity of both targets in cultured adipocytes to insulin (Table 1). These results indicate that in cultured adipocytes, SHIP2 does not function in the pathway of insulin signalling to GLUT4, whereas PTEN is indeed a potent negative regulator of the PI3K pathway in these cells.

Additional signalling pathways have been implicated in insulin regulation of GLUT4 over the past several years. These have been proposed to potentially act in parallel with the PI3K pathway. The results supporting this general concept, from our laboratory and others, include the apparent failure of an exogenously administered PtdIns(3,4,5)P3 analogue alone to mimic insulin action on deoxyglucose transport [27], and the apparent failure of integrin-mediated Akt activation to mimic insulin signalling on hexose uptake [28]. Opposed to these observations is the finding that expression of a membrane-directed form of Akt/PKB can significantly activate GLUT4 translocation [16]. Nonetheless, two specific signalling elements that have been suggested to be required for optimal insulin signalling to GLUT4 are phospholipase Cγ [29] and the GTPase TC10 [30,31]. We therefore tested the role of these pathways in 3T3-Ll adipocytes using RNAi. From Affymetrix GeneChip array analysis of gene expression in these cultured adipocytes, we determined that phospholipase Cγ1 is the isoform of this family that is expressed in these cells. RNAi-mediated depletion of this protein resulted in approx. 90% loss of abundance, but no detectable decrease in insulin responsiveness of deoxyglucose transport was observed (Table 1). These results suggest the phospholipase Cγ pathway is not required for GLUT4 regulation by insulin in 3T3-L1 adipocytes.

Similarly we tested the role of the TC10 pathway in 3T3-L1 adipocytes. This putative signalling system is proposed to involve the recruitment of CAP and APS proteins to the insulin receptor, which then attract Cbl proteins that are tyrosine-phosphorylated by the receptor tyrosine kinase [30,31]. This in turn recruits CrkII through its SH2 domain, which interacts with C3G, a proposed exchange factor for TC10 [30,31]. Two isoforms of Cbl which can bind CrkII exist, c-Cbl and Cbl-b, and we found that we could virtually and completely block expression of these proteins, individually or in combination, by RNAi treatment of 3T3-L1 adipocytes [32]. However, loss of c-Cbl plus Cbl-b proteins failed to attenuate detectably insulin signalling to stimulate hexose transport (Table 1) or GLUT4 translocation [32]. Similarly, depletion of CAP or CrkII proteins by 90% or more had no effect on insulin responsiveness of deoxyglucose transport in 3T3-L1 adipocytes (Table 1 and [32]). These results indicate that Cbl, CAP and CrkII proteins are not required for insulin signalling to GLUT4, suggesting that the TC10 pathway may not be physiologically linked to hexose transport regulation in adipocytes. This same conclusion has been recently published in relation to muscle cells [33].

The utility of RNAi-based gene silencing in fully differentiated adipocytes can also be extended to identifying novel components that function in insulin signalling or other metabolic pathways. Our laboratory has miniaturized the method for transfection of adipocytes with siRNA such that 96-well plates can be used for functional assays in a relatively high throughput format. With this technology, approx. 30–50 selected genes per week can be silenced and studied for effects on insulin signalling or other processes in 3T3-L1 adipocytes by a small group of researchers. This screen has been used to scan through large numbers of genes that are highly up-regulated in cultured adipocytes. One highly expressed gene that we unexpectedly found to function in these cells is the transcriptional co-repressor Rip140, previously found to control fertility [34]. Depletion of this protein with RNAi leads to markedly enhanced insulin-stimulated deoxyglucose transport in 3T3-L1 adipocytes, presumably due to de-repression of one or more nuclear receptors (A.M. Powelka, J.V. Virbasius, A. Guilherme, S.M.C. Nicoloro and M.P. Czech, unpublished work). Remarkably, recently, another laboratory independently reported that Rip140 knockout mice show enhanced energy expenditure and increased expression of the uncoupling protein UCP1 in white fat cells [35]. Our experiments in progress are designed to unravel the underlying molecular details of Rip140 function in fat cells and to identify additional gene products that regulate insulin signalling and metabolic pathways in cultured adipocytes using RNAi.


We thank J. Erickson for expert help in preparing this paper. This work was supported by the National Institutes of Health (grants DK30898, DK30648 and DK60837) and the Diabetes and Endocrinology Research Center (grant 5 P30 DK32520), and an American Diabetes Association grant to Z.Y.J. (7-03-JF-22).


  • Signalling Outwards and Inwards: A Focus Topic at BioScience2004, held at SECC Glasgow, U.K., 18–22 July 2004. Edited by J. Challiss (Leicester, U.K.), A. Harwood (University College London, U.K.), M. Humphries (Manchester, U.K.), C. Isacke (Institute of Cancer Research, London, U.K.), R. Liddington (Burnham Institute, La Jolla, CA, U.S.A.), T. Palmer (Glasgow, U.K.), K. Siddle (Cambridge, U.K.), C. Sutherland (Dundee, U.K.), H. Wallace (Aberdeen, U.K.) and M. Welham (Bath, U.K.).

Abbreviations: PI3K, phosphoinositide 3-kinase


View Abstract