Fat Cell–Specific Ablation of Rictor in Mice Impairs Insulin-Regulated Fat Cell and Whole-Body Glucose and Lipid Metabolism

Rictor^flox/flox mice (10) were crossed with aP2-cre transgenic mice [strain, B6-Cg-Tg(Fabp4-cre)1Rev/J; The Jackson Laboratory, Bar Harbor, ME] to obtain heterozygous aP2-cre;rictor^flox/WT offspring (where WT refers to wild type) in the F1 generation. These heterozygous mice were crossed with rictor^flox/flox mice to obtain the FRic^−/− mice with genotype aP2-cre;rictor^flox/flox and their rictor^flox/flox littermates lacking aP2-Cre expression (referred to here as FRic^+/+). Age-matched FRic^+/+ and FRic^−/− mice of both sexes were studied at 3–5 months of age (young) and when they were aged >9 months (old). For all mice used, the genotypes were determined by PCR analysis of tail genomic DNA as described previously (10). Both rictor^flox/flox mice as well as aP2-cre transgenic mice used for breeding had been backcrossed for 6 and 9 generations to the C57BL/6 strain, respectively. The mice were maintained under temperature- and humidity-controlled conditions with a 12-h light/dark cycle and were allowed food (total calories from fat 17%, LM-485; Harlan-Teklad) and water ad libitum. All animal studies performed in this investigation were approved by the University of Virginia Animal Care and Use Committee.

Following an overnight fast (∼15 h), a 2-h hyperinsulinemic (insulin at 150 mU/kg body wt priming followed by 2.5 mU/kg/min)-euglycemic clamp was conducted in awake mice using [3-³H]-glucose and 2-deoxy-d-[1-¹⁴C]-glucose to assess glucose metabolism in individual tissues as described in Kim et al. (17). Other experimental protocols are described in detail in the online appendix (available at http://diabetes.diabetesjournals.org/cgi/content/full/dc09-1061/DC1).

Values given are means ± SE for the numbers of animals indicated in the figure legends. Significance was determined by unpaired two-tailed t tests. Differences were considered significant if the P value was <0.05.

FRic^−/− mice were generated by breeding rictor^flox/flox mice with mice expressing Cre recombinase under the control of the aP2 promoter. Rictor protein levels were measured in isolated fat cells, liver, and skeletal muscle from FRic^−/− and FRic^+/+ littermates. While fat cells from FRic^−/− mice showed an ∼90% reduction (P < 0.0001) in rictor protein levels when compared with FRic^+/+ fat cells, there was no change in rictor protein levels in liver and skeletal muscle of FRic^−/− mice (Fig. 1,A). This result confirmed that the aP2-Cre–mediated recombination event was restricted to fat cells. mTOR, the binding partner of rictor in mTORC2, and Akt, a known substrate of mTORC2 kinase, were expressed at similar levels in FRic^−/− and FRic^+/+ fat cells (Fig. 1 A).

To test the effect of loss of rictor on mTORC2 activity, we measured levels of Akt phosphorylation at the S473 residue in insulin-stimulated FRic^+/+ and FRic^−/− fat cells. Consistent with the data we obtained in MRic^−/− muscles (11), in FRic^−/− fat cells insulin caused only a threefold increase in Akt phosphorylation at S473, whereas in FRic^+/+ fat cells an ∼10-fold increase above the basal level was observed (Fig. 1 B). This result suggests that mTORC2 is the major kinase responsible for phosphorylating Akt at the S473 residue in response to insulin in fat cells.

Similar to what we had previously reported in MRic^−/− mice (11), the level of Akt phosphorylation at T308 (a site phosphorylated by PDK1) in insulin-stimulated FRic^−/− fat cells was comparable with that seen in FRic^+/+ fat cells (Fig. 1,B). Also, insulin-stimulated phosphorylation of the insulin receptor (IR) at the tyrosine (Y) 972 residue was normal (Fig. 1,B). When evaluating phosphorylation of Akt substrates in response to insulin, we found that in FRic^−/− and FRic^+/+ fat cells phosphorylation of glycogen synthase kinase (GSK) 3β at S9 was similar between the two genotypes (Fig. 1,B). However, phosphorylation of both FoxO3a at T32 (Fig. 1,B) and AS160 at T642 was dramatically reduced (69% reduction in insulin-stimulated AS160 T642 phosphorylation in FRic^−/− fat cells, P < 0.03) (Fig. 1 B). FoxO3a is also shown to be phosphorylated at T32 by serum- and glucocorticoid-induced protein kinase (SGK) 1, another target of mTORC2 kinase activity (18). However, in insulin-stimulated fat cells, whether mTORC2 mediates SGK1 phosphorylation is not known. Recently, mTORC2 has been reported to promote the phosphorylation of turn motif sites in protein kinase C (PKC) α, thus mediating its stabilization (19,20). In mTORC2-null FRic^−/− fat cells, we observed a decrease in PKCα levels (P < 0.001; supplemental Fig. 1).

Body composition of FRic^−/− and FRic^+/+ mice, assessed by proton–magnetic resonance spectroscopy, showed similar fat and lean mass (supplemental Fig. 2A). Also, fat cell sizes of FRic^−/− and FRic^+/+ mice were not different (supplemental Fig. 2B and C). However, as previously reported (21), parametrial fat pads (40%), pancreas (20%), kidney (24%), liver (19%), and heart (22%) were heavier in FRic^−/− mice compared with FRic^+/+ mice (supplemental Fig. 3).

Consistent with previous findings demonstrating decreased glucose transport in skeletal muscle lacking rictor (11), glucose transport was reduced in FRic^−/− fat cells. While insulin caused a three- to fourfold increase (P < 0.004) in glucose uptake above basal levels in both FRic^+/+ and FRic^−/− fat cells (Fig. 2 A), in FRic^−/− fat cells basal as well as insulin-stimulated glucose uptake were reduced by ∼75% (P < 0.0002) and ∼65% (P < 0.002), respectively, when compared with FRic^+/+ fat cells.

To elucidate the possible defect underlying the reduction in glucose uptake in FRic^−/− fat cells, we analyzed the expression of proteins known to be essential for insulin-stimulated glucose uptake. Total levels of the GLUT4 (Fig. 2,B), the insulin-regulated aminopeptidase (IRAP), a protein associated with GLUT4 (22), and myosin 1c (Myo1c), a protein involved in insulin-regulated GLUT4 translocation to the plasma membrane (23), were similar in FRic^−/− and FRic^+/+ fat cells (Fig. 2,B). However, the amount of GLUT4 in plasma membranes prepared from basal and insulin-stimulated FRic^−/− fat pads was reduced by ∼40% when compared with plasma membranes prepared from FRic^+/+ fat pads (n = 3, P < 0.01 for basal and P < 0.05 for insulin stimulated) (Fig. 2 C, upper and lower panels, and D).

When measuring serum metabolite levels, we found normal fed and slightly decreased fasting triglyceride (TAG) levels (P < 0.04) (Table 1) in FRic^−/− mice (Table 1). Glycerol levels in serum from fasted (P < 0.01) (Table 1) and fed (P < 0.02) (Table 1) male FRic^−/− mice were higher compared with FRic^+/+ mice. Similarly, female FRic^−/− mice had elevated glycerol levels (3- to 5-month-old fed mice, 12.9 ± 2.4 μg/ml in FRic^+/+ mice and 24.5 ± 4.3 μg/ml in FRic^−/− mice, n = 5–6, P = 0.057). In addition, fasting NEFA levels were increased significantly (P < 0.04) (Table 1) in FRic^−/− mice, while fed NEFA levels were unchanged (Table 1).

Since serum levels of both NEFAs and glycerol, the products of lipolysis, were elevated in FRic^−/− mice (Table 1), we examined whether FRic^−/− mice are resistant to insulin-induced inhibition of lipolysis. To test this in vivo, we determined the levels of glycerol and NEFAs in serum taken from FRic^−/− and FRic^+/+ mice before and 15 min after an intraperitoneal insulin injection (16). In FRic^+/+ mice, insulin caused an ∼50% reduction in both glycerol and NEFA levels, indicating inhibition of lipolysis by insulin (Fig. 3,A), whereas in FRic^−/− mice there was only a 13% reduction in glycerol levels (n = 8, P < 0.002 vs. FRic^+/+ mice) (Fig. 3,A). Serum NEFA levels also declined significantly less after insulin injection in FRic^−/− mice compared with FRic^+/+ mice (n = 5–8, P < 0.004) (Fig. 3,B). We next evaluated lipolysis ex vivo by determining glycerol release from isolated FRic^−/− and FRic^+/+ fat cells in the presence of either CL316243 (specific β3-adrenergic receptor agonist) alone to induce lipolysis and CL316243 together with insulin to suppress lipolysis. As shown in Fig. 3,C, FRic^−/− fat cells showed twofold higher levels of basal lipolysis (n = 4, P < 0.01) when compared with FRic^+/+ fat cells, but the levels of CL316243-stimulated lipolysis was similar between FRic^−/− and FRic^+/+ fat cells (Fig. 3,C). Insulin suppressed CL316243-induced lipolysis in FRic^+/+ fat cells by ∼50% (n = 4, P < 0.0001) (Fig. 3,C) but in FRic^−/− fat cells only by 15% (n = 4) (Fig. 3 C). These data clearly demonstrate a defective regulation of lipolysis in FRic^−/− mice.

Insulin suppresses lipolysis largely through the regulation of hormone-sensitive lipase (HSL). Phosphorylation of HSL at S563 by cAMP-dependent protein kinase (also known as protein kinase A, PKA) increases the lipolytic activity of HSL, and insulin suppresses HSL activity by inhibiting the activation of PKA (24,25). Although HSL levels were similar in isolated fat pads of both genotypes (Fig. 3,D), there was an increased basal level of phosphorylation of HSL at S563 in FRic^−/− compared with FRic^+/+ fat pads (Fig. 3,D). Furthermore, whereas an ∼30% reduction in HSL S563 phosphorylation was observed in FRic^+/+ fat pads, no repression of HSL S563 phosphorylation was found in FRic^−/− fat pads in response to insulin (n = 3) (Fig. 3,D). Isoproterenol (a β-adrenergic receptor agonist) and CL316243 are known to stimulate HSL S563 phosphorylation by inducing cAMP production and consequently activating PKA. In isolated fat pads, isoproterenol stimulated HSL S563 phosphorylation in both FRic^−/− and FRic^+/+ fat pads (supplemental Fig. 4). However, the insulin-mediated inhibition of isoproterenol-induced HSL S563 phosphorylation was seen only in FRic^+/+ fat pads (supplemental Fig. 4). Consistent with these findings, basal PKA activity in isolated FRic^−/− fat cells was 50% higher (n = 4, P < 0.01) (Fig. 3,E) than in FRic^+/+ fat cells, but CL316243 stimulated PKA activity in FRic^−/− and FRic^+/+ fat cells to a similar level. Insulin suppressed CL316243-mediated activation of PKA in FRic^+/+ fat cells by 50% (n = 4, P < 0.001) (Fig. 3,E) but had no effect in FRic^−/− fat cells (n = 4) (Fig. 3 E). This finding, together with reduced repression of lipolysis in insulin-stimulated FRic^−/− fat cells, strongly supports a role for rictor/mTORC2 in the regulation of PKA activity to suppress lipolysis in fat cells.

Fat cell function affects whole-body glucose homeostasis and insulin sensitivity (26). In FRic^−/− mice, serum insulin levels were significantly higher than in FRic^+/+ mice (for fasting, P < 0.008; for fed, P < 0.0001) (Table 1). Furthermore, young FRic^−/− mice showed impaired insulin sensitivity in insulin tolerance tests (supplemental Fig. 5C). Younger FRic^−/− mice (3–5 months old) of either sex showed slightly better glucose clearance during a glucose tolerance test when compared with FRic^+/+ controls (supplemental Fig. 5A) as previously reported (21). However, old FRic^−/− mice (>9 months old) showed significantly higher blood glucose levels after an overnight fast and a dramatic impairment in glucose clearance after an intraperitoneal glucose load (Fig. 4 A and supplemental Fig. 5B). These data indicate the presence of severe insulin resistance in young and old FRic^−/− mice.

To measure insulin sensitivity in individual tissues, we performed a 2-h hyperinsulinemic-euglycemic clamp in conscious young male FRic^−/− mice. During the clamp, the glucose infusion rate required to maintain euglycemia (Fig. 4,B), and insulin-stimulated whole-body glucose turnover rates (Fig. 4,C) were significantly lower in the FRic^−/− mice compared with FRic^+/+ mice (glucose infusion rate, 80% reduction P < 0.0001; glucose turnover rate, 49% reduction, P < 0.0001), indicating pronounced systemic insulin resistance in FRic^−/− mice. Insulin-stimulated whole-body glycolysis (Fig. 4,D) and glycogen plus lipid synthesis (Fig. 4,E) were also reduced in FRic^−/− mice when compared with FRic^+/+ mice (glycolysis, ∼44% reduction, P < 0.01; glycogen plus lipid synthesis, ∼59% reduction, P < 0.001). As shown in Fig. 4,F and G, in vivo insulin-stimulated glucose uptake in adipose tissue and skeletal muscle during the hyperinsulinemic-euglycemic clamp was also significantly reduced in FRic^−/− mice (∼50% reduction in adipose tissue, P < 0.01; ∼33% reduction in skeletal muscle, P < 0.04). In addition, insulin induced a marked suppression of hepatic glucose production (HGP) during the clamp in FRic^+/+ mice (46% reduction compared with basal levels, P < 0.001) (Fig. 4,H); however, insulin completely failed to suppress HGP in FRic^−/− mice (Fig. 4 H). These results point not only to severe insulin resistance in adipose tissue but also marked insulin resistance in skeletal muscle and liver of FRic^−/− mice.

Adipokines can affect whole-body insulin sensitivity. However, we found that serum levels of leptin, adiponectin, and resistin were similar in FRic^−/− and FRic^+/+ mice (Table 1). This result shows that changes in at least these adipokines do not play a causative role in the reduced insulin sensitivity in FRic^−/− mice.

Since fat cell–specific loss of rictor/mTORC2 decreased skeletal muscle glucose transport in response to insulin, we measured insulin-stimulated glycogen synthesis in isolated soleus and extensor digitorum longus (EDL) muscles. We found that insulin-stimulated glycogen synthesis was markedly reduced in both soleus (∼41% reduction, P < 0.0001) (Fig. 5,A) and EDL (∼27% reduction, P < 0.0004) (supplemental Fig. 6) muscles from FRic^−/− mice when compared with FRic^+/+ mice. The isolated EDL muscle from FRic^−/− mice showed normal insulin-stimulated phosphorylation of the IR at Y972. However, a marked reduction in insulin-stimulated phosphorylations of insulin receptor substrate (IRS) 1 at Y896 and of Akt at both T308 and S473 (Fig. 5,B) was observed. Interestingly, basal phosphorylation of IRS1 at S302 in isolated EDL muscles from FRic^−/− mice was slightly increased when compared with FRic^+/+ mice (∼30%, n = 4, data not shown). In addition, we found a twofold increase (P < 0.03) in levels of IRS1 phosphorylated at S302 (Fig. 5,C, top panel) and a 37% reduction (P < 0.04) in the total IRS1 (Fig. 5 C, middle panel) levels in tibialis anterior (TA) muscles from FRic^−/− mice.

Since the FRic^−/− mice showed increased lipolysis and muscle is known to take up NEFAs to store as TAG, we measured TAG levels in TA muscle homogenates. Indeed TAG levels in muscles from FRic^−/− mice were significantly increased when compared with FRic^+/+ mice (P < 0.04) (Fig. 5,D). Consistent with the increase in TAG levels, we found that lipin 1, an enzyme involved in the synthesis of TAG (27), was increased (1.4-fold, P < 0.002) in the TA muscles of FRic^−/− mice (Fig. 5 E, top panel). Thus, the loss of rictor in fat cells leads to increased lipid deposition and impaired insulin signaling in skeletal muscle.

Livers from both young and old FRic^−/− mice showed increased lipid accumulation (hepatic steatosis) when compared with age-matched FRic^+/+mice (Fig. 6,A and B). Consistent with this finding, gene expression of two important hepatic lipogenic enzymes, l-type pyruvate kinase (l-PK) and fatty acid synthase (FAS) were increased two- and threefold, respectively, in livers from old FRic^−/− mice (P < 0.05) (Fig. 6 C and D).

The loss of fat cell rictor expression in mice results in a metabolic phenotype similar to type 2 diabetes with impaired fat cell function associated with severe insulin resistance in adipose tissue, skeletal muscle, and liver. This suggests that rictor plays an important role in regulating fat cell metabolism and, consequently, whole-body glucose and lipid homeostasis and insulin sensitivity.

FRic^−/− fat cells showed impaired regulation of glucose transport and lipolysis in response to insulin. Defective insulin signaling in FRic^−/− fat cells most likely causes these metabolic abnormalities. Our analyses of insulin signaling events that lie downstream of mTORC2 and mediate metabolic responses revealed a dramatic reduction in phosphorylation of Akt at S473 and, consequently, FoxO3a at T32 and AS160 at T642. At this time, it is not clear whether FoxO3a has a role in the observed metabolic defects in FRic^−/− fat cells. However, the Akt-mediated AS160 phosphorylation at T642 and consequent inhibition of the Rab-GAP (GTPase activating protein) activity of AS160 is essential for insulin-stimulated GLUT4 translocation to the plasma membrane (28,29). Consistent with decreased phosphorylation of AS160, GLUT4 translocation to the plasma membrane in response to insulin is impaired and glucose uptake is decreased in FRic^−/− fat cells.

Rictor/mTORC2 activity has not been linked to regulation of either basal lipolysis or insulin-mediated suppression of lipolysis. However, our observations that in FRic^−/− fat cells increased phosphorylation of HSL at S563 concomitant with upregulated PKA activity and that insulin failed to inhibit these events suggests a role for mTORC2 in the regulation of lipolysis by modifying PKA and HSL activities. It is not clear how mTORC2 regulates basal PKA activity. In response to insulin, Akt directly phosphorylates and activates phosphodiesterase 3B (PDE3B) (30) increasing cAMP to AMP conversion and thereby decreasing PKA activity (31). Since PKA activates lipolysis via phosphorylation of HSL, a defect in insulin activation of PDE3B would impair downregulation of lipolysis. In fact, defective regulation of lipolysis in response to insulin is reported in PDE3B knockout mice (32). While our data supports a model whereby mTORC2 activity controls Akt-mediated activation of PDE3B by insulin, this remains to be established. Rapamycin-mediated inhibition of mTOR has implicated mTORC1 in the regulation of lipolysis (14,15); however, eliminating mTORC1 by ablating raptor in fat cells does not affect lipolysis (33). Considering that prolonged rapamycin treatment inhibits the formation of mTORC2 in some cell types (4), our studies suggest that clinical observations of elevated NEFA levels during rapamycin treatment (14,15) could be due to inhibition of mTORC2.

How do fat cell–specific genetic alterations in the insulin signaling pathway lead to impairment of whole-body glucose homeostasis? Although FRic^−/− mice show a profound reduction in glucose transport in fat cells, this defect is likely to be only partly responsible for impaired glucose homeostasis in the FRic^−/− mice. Adipose tissue contributes only ∼10% to whole-body glucose clearance (34). The major contributors to impaired glucose homeostasis in FRic^−/− mice are reduced insulin-stimulated skeletal muscle glucose transport and impaired suppression of hepatic glucose production by insulin. We propose that elevated circulating NEFAs due to upregulation of lipolysis in fat cells cause the insulin resistance in skeletal muscle and liver observed in the FRic^−/− mice.

The increased serum NEFA levels in FRic^−/− mice are most likely responsible for the elevated intramuscular accumulation of TAG, which in turn may be mediated by increased lipin 1 levels in FRic^−/− skeletal muscles. Increased NEFAs flux into skeletal muscle leads to elevated intramuscular levels of lipid metabolites, such as fatty acyl CoA and diacylglycerol (35). These lipids induce insulin resistance in skeletal muscle by causing defects in insulin signaling (36). Previously, NEFAs have been shown to activate serine/threonine kinases such as IκB kinase-2 (37), Jun NH₂-terminal kinase (38), and protein kinase Cθ (39). All of these kinases can phosphorylate IRS1 on critical serine residues, thereby blocking phosphorylation of IRS1 by IR kinase on tyrosine sites that are required for PI3K association and activation (40). IRS1 S302 phosphorylation through IκB kinase-2 (41) or Jun NH₂-terminal kinase (42) can lead to IRS1 degradation. Since we see an increase in phosphorylation of IRS1 at S302 in FRic^−/− muscle, a decrease in IRS1 Y896 phosphorylation, as well as a reduction in total IRS1 levels, we suggest that elevated lipids are responsible for the insulin resistance observed in vivo and ex vivo in the skeletal muscles of FRic^−/− mice.

The development of hepatic steatosis in FRic^−/− mice may be due to the prolonged elevation of NEFAs and glycerol in serum, as both are used for synthesis of TAG in the liver. Furthermore, under hyperglycemic conditions, an increased glucose flux into the liver in FRic^−/− mice could compensate for skeletal muscle insulin resistance to maintain normoglycemia (43). The surplus glucose is converted to TAG and contributes to the development of hepatic steatosis. Thus, liver function also becomes affected in FRic^−/− animals after chronic exposure to elevated circulating NEFAs, glycerol, and glucose.

A different line of fat cell–specific rictor knockout mice (referred to as rictor^ad−/−) has recently been described (21). Rictor^ad−/− mice showed that visceral organs such as liver, pancreas, kidney, fat pads, spleen, and heart increased in size with concomitantly increased lean mass that was more apparent on a high-fat diet. This finding suggests that rictor/mTORC2 in fat cells is important for organismal growth (21). Our FRic^−/− mice also showed increased organ size (supplemental Fig. 3). The rictor^ad−/− mice were hyperinsulinemic but were normoglycemic in the fed and fasting state, similar to young FRic^−/− mice. Increased pancreatic β-cell mass could contribute to hyperinsulinemia in FRic^−/− mice, as suggested for the rictor^ad−/− mice. However, increased β-cell mass alone is not sufficient to induce hyperinsulinemia (44) but requires a factor that would induce insulin secretion. It is well known that insulin resistance is compensated by increased β-cell mass and insulin secretion (45). Our results from hyperinsulinemic-euglycemic clamp studies in FRic^−/− mice show severe whole-body insulin resistance including adipose tissue, skeletal muscle, and liver. While the insulin sensitivity in rictor^ad−/− mice was not directly measured, results from insulin tolerance tests suggest that these mice are also insulin resistant. Cybulski et al. (21) ascribe the insulin resistance in the rictor^ad−/−mice to decreased circulating adiponectin levels. There was no change in adiponectin levels in the FRic^−/− animals on a standard diet. The reason for this discrepancy is not clear, as both fat cell–specific rictor knockout models have very similar genetic backgrounds of 129 crossed to C57BL/6J, although the 129 substrains differ (129S6 for FRic^−/− mice and 129S1/Svlmj for rictor^ad−/− mice). However, since the genetic loss of adiponectin does not affect insulin sensitivity of standard diet–fed mice (46), it is unlikely that decreased adiponectin is the cause for decreased insulin sensitivity arising from fat cell–specific loss of rictor in mice. The elevated circulating NEFA levels that we observe in the FRic^−/− mice could not only contribute to peripheral insulin resistance but also directly stimulate insulin secretion from β-cells (data not shown) as well as increase pancreatic β-cell mass (47). Therefore, we propose that hyperinsulinemia and increased β-cell mass in FRic^−/− mice is predominantly a consequence of insulin resistance and increased NEFAs.

In conclusion, we demonstrate that mTORC2 is essential for insulin action on glucose transport and lipolysis in fat cells. Our studies further demonstrate the important role for mTORC2 in the regulation of fat cell function and control of whole-body glucose and lipid homeostasis and insulin sensitivity. Thus, mTORC2 may serve as a drug target for the treatment of obesity and type 2 diabetes.

This work was supported by National Institutes of Health Grants DK52753 and DK28312 (to J.C.L.) and DK80756 (to J.K.K.), the American Diabetes Association (1-04-RA-47 and 7-07-RA-80 to J.K.K.), and University of Massachusetts Diabetes Endocrinology Research Center Grant DK32520 (to J.K.K.).

No potential conflicts of interest relevant to this article were reported.

Parts of this study were presented at the 68th Scientific Sessions of the American Diabetes Association, San Francisco, CA, 6–10 June 2008.

We thank Drs. James Garrison, Thomas Sturgill (University of Virginia), and Sue Bodine (University of California, Davis) for helpful comments on the manuscript; Dr. Peter Gillespie (Oregon Health and Science University) for providing the Myo1c antibody; and Olesya Shcheglovitova for assistance with mouse husbandry and genotyping.