Attenuated Pik3r1 Expression Prevents Insulin Resistance and Adipose Tissue Macrophage Accumulation in Diet-Induced Obese Mice

All studies were approved by the Institutional Animal Care and Use Committee at the University of Colorado School of Medicine. Six-week-old, male Pik3r1^+/− (αHZ) and Pik3r2^−/− (βKO) and their wild-type (WT) littermates were placed on an HFD (45% of calories from fat) or control diet (CON; 10% of calories from fat) from Research Diets (New Brunswick, NJ) for 16 weeks. Basal and insulin-stimulated tissue was collected as previously described (21). Blood samples were collected after a 4-h fast through the retro-orbital sinus, and plasma cytokines/adipokines were measured by multiplex assay (Bio-Rad Laboratories, Hercules, CA).

Male C57BL/6J (CD45.1) WT recipient mice (Jackson Laboratory, Bar Harbor, ME) and male C57BL/6SVJ Pik3r1^+/− (CD45.2; αHZ) recipient mice received 1,000 rads of whole-body irradiation. BM was extracted from the tibia and femur of WT PIk3r1^+/+ (CD45.2), WT (CD45.1), and αHZ (CD45.2) donor mice, and 2.5 × 10⁶ cells were injected into the retro orbital sinus cavity of irradiated mice. At 6 weeks, engraftment of donor BM was tested by flow cytometry for CD45.1 or CD45.2 antigens (anti–CD45.1-PE, anti–CD45.2-APC, and ter119-FITC; BD Pharmingen) in peripheral blood (see Supplementary Fig. 3).

The hyperinsulinemic-euglycemic (H-E) clamps were conducted using two jugular vein cannulas as previously described (22), except that clamps were conducted on overnight-fasted mice that were anesthetized with a drug cocktail (acepromazine, 10 mg/mL; midazolam, 5 mg/mL; and fetanyl, 0.05 mg/mL) as previously outlined (23). This anesthesia protocol does not alter glucose metabolism during an H-E clamp (23).

Epididymal adipose tissue (eAT) was homogenized (3 g:1 mL) in ice-cold lysis buffer as previously described (22). Samples (50 μg) were run on a 7% acrylamide Bis-Tris gel and transferred to polyvinylidene fluoride membranes (Bio-Rad Laboratories). Membranes were then exposed to the appropriate primary and secondary antibodies. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) or actin was probed as a loading control.

PI3K was immunoprecipitated from 500 μg AT or skeletal muscle (SkM) homogenates by incubating overnight with a mouse anti-pY100 antibody (Cell Signaling, Danvers, MA). The kinase reaction was initiated with the addition of 2 μg phosphatidylinositol (PI) and 20 µCi [γ-³²P]-ATP, incubated for 30 min, and then terminated by addition of 20 μL of 8 N HCL as previously described (22). The lipid products were run on a thin-layer chromatography plate, and phosphoinositol-3-phosphate [PI(3)P] was visualized on film.

eAT for immunohistochemistry was fixed in 10% neutral, phosphate-buffered formalin for 24–48 h and paraffin embedded, and 4-µm sections were stained with DAPI and a rat anti-mouse F4/80 monoclonal antibody (1:250; Abcam, Cambridge, MA).

RNA was isolated from eAT and SkM using the RNeasy Plus Mini Kit (QIAGEN, Inc., Valencia, CA). RNA was quantitated and reverse transcribed (ImPromII reverse transcriptase; Promega), and quantitative real-time PCR was run on a Roche LightCycler 480 instrument. mRNA was normalized to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) or 18S. Primer sequences are presented in Supplementary Table 3.

SkM tibialis anterior (30 mg) and liver (70 mg) were lyophilized, weighed, and homogenized in MeOH along with an internal standard of tripentadecanoic acid (Omni International, Marietta, GA). SkM lipid extraction, isolation, and analysis were performed as previously described (24).

eAT was digested in Dulbecco’s modified Eagle’s medium (DMEM) + 0.4% collagenase I, and cells were filtered and centrifuged to collect stromal vascular cells (SVCs). Cells were incubated for 10 min with Fc block, followed by a 30-min incubation with anti–F4/80-PE and anti–CD11c-APC (eBioscience, San Diego, CA). Cells were then fixed with 1% paraformaldehyde in PBS. Analysis was performed using the FACS Calibur system with Cell Quest software.

eAT was extracted under sterile conditions, and 1 g of tissue was minced, placed in a six-well tissue culture plate with serum-free DMEM (2 mL), and incubated for 48 h (37°C). Media was collected and stored at −80°C until use. Media was analyzed using RayBiotech, Inc. (Norcross, GA) Mouse Cytokine Array 3 according to the manufacturer’s directions. Arrays were visualized on Ox-Mat film and quantitated by densitometry.

eAT was digested with 1 mg/mL collagenase type I (Sigma-Aldrich) in a Krebs-Ringer digestion buffer and isolated as previously described (25). Adipocytes were removed and immediately stained with 0.2% methylene blue for cell integrity. The imaging methods were adapted from Higgins et al. (26) using an Olympus Max U-CMAD3 microscope with digital camera and analyzed using AdCount software (Mayo Clinic, Rochester, MN).

Isolated adipocytes were washed in Krebs-Ringer phosphate buffer, spun, and distributed as 100-μL packed cells for basal or insulin-stimulated conditions. Insulin-stimulated 2-deoxyglucose (2DG) uptake was measured, as previously described (22), in the presence and absence of 1.2 nM insulin.

Isolated-SkM, insulin-stimulated 2-DG uptake was measured in WT and αHZ extensor digitorum longus (EDL) as previously described (27). For tumor necrosis factor-α (TNF-α) experiments, the same protocol was used except that isolated EDL muscles were initially incubated in Krebs-Henseleit buffer for 1 h with or without 10 μg/mL TNF-α in the presence and absence of 0.3 nM insulin.

3T3-L1 preadipocytes were plated in DMEM with 1% FCS and 1 mM l-glutamate. Cells were differentiated into adipocytes by the addition of DMEM containing 10% FCS, 1 mM L-glutamate, 300 μM isobutylmethylxanthine, 1 μM dexamethasone, and 1 μg/mL insulin for 3 days. After differentiation, cells were maintained as previously described (28). For time course experiments, cells were serum starved overnight and then exposed to interleukin-6 (IL-6; 20 ng/mL) (Sigma-Aldrich) for indicated times.

Data were analyzed by two-way ANOVA for main effects of diet and genotype with Tukey post hoc analysis. Significant differences within genotype are indicated with *, and significant differences within diet group are indicated with #. Data from BMT studies were analyzed only within the HFD groups by two-way ANOVA for main effects of BM genotype and body genotype with Tukey post hoc analysis. The control group (WT/WT-CON) was used as reference to show that HFD induced a change independent of transplant. For BMT studies, * indicates significant differences within BM genotype and # indicates significant differences within body genotype. Pearson χ² test of independence was used to compare cell size frequency distribution profiles. 3T3-L1 adipocyte studies were analyzed by one-way ANOVA with Bonferroni post hoc test, and * indicates significant differences compared with t = 0.

After 16 weeks on HFD, αHZ and βKO, and their respective WT littermates (i.e., Pik3r1^+/+ and Pik3r2^+/+), gained significant weight compared with mice on CON, with the majority of weight gain due to an increase in body fat (Fig. 1A and B). Despite comparable obesity, whole-body insulin sensitivity was significantly greater in αHZ-HFD compared with WT-HFD mice as measured by the steady-state glucose infusion rate (GINF) during the H-E clamp (Fig. 1C and Supplementary Fig. 1A). The enhanced GINF in αHZ-HFD was primarily due to an increase in insulin-stimulated glucose disposal rate (IS-GDR), as suppression of hepatic glucose production by insulin was not different between αHZ-HFD and WT-HFD (Fig. 1D and E and Supplementary Fig. 1B). AT insulin sensitivity, measured by the ability of insulin to suppress FFA and glycerol levels during the clamp, was significantly impaired in WT-HFD mice, but was similar to CON in αHZ-HFD mice (Fig. 1F and G and Supplementary Fig. 1C and D). Surprisingly, the metabolic improvements did not manifest in βKO-HFD mice, which were indistinguishable from WT-HFD mice (Fig. 1C–G and Supplementary Fig. 1A–D). Supporting the clamp data, lipid analysis of SkM showed a significant twofold increase in triglyceride and diacylglyceride concentrations in WT-HFD, but not αHZ-HFD, compared with CON-fed mice, consistent with improved IS-GDR in αHZ-HFD (Fig. 1H). Hepatic triglycerides and diacylglycerides were elevated twofold in both WT and αHZ on HFD (Fig. 1I). In isolated adipocytes (Fig. 1J) and SkM (Fig. 1K), again, WT-HFD, but not αHZ-HFD, had reduced insulin-stimulated 2DG uptake. Because AT inflammatory cytokine secretion is thought to drive systemic insulin resistance, we measured insulin sensitivity in isolated EDL muscles after exposure to TNF-α. Insulin-stimulated 2DG uptake in the EDL was equally reduced in WT and αHZ after TNF-α treatment (Fig. 1K), which suggests that improved SkM insulin sensitivity in αHZ-HFD may occur secondary to reduced systemic levels of inflammatory cytokines.

Inflammatory and macrophage markers were significantly elevated in AT and SkM from all groups compared with their respective CON groups, but were greatly reduced in αHZ-HFD vs. WT-HFD (Fig. 2A and Supplementary Fig. 2). Paralleling these changes, more F4/80+ cells were present in characteristic crown-like structures in WT-HFD and βKO-HFD, but not αHZ-HFD, compared with WT-CON (Fig. 2B). By flow cytometry, we found significant, two- and sevenfold increases in F4/80+ and F4/80+/CD11c+ cells, respectively, in the SVC fraction of WT-HFD and βKO-HFD versus WT-CON (Fig. 2C). In contrast, there were significantly less F4/80+ and F4/80+/CD11c+ cells in AT from αHZ-HFD compared to WT-HFD (Fig. 2C). Representative gated flow cytometry scatter plots with average percent of cells per quadrant for each group are presented in Supplementary Fig. 3.

Plasma insulin, resistin, and plasminogen activator inhibitor-1 were significantly elevated in WT-HFD versus WT-CON mice, but not in αHZ-HFD mice. Additionally, plasma adiponectin levels were significantly decreased in WT-HFD, but not αHZ-HFD, mice as compared with CON (Table 1). Chemokine (C-C motif) ligand (CCL) 2 (CCL2)/monocyte chemotactic protein-1 (MCP-1) was not increased in the plasma of HFD mice; however, both IL-6 and leptin were significantly increased in HFD, regardless of genotype (Table 1).

We profiled media collected from αHZ-HFD, WT-HFD, and WT-CON AT explants. Of the 62 cytokines measured, 10 inflammatory chemokines were differentially secreted in AT from WT-HFD versus αHZ-HFD mice (Table 2). Six of these secreted factors, chemokine (C-X-C motif) ligand (CXCL) 13 (CXCL13), CXCL12, CXCL4, CXCL16, CCL1, and l-selectin are chemoattractants for immune cells (2,29,30) and were elevated in WT-HFD but not αHZ-HFD. Macrophage colony stimulating factor (M-CSF), a cytokine that stimulates macrophage differentiation (31), and CXCL10, a proinflammatory cytokine released by macrophages (32), were also increased in conditioned media from WT-HFD only. IL-1α, IL-3, and tissue inhibitor of metalloproteinases 1 (TIMP-1) were differentially secreted in αHZ-HFD as compared with WT mice in a pattern that would positively influence extracellular matrix (ECM) remodeling and collagen synthesis (33,34). The full list of secreted factors can be found in Supplementary Table 1. Paralleling the changes in cytokine secretion, phosphorylation of IKKα/β(S180/181) in AT was increased twofold in WT-HFD versus WT-CON, but was not increased in αHZ-HFD (Supplementary Fig. 7A).

Adipocytes from HFD versus CON were larger in size but fewer in number (Supplementary Fig. 4A–D). In CON groups, a greater percentage of αHZ adipocytes were larger than WT, whereas βKO had a greater percentage of cells that were smaller (Supplementary Fig. 4A and B). In HFD groups, αHZ and βKO had a greater percentage of larger adipocytes compared with WT (Supplementary Fig. 4A and B). Notably, despite the larger cell size in both αHZ-HFD and βKO-HFD, only the αHZ-HFD mice were protected against ATM infiltration and adipocyte insulin resistance.

BMT studies were performed in lethally irradiated Pik3r1^+/− and WT mice. We created four chimeras: WT BM into WT mice (WT→WT; donor BM genotype→recipient mouse genotype), Pik3r1^+/− BM into WT mice (αHZ→WT), Pik3r1^+/− BM into Pik3r1^+/− mice (αHZ→αHZ), and WT BM into Pik3r1^+/− mice (WT→αHZ) (Fig. 3A). Flow cytometry on peripheral blood monocytes shows that engraftment was complete within 6 weeks of transplant (Supplementary Fig. 5A–D). Mice with >80% engraftment were placed on an HFD for 12 weeks and insulin sensitivity was measured by clamp. Body weight and percent body fat were significantly increased in all BMT-HFD mice compared with WT→WT-CON mice (Fig. 3B–C). Insulin sensitivity (GINF) was not impaired in obese αHZ→αHZ-HFD, but was significantly reduced in WT→WT-HFD mice versus WT→WT-CON (Fig. 3D). Replacing WT BM with Pik3r1^+/− BM did not improve insulin sensitivity in obese WT mice (αHZ→WT), and GINF was similar to WT→WT-HFD mice (Fig. 3D). Replacing Pik3r1^+/− BM with WT BM did not impair insulin sensitivity as GINF in WT→αHZ-HFD was similar to αHZ→αHZ-HFD mice (Fig. 3D).

Complementing the insulin sensitivity data, gene expression of inflammatory and macrophage markers in AT was significantly increased in WT→WT-HFD and αHZ→WT-HFD mice compared with WT→WT-CON, but were attenuated in αHZ→αHZ-HFD and WT→αHZ-HFD mice (Fig. 4A). In addition, only AT from WT→WT-HFD and αHZ→WT-HFD mice had a significant increase in F4/80+ cells forming crown-like structures by immunohistochemistry and more F4/80+ (approximately twofold) and F4/80+/CD11c+ (approximately eightfold) cells counted in the SVCs compared with WT→WT-CON mice (Fig. 4B and C). Macrophage infiltration was markedly reduced and essentially absent in WT→αHZ-HFD and αHZ→αHZ-HFD mice (Fig. 4B and C).

Insulin-stimulated tyrosine phosphorylation of IRS1 (pIRS1[Y612]) was significantly and equally impaired (approximately threefold) in AT of all HFD versus CON mice (Fig. 5A and B and Supplementary Fig. 6A). The decrease in IRS1 activation was accompanied by a significant reduction (∼50%) in IRS1 abundance (Fig. 5A and Supplementary Fig. 6A). This reduction was likely due to increased IRS1 serine phosphorylation as JNK1(T138/Y185) phosphorylation was equally upregulated in AT from WT-HFD and αHZ-HFD mice (Supplementary Fig. 7B). X-box binding protein 1 splicing and gene expression of the ER stress markers growth arrest and DNA damage-inducible gene-153 (GADD153) and glucose-regulated protein, 78kDa (GRP78) and hypoxia markers, vascular endothelial growth factor (VEGF), GLUT1, and hypoxia-inducible factor (HIF) 1α, were elevated in AT with HFD regardless of genotype (Supplementary Fig. 7C–E). These findings are important in that the decrease in IRS1 activation and upregulation of pJNK1(T138/Y185), ER stress, and hypoxia occurred without profound ATM accumulation in αHZ-HFD. Despite reduced insulin-stimulated IRS1 activation, insulin-stimulated Akt(S473) and Akt(T308) phosphorylation was not impaired in αHZ-HFD, but was impaired in WT-HFD versus CON mice (Fig. 5A, D, and E). Enhanced insulin-stimulated Akt activation did not occur in βKO-HFD mice (Supplementary Fig. 6A). Insulin-stimulated, pY-associated PI3K activity in AT was also reduced twofold in WT-HFD versus WT-CON mice, but was not reduced in αHZ-HFD mice (Fig. 5F). There was no difference in basal pY-associated PI3K activity (Fig. 5F).

In parallel with the insulin-stimulated PI3K activity, there was a significant increase in the abundance of the shorter isoforms of Pik3r1 (p55α and p50α; four- and twofold, respectively) and a smaller (40%) increase in p85α in AT from WT-HFD and βKO-HFD versus WT-CON, which did not occur in αHZ-HFD mice (Fig. 5A and C and Supplementary Fig. 6A). Changes in p55α/p50α abundance in WT-HFD were paralleled by an increase in mRNA levels of p55α and p50α, but not p85α (Supplementary Fig. 7F). p110α abundance was not different across groups (Fig. 5A). Interestingly, we observed only a modest, nonsignificant decrease in p85α abundance in AT from Pik3r1^+/− mice despite a significant, 50% decrease in mRNA expression, suggesting that Pik3r1 protein stability is likely increased. We speculate that increased protein stability may be due to a greater number of the Pik3r1 regulatory subunits binding to the catalytic subunit to form heterodimers or to other p85α binding partners. Indeed, association with the catalytic subunit has been previously shown to increase heterodimer stability (35). Thus, these data suggest that HFD appears to reduce insulin-stimulated PI3K activity and downstream signaling to Akt, at least in part through upregulation of p85α, p55α, and p50α expression and protein abundance in AT.

Despite severe obesity, SkM insulin signaling was not impaired in αHZ-HFD mice compared with αHZ-CON, whereas WT-HFD mice had significant impairments in insulin-stimulated tyrosine phosphorylation of pIRS1(Y612) (∼50%), insulin-stimulated pY-associated PI3K activity, and Akt(S473) phosphorylation compared with WT-CON (Supplementary Fig. 8A, B, E, and F). Similar to AT, p85α abundance was increased in WT-HFD SkM, with no change in p110 abundance (Supplementary Fig. C–E). Although there was no significant decrease in p85α abundance in αHZ SkM, we did find an ∼50% decrease in the p110α catalytic subunit (Supplementary Fig. C–E). This decrease in p110 abundance was reflected in a 50% decrease in insulin-stimulated, pY-associated PI3K activity in both αHZ-CON and αHZ-HFD SkM; however, correcting for catalytic subunit abundance in all groups, insulin-stimulated, pY-associated PI3K activity/p110 was significantly higher in αHZ groups compared with their diet-matched WT group (Supplementary Fig. 6F). The adjusted insulin-stimulated PI3K activity matches the increase in Akt phosphorylation seen in αHZ mice compared with WT groups (Supplementary Fig. 8B and F).

Nuclear localization of signal transducer and activator of transcription 3 (STAT3) (Fig. 6A) was increased in AT of WT-HFD and αHZ-HFD versus WT-CON. Because IL-6 was elevated in both WT-HFD and αHZ-HFD and is known to activate STAT3 (36), we studied the effects of IL-6 treatment on p85α, p55α, and p50α expression in differentiated 3T3-L1 adipocytes. IL-6 rapidly activated STAT3(Y705) phosphorylation within 15 min (Fig. 6B), corresponding to an acute rise in STAT3 genes IL-6 and SOCS3 (Fig. 6C). Interestingly, STAT3 activation by IL-6 appeared biphasic, with peak phosphorylation at 1 and 24 h (Fig. 6B). This pattern of activation suggests the possibility that acute IL-6 stimulation of STAT3 may lead to secretion of a second factor that then reactivates STAT3 signaling. Only chronic IL-6 stimulation led to a significant increase in p55α and p50α expression (Fig. 6D). No increase in p85α expression was found in response to IL-6 stimulation (Fig. 6D).

Despite strong evidence for a link between macrophage accumulation in AT and insulin resistance in obesity, the factors that initiate chemokine secretion for macrophage recruitment remain elusive. In this study, we introduce the idea that AT insulin resistance is an important factor that links cellular metabolism, particularly insulin action, to chemokine secretion and initiation of macrophage recruitment.

Elegant transgenic mouse studies have demonstrated that AT inflammation and macrophage chemotaxis are reduced when signaling pathways, such as JNK, toll-like receptor 4, and IKK, are manipulated in the hematopoietic compartment (6–8). The underlying conclusion from these and comparable studies (1) is that AT inflammation is driven by macrophage infiltration, and secondary to this infiltration, AT insulin resistance manifests. In contrast, however, several studies have found dissociation between ATM infiltration and the development of insulin resistance. For example, in lean mice null for transforming growth factor-β receptor and apolipoprotein E, there is massive T-cell activation, hyperlipidemia, ATM infiltration, and upregulation of inflammatory cytokines but no evidence of adipocyte or systemic insulin resistance (37). In our study, AT from WT-HFD had significant increases in F4/80+ and F4/80+/CD11c+ cells, whereas accumulation of these cells in AT was attenuated in αHZ-HFD mice. Our BMT experiments demonstrate that the metabolic improvements in αHZ-HFD mice are independent of changes in Pik3r1 expression in myeloid cells. In fact, even when αHZ mice were transplanted with WT BM (WT→αHZ), which would be expected to facilitate F4/80+/CD11c+ infiltration and AT inflammation in obese mice, F4/80+/CD11c+ cells were significantly reduced compared with obese WT→WT mice. Taken together, these experiments demonstrate that the impact of reduced PI3K regulatory subunits on improved insulin sensitivity and macrophage recruitment in obese Pik3r1^+/− mice is driven by non-hematopoietic–derived cells.

Local hypoxia, upregulation of proinflammatory JNK1 and IKKβ pathways, and induction of ER stress have all been shown to be key signaling pathways that contribute to AT insulin resistance, primarily through inhibition of insulin signaling at the level of IRS1 (1). We found that activation of JNK, ER stress, and hypoxia were induced similarly in AT from WT-HFD, βKO-HFD, and αHZ-HFD mice, in parallel with reduced insulin-stimulated activation of IRS1. Despite these changes in IRS1, insulin-stimulated PI3K activity was not reduced in AT from αHZ-HFD mice but was impaired in WT-HFD. Thus, these data demonstrate that partial deletion of Pik3r1, and subsequent maintenance of PI3K function, bypasses the obesity- and inflammation-mediated reductions in tyrosine activation of IRS1. Moreover, our results clearly demonstrate that PI3K-dependent signaling is critical for ATM recruitment in obesity. A separate study found that increased AT lipolysis leads to a rapid increase in ATM accumulation in lean mice after fasting or after initial weight loss in HFD-fed mice, suggesting that increased FFA may be an important cue for macrophage recruitment (38). Like the findings of Kosteli et al. (38), αHZ-HFD mice had no impairments in insulin suppression of FFAs or glycerol during a clamp and significantly less ATM accumulation as compared with WT-HFD mice, which failed to suppress lipolysis during the clamp. Mechanistically, we believe that the local tissue environment plays a major role in reducing ATM recruitment in Pik3r1^+/− mice. To this end, AT explants from αHZ-HFD mice secreted less proinflammatory cytokines compared with WT-HFD mice. This reduction in chemokine and cytokine secretion may be linked to the reduction in IKK(S180/S181) phosphorylation in AT from αHZ-HFD compared with WT-HFD mice. Activation of IKKβ is thought to be a key regulator of inflammatory gene expression in obesity through regulation of nuclear factor-κB activity (7). Alternatively, reduced inflammation in αHZ-HFD compared with WT-HFD mice may be linked to the increased circulating adiponectin, which has potent anti-inflammatory effects (39).

Although PI3K is a key regulator of insulin signaling, surprisingly, little is known about the signaling mechanisms that modulate the expression of the regulatory subunits of PI3K in response to nutrient flux. Increased expression of p85α in peripheral tissues occurs in response to increased insulin (40), growth hormone (21), and dexamethasone (41). Additionally, rosiglitazone increases p85α expression in human adipocytes through peroxisome proliferator–activated receptor γ activation (42), and SkM p85α expression is strongly induced by peroxisome proliferator–activated receptor α activation (43). We speculate that the increase in p85α in WT-HFD AT and SkM may be due to increased fasting insulin in WT-HFD mice. Regarding p55α/p50α transcription, Abell et al. (44) reported that STAT3 activation increases p55α/p50α transcription during mammary gland differentiation, stimulating apoptosis through inhibition of PI3K activity. A similar pattern of PI3K regulation has recently been proposed in SkM in response to caloric restriction (22). In agreement, we found that AT p55α/p50α abundance was increased in parallel with increased STAT3 nuclear localization in WT-HFD mice and in 3T3-L1 cells in response to chronic IL-6 exposure. Limiting STAT3 transcription of Pik3r1 in AT could be an attractive target for preventing obesity-induced insulin resistance. In support of this, mice with AT-specific knockout of STAT3 have increased body weight and adiposity with no impairment in insulin sensitivity (45).

In summary, our results suggest that impaired AT insulin sensitivity in obesity at the level of PI3K is a critical step necessary for ATM infiltration and inflammation. Moreover, they suggest a model in which reduced PI3K function and AT insulin resistance manifests due to increased adipocyte activation of STAT3 and p55α/p50α transcription. These data also raise new questions as to whether the reduced inflammatory response in AT of Pik3r1^+/− mice is due directly to improvements in insulin action or, alternatively, related to improved PI3K activity independent of the insulin signaling pathway. Thus, the molecular mechanism of how PI3K signaling affects macrophage function remains to be further defined; however, the present results suggest that modulating AT Pik3r1 expression and PI3K function is a potentially attractive avenue for the development of novel therapies to treat AT insulin resistance and inflammation in obesity.

This work was supported by grants from the National Institutes of Health (P30-DK-048520 and P30-DK-57516 to C.E.M., P30-AR-058878 and R24 HD050837 to S.S., DK-059767 to J.E.F., and DK-053969 to D.J.K.) and the Agence Nationale de la Recherche (ANR-09-RPDOC-018-01 to D.P.). C.E.M. is supported by the Office of Research in Women’s Health (K12-HD-057022).

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

C.E.M., S.S., and A.P. researched data, contributed to the discussion, and wrote and edited the manuscript. M.J.H. and J.A.H. researched data and contributed to the discussion. D.P., P.S.M., S.M.M., and D.J.K. researched data, contributed to the discussion, and edited the manuscript. J.E.F. contributed to the discussion and edited the manuscript. C.E.M. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Parts of this study were presented in abstract form at the 70th Scientific Sessions of the American Diabetes Association, Orlando, Florida, 25–29 June 2010, and at the 68th Scientific Sessions of the American Diabetes Association, San Francisco, California, 6–10 June 2008.

The authors thank Dr. C. Ronald Kahn (Joslin Diabetes Center at Harvard Medical School, Boston, MA) for providing the transgenic mice used in this study, Dr. Jerrold M. Olefsky (University of California, San Diego) for guidance, and Keith J. Fox and Heidi Miller (University of Colorado School of Medicine) for technical assistance with 3T3-L1 adipocyte cultures.