Skeletal Muscle Insulin Signaling Defects Downstream of Phosphatidylinositol 3-Kinase at the Level of Akt Are Associated With Impaired Nonoxidative Glucose Disposal in HIV Lipodystrophy

Eighteen HIV-1–infected patients who displayed fat redistribution (LIPO patients) and 18 HIV-1–infected patients without fat redistribution (NONLIPO patients, control subjects) were recruited from the outpatient clinic of infectious diseases at Hvidovre University Hospital (Copenhagen, Denmark). All patients were on HAART and of male sex. A description of selection, anthropometry, immunology, components of HAART, and glucose metabolism of all subjects has been presented (11,21). In brief, for case subjects, a questionnaire and a physical examination had to be positive for signs of lipodystrophy, whereas for control subjects, both the questionnaire and physical examination had to be negative. All case subjects, except one who had multiple lipomatosis, displayed lipodystrophy consistent with peripheral fat atrophy and central fat accumulation (i.e., mixed lipodystrophy [2]). Except from HAART, none of the subjects received medication known to affect glucose metabolism, and all subjects had to present fasting plasma glucose concentrations <7.0 mmol/l. All participants except two NONLIPO patients had a negative family history of diabetes. All participants except two NONLIPO patients (black Africans) were of Caucasian ethnicity. Subjects gave their written informed consent, and the protocol was approved by the ethical committee in Copenhagen, Denmark, and performed in accordance with the Helsinki Declaration II.

All subjects were instructed to abstain from strenuous exercise for at least 3 days before the metabolic assessments. The HIV-infected patients reported to our laboratory at 0800 after a 12-h overnight fast, including a 16-h withdrawal of HAART. A euglycemic-hyperinsulinemic clamp preceded by an intravenous glucose tolerance test was undertaken as described previously (11). In brief, a primed continuous infusion of [3-³H]tritiated glucose (bolus, 2.05 μCi · mmol⁻¹ · l⁻¹ fasting plasma glucose; continuous, 0.11 μCi/min; New England Nuclear, Boston, MA) was initiated at 0900, at the start of a 150-min basal period (−150 to 0 min) and continued throughout the study, which included an intravenous glucose tolerance test (0–30 min, 300 mg glucose/kg body wt, performed to characterize the first-phase insulin response [22]; data from the intravenous glucose tolerance test and a 120-min euglucemic-hyperinsulinemic clamp (30–150 min) have been presented previously [11,21]). Two steady-state periods were predefined, i.e., at −60 to −30 min as basal and 120–150 min as clamp. At both steady-state periods, indirect calorimetry was undertaken using the flow-through canopy gas analyzer system (Deltatrac; Datex, Helsinki, Finland). An insulin (Actrapid; Novo Nordisk, Bagsvaerd, Denmark) infusion was started at 30 min with a stepwise reduction in the infusion rate every 3rd min from 100 to 80 to 60 to 40 mU · m⁻² · min⁻¹. Hereafter (39–150 min), the insulin infusion rate was fixed at 40 mU · m⁻² · min⁻¹. The plasma glucose concentration was kept at ∼5 mmol/l by adjusting the infusion rate of glucose (180 g/l), which was enriched with tritiated glucose (110 μCi/l). Total glucose disposal rates (GDRs) were calculated using Steele’s non–steady-state equations adapted for labeled glucose infusates (23). Distribution volume of glucose was taken as 200 ml/kg body wt and pool fraction as 0.65. Rates of glucose oxidation (GOX) and lipid oxidation were calculated from Frayn’s equation (24). NOGM was calculated as the difference between GDR and GOX. Plasma glucose, insulin, free fatty acids (FFAs), glycerol, triglyceride, total cholesterol, and HDL cholesterol were measured as described previously (11). Muscle biopsies were obtained from the vastus lateralis muscle immediately after both steady-state periods using a modified Bergström needle with suction under local anesthesia (glucose clamp was continued during sampling of the insulin-stimulated muscle biopsy). Muscle samples were immediately blotted free of visible nonmuscle tissue and frozen in liquid nitrogen within 20–30 s. The muscle samples were stored at −80°C until processed as follows. The muscle samples were freeze-dried and were carefully dissected free of eventually remaining blood, fat, and connective tissue before analysis of muscle enzyme activity and phosphorylation, which were measured in muscle homogenates (25). Body composition was estimated by dual-energy X-ray absorptiometry scanning (XR-36; Norland Medical System, Fort Atkinson, WI) as previously described (21).

Muscle glycogen synthase activity was measured in muscle homogenates using a microtitter plate assay (Unifilter 350 Plates; Whatman, Cambridge, U.K.). The assay ran in triplicates based on the original protocol described by Thomas et al. (26). Glycogen synthase activity was determined in the presence of 8, 0.17, or 0.02 mmol/l glucose-6-phosphate (G6P). Glycogen synthase activity is given either as the percentage of G6P-independent glycogen synthase activity (percent I-form activity) (100 × activity in the presence of 0.02 mmol/l G6P divided by the activity at 8 mmol/l G6P [saturated]) or as the percentage of fractional velocity (percent FV) (100 × activity in the presence of 0.17 mmol/l G6P divided by the activity at 8 mmol/l G6P).

Protein expression and protein phosphorylation were examined in muscle homogenates by SDS-PAGE (10 or 7.5% Criterion gels; Bio-Rad, Hercules, CA) and Western blotting using polyvinylidene fluoride membrane and semidry transfer. After membrane blocking (Tris-buffered saline with 1% Tween [TBST] and 2% skim milk), glycogen synthase protein was detected as a single band at ∼90 kDa using a polyclonal glycogen synthase antibody (TBST and 2% skim milk) (provided by Prof. Oluf Pedersen, Steno Diabetes Center, Copenhagen, Denmark) (27). Glycogen synthase phosphorylation was determined using five phospho-specific polyclonal antibodies against glycogen synthase as described previously (28–30), i.e., anti-site 2 (Ser7) and anti-site 2+2a (Ser7 and Ser10) antibodies raised against the peptides PLSRSLSpMSSLPGLED and PLSRSLSpMSSpLPGLED (residues 1–16 of rat glycogen synthase) and an anti-site 3a+3b (Ser640 and Ser644) antibody raised against the peptide RYPRPASpVPPSpPSLSR (residues 634–50 of human glycogen synthase), an anti-site 3a (Ser640) antibody raised against the peptide RYPRPASpVPPSPSLSR (residues 634–50 of human glycogen synthase), and an anti-site 3b (Ser644) antibody raised against the peptide RYPRPASVPPSpPSLSR (residues 634–50 of human glycogen synthase). Besides previous evaluation (28–30), all five antibodies used have proven to be phospho-site–specific by phosphopeptide-competition assay performed toward the remaining eight known regulatory serine residues on glycogen synthase (J.N. Nielsen, J.F.P.W., unpublished observations). Akt 1+2 protein, Akt Ser473, and Akt Thr308 phosphorylation was determined using primary antibodies from New England Biolabs (Beverly, MA) and Upstate Biotechnology (Lake Placid, MA). GSK-3α Ser21 and GSK-3β phosphorylation was determined using primary antibodies from Upstate Biotechnology and Transduction Laboratories (San Diego, CA). All primary antibodies were in TBST and 2% skim milk and detected using horseradish peroxidase–conjugated secondary antibodies (in TBST and 2% skim milk) and enhanced chemiluminescence. The bands were visualized and analyzed using a Kodak Image Station E440CF (Kodak Image Station, E440 CF; Kodak, Ballerup, Denmark).

IRS-1–associated PI 3-kinase activity was measured in IRS-1 immunoprecipitated from muscle homogenates (400 μg protein) using an anti–IRS-1 antibody raised against the COOH terminus of IRS-1 provided by Dr. K. Siddle (Cambridge University) (31) and protein G agarose beads (Pierce, Rockford, IL). Pellet was washed twice in PBS containing 1% NP-40 and 100 μmol/l NA₃VO₄ (pH 7.5) and twice in 10 mmol/l Tris buffer containing 100 mmol/l NaCl, 1 mmol/l EDTA, and 100 μmol/l NA₃VO₄ (pH 7.5). The pellet was resuspended in 50 μl of the latter buffer, and the PI 3-kinase assay (30°C for 20 min) was performed as described previously (32).

Data are presented as means ± SE and as medians and interquartile ranges when distributions were skewed. ANOVA was performed to compare distribution of data between LIPO patients and NONLIPO patients. Because age was slightly increased in LIPO patients compared with NONLIPO patients, age was adjusted for by univariate ANOVA. If data distribution was skewed, data were log transformed before applying ANOVA. Pearson correlation coefficient (r) and Spearman correlation coefficient (ρ) were applied to estimate associations between variables, when appropriate. Calculations were performed by SPSS (SPSS ver. 12.0; SPSS, Chicago, IL). Two-sided P values <0.05 were defined as statistically significant. A trend was noted if P was ≥0.05 but <0.2.

Because LIPO patients were slightly older than NONLIPO patients, age was corrected for in the comparisons between groups (Table 1). LIPO patients had slightly higher BMIs attributable to an increased lean body mass, whereas total fat mass did not differ significantly between study groups (Table 1). The ratio of limb fat to trunk fat was significantly reduced in LIPO compared with NONLIPO patients (P < 0.001), consistent with lipodystrophy in LIPO patients. Duration of HIV infection, duration of HAART, and CD4 cell number did not differ significantly between study groups. HIV-RNA was fully suppressed in both study groups (Table 1). All LIPO and NONLIPO patients were treated with nucleoside reverse transcriptase inhibitors as part of HAART. Nucleoside reverse transcriptase inhibitors used in LIPO and NONLIPO patients were those of lamivudine (83 and 83%), zidovudine (39 and 50%), stavudine (56 and 39%), didanosine (6 and 11%), and abacavir (6 and 0%), respectively. The following HIV-1 protease inhibitors were used by 89% of LIPO and NONLIPO patients: indinavir (44 and 22%), ritonavir (22 and 39%), nelfinavir (22 and 22%), and saquinavir (17 and 11%), respectively.

Basal and clamp glucose concentrations were similar between the study groups (Table 1). All subjects displayed basal plasma glucose ≤6.1 mmol/l (four LIPO and two NONLIPO patients had fasting plasma glucose ≥5.6 mmol/l; data not shown). Basal, clamp, and incremental insulin concentrations were increased in LIPO patients (Table 1). Insulin-mediated suppression of plasma FFAs, glycerol, and lipid oxidation was attenuated in LIPO patients, despite a greater increment in plasma insulin (Table 1). Basal and clamp non-HDL cholesterol were increased in LIPO patients, whereas plasma triglyceride tended to be increased in LIPO patients (Table 1).

Steady state of plasma glucose and insulin was established in both study groups during the latter 75 min of the clamp period (data not shown). In the basal state, the GDR was similar between LIPO and NONLIPO patients, whereas during clamp, the steady-state GDR was 34% higher in NONLIPO compared with LIPO patients (P < 0.001, P = 0.002 age adjusted; Fig. 1). The GDR increased significantly during insulin stimulation in both groups, but the increase was significantly lower in LIPO patients (P < 0.001, P = 0.002 age adjusted). Basal GOX and NOGM did not differ between study groups, whereas clamp GOX was 17% lower (P = 0.056, P < 0.05 age adjusted) and clamp NOGM was 49% lower (P = 0.002, P = 0.022 age adjusted) in LIPO patients. Insulin increased GOX and NOGM significantly in both groups, but the increases in NOGM (P = 0.005, P = 0.054 age adjusted) and GOX (P < 0.001, P = 0.002 age adjusted) were both impaired in LIPO patients (Fig. 1). The lower NOGM accounted for 72% of the reduction in incremental GDRs in LIPO patients, demonstrating that the reduced glucose uptake was caused primarily, but not solely, by an impaired stimulation of glucose storage.

Basal glycogen synthase activities, both as the percent I-form and as percent FV, were almost identical in both study groups (Fig. 2A and B). Insulin-stimulated glycogen synthase activity (percent I-form) increased significantly in both LIPO (80%) and NONLIPO (127%) patients; the increment was higher in NONLIPO patients (P = 0.032, P = 0.14 age adjusted). Insulin also stimulated glycogen synthase activity (percent FV) significantly in both LIPO (29%) and NONLIPO (47%) patients, with the increment tending to be higher in NONLIPO patients (P = 0.072, P = 0.12 age adjusted). Correlation analyses were performed to examine whether the level of insulin-stimulated and incremental glycogen synthase activity both as percent I-form and as percent FV could account for the rate of insulin-stimulated and incremental glucose storage. Strong positive correlations were demonstrated for all of these variables (Fig. 2C–F). Total glycogen synthase protein and glycogen content in muscle tissue did not differ significantly between LIPO and NONLIPO patients at baseline (118 ± 8 vs. 144 ± 11 arbitrary units (AU), NS; 465 ± 30 vs. 464 ± 23 mmol/kg dry wt, NS) or during clamp (127 ± 7 vs. 140 ± 10 AU, NS; 489 ± 30 vs. 467 ± 8 mmol/kg dry wt, NS).

Because phosphorylation of the glycogen synthase NH₂-terminal sites 2+2a and the COOH-terminal sites 3a and 3a+b have been shown to be essential for glycogen synthase activity (28,33), we investigated basal and insulin-induced phosphorylation of these sites. Basal glycogen synthase phosphorylation at sites 2+2a was not different between study groups (Fig. 3A). Insulin decreased glycogen synthase phosphorylation at sites 2+2a in NONLIPO patients (−17%), whereas insulin insignificantly increased phosphorylation at this glycogen synthase site in LIPO patients (13%, P = NS), such that the incremental changes were significantly different between study groups (P = 0.036, although the difference attenuated after age adjustment, P = 0.19). Basal phosphorylation of glycogen synthase at site 3a tended to be increased in NONLIPO compared with LIPO patients (P = 0.16, P = 0.19 age adjusted; Fig. 3B) but was significantly reduced by insulin in NONLIPO patients only (−68%, P < 0.01), such that the incremental decrement tended to be greater in NONLIPO patients (P = 0.051, P = 0.085 age adjusted). Basal phosphorylation of glycogen synthase at sites 3a+b was significantly increased in NONLIPO patients (Fig. 3C). However, phosphorylation of glycogen synthase sites 3a+b showed a significant decrease in NONLIPO patients after insulin stimulation (−49%, P < 0.01), whereas LIPO patients did not change, making the incremental change highly significant (P = 0.001, P = 0.004 age adjusted). Phosphorylation of sites 2 and 3b was similar between study groups during basal and clamp conditions and was not influenced by insulin stimulation (data not shown). Correcting for total glycogen synthase protein did not significantly change the results given above (data not shown).

Strong inverse correlations were observed between the degrees of phosphorylation (log transformed) of these glycogen synthase sites versus glycogen synthase activity (percent I-form) during clamp (Fig. 3D–F), suggesting that the degree of phosphorylation of glycogen synthase sites 2a+b, 3a, and 3a+b could account for glycogen synthase activity level. A multivariate linear regression analysis, which included these sites as potential predictors of glycogen synthase activity, revealed that 43% of the variation in clamp glycogen synthase activity could be accounted for by degree of phosphorylation of sites 2a+b and 3a+b during clamp (r² = 0.43, P < 0.0001).

We next examined whether these defects of muscle glycogen synthesis and glycogen synthase phosphorylation could be attributed to a defect(s) in the proximal or distal steps of the signal transduction cascade currently believed to mediate insulin activation of glycogen synthase.

Basal IRS-1–associated PI 3-kinase activity tended to be increased in LIPO patients compared with NONLIPO patients (P = 0.079, P = 0.063, age adjusted; Fig. 4A). Insulin stimulation significantly increased the IRS-1–associated activity in LIPO patients (77%, P < 0.001) and in NONLIPO patients (103%, P < 0.001; Fig. 4A), with no differences in clamp and incremental values between study groups.

Basal levels of Akt Ser473 and Akt Thr308 phosphorylation were low and did not differ between study groups (NS). During clamp steady state, Akt Ser473 tended to be higher in NONLIPO patients (P = 0.069, P = 0.015 age adjusted), and insulin-induced Akt Thr308 was increased in NONLIPO patients (P = 0.033, P = 0.015 age adjusted). Akt protein did not differ between study groups and was not changed by insulin stimulation (data not shown). Correcting the phosphorylation levels of Akt for Akt protein content, however, made the differences of clamp levels of Akt Ser473 (P = 0.029, P = 0.007 age adjusted) and Akt Thr308 (P = 0.009, P = 0.015 age adjusted) phosphorylation between study groups more distinct, leaving NONLIPO patients as those patients with the highest Akt phosphorylation during clamp steady state (Fig. 4B and C). Basal Akt protein was used for correction of Δ values.

Basal GSK-3α phosphorylation tended to be increased in LIPO patients (P = 0.14, P = 0.028 age adjusted). Insulin-induced phosphorylation of GSK-3α was significant in NONLIPO patients only (48%; Fig. 4D). Basal GSK-3β did not differ between study groups (Fig. 4E). Insulin stimulated the degree of GSK-3β phosphorylation significantly in LIPO patients (34%) and in NONLIPO patients (69%), with the increment being most pronounced in NONLIPO patients (P = 0.015, P = 0.007 age adjusted).

These results indicate that a defect in insulin signaling in LIPO patients could reside downstream from IRS-1–associated PI 3-kinase activity at the level of Akt. This interpretation was strengthened by the observation of significant positive correlations between protein-corrected Akt Thr308 phosphorylation versus NOGM (r = 0.42, P = 0.01; Fig. 5A) and glycogen synthase activity (I-form, ρ = 0.39, P < 0.02) during clamp. Trends were shown for correlations between protein-corrected Akt Ser473 versus NOGM (r = 0.33, P = 0.050) and glycogen synthase activity (I-form, ρ = 0.25, P = 0.14) during clamp. Clamp IRS-1–associated PI 3-kinase activity and clamp NOGM did not correlate (r = 0.15, P > 0.3; Fig. 5B), suggesting that the defect in insulin signaling did not reside at this level. Incremental Akt Thr308 phosphorylation correlated positively with both incremental GSK-3α and -3β phosphorylation (Fig. 5C and D), which was also observed for incremental Akt Ser473 versus GSK-3α and -3β phosphorylation (r = 0.52, P < 0.002 and r = 0.32, P = 0.058, respectively). Thus, degree of Akt phosphorylation may account for the defects at the GSK-3α and -3β level.

Because rate of lipolysis and dyslipidemia may influence glucose uptake and glycogen syntheses (34), we examined whether markers of lipolysis (i.e., clamp levels of FFAs and glycerol) and clamp plasma levels of non-HDL cholesterol and triglyceride correlated with clamp glycogen synthase activity in both percent I-form and percent FV form (Table 2). Several significant inverse correlations were observed in support of this view. Insulin-stimulated Akt phosphorylation and fat distribution (ratio of limb fat to trunk fat) correlated significantly with clamp levels of FFAs and glycerol. Basal FFAs and basal glycerol, however, did not correlate significantly with insulin-stimulated glycogen synthase activity, Akt phosphorylation, or fat distribution (all P > 0.2; data not shown).

Fat redistribution as depicted by the ratio of limb fat to trunk fat may be more related to insulin resistance than trunk fat and limb fat masses, themselves (11,35). This relationship may also be reflected in insulin signaling. Therefore, we tested whether fat redistribution (ratio of limb fat to trunk fat) would correlate more strongly with insulin-stimulated glycogen synthase activity/phosphorylation and upstream insulin signaling effectors than trunk fat and limb fat masses (Table 3). This might be the case, because the ratio of limb fat to trunk fat showed several significant correlations with these effectors of insulin signaling, whereas trunk fat and limb fat themselves failed to demonstrate any significant correlations.

The overall picture of the present study is that decreased insulin-induced NOGM is associated with decreased glycogen synthase activity in lipodystrophic HIV-infected patients. This defect in insulin action on glycogen synthase activity is associated with reduced dephosphorylation on sites 2+2a, 3a, and 3a+b and decreased phosphorylation of GSK-3α and -3β and Akt Ser473 and Akt Thr308 but not impaired IRS-1–associated PI 3-kinase activity.

Patients were studied after abstinence from HAART for 16 h. Therefore, the acute effect of antiretroviral drugs on glucose metabolism was likely attenuated (7). In that respect, we probably studied more the effect of fat redistribution than that of HAART on insulin signaling pathway. The present results indicate that defective activation of glycogen synthase may explain a major part of the reduced NOGM observed in lipodystrophic HIV-infected patients (10–12). In concord with recent observations in obese patients with type 2 diabetes, defects in dephosphorylation of glycogen synthase at sites 2+2a (28) was observed in HIV lipodystrophy. Moreover, the association observed between defective dephosphorylation of glycogen synthase site 3a and sites 3a+b versus glycogen synthase activity fits previous results demonstrating that the glycogen synthase COOH-terminal sites 3a and 3b, including glycogen synthase NH₂-terminal sites 2 and 2a, are the sites most important for glycogen synthase activity (28,33,36). The present results suggest that aberrant phosphorylation at specific glycogen synthase sites may convey derangement in glucose homeostasis and increase risk for type 2 diabetes in HIV lipodystrophy.

Significant correlations were observed between insulin-stimulated Akt phosphorylation versus NOGM and glycogen synthase activity, including incremental Akt versus incremental GSK-3 phosphorylation, suggesting a causal interplay (33,37,38). The more proximal steps of insulin signaling were addressed in the present study by analyzing IRS-1–associated PI 3-kinase activity only. This proximal step appears to be crucial, however, and defects in yet more proximal steps in insulin signaling are likely to be reflected in an impaired stimulation of IRS-1–associated PI 3-kinase (39,40). Obese type 2 diabetic patients and obese glucose-intolerant relatives of type 2 diabetic patients have been shown to exhibit an increased basal activity of IRS-1–associated PI 3-kinase (28,41), which was observed also in lipodystrophic patients in the present study, although the difference did not reach statistical significance. Insulin-stimulated IRS-1–associated PI 3-kinase activity has been shown to be blunted in obese type 2 diabetic patients, obese normoglycemic patients, and obese glucose-intolerant relatives to type 2 diabetic patients (28,41), which contrast the present findings in lipodystrophic as well as nonlipodystrophic HIV-infected patients. Both study groups showed a similar and considerable increment in IRS-1–associated PI 3-kinase activity after insulin stimulation consistent with previous findings in subjects with normal insulin action (32), strongly suggesting that insulin signaling as far as this proximal step was normal in these normoglycemic HIV-infected patients.

Insulin signaling downstream from PI 3-kinase to Akt was attenuated in lipodystrophic HIV-infected patients, suggesting that defects in insulin signaling in these patients may reside at Akt, possibly by involving mediators of Akt phosphorylation downstream of PI 3-kinase, e.g., PDK activity (42). Interestingly, as a model for HIV-1 protease inhibitors, nelfinavir has been shown to induce impaired insulin stimulation at the Akt Ser473 level, which may be mediated through reduced PDK activity in differentiated 3T3-L1 adipocytes, whereas signaling to the IRS-1 and PI-3 kinase levels remained intact (15,43). On the other hand, in vitro studies have shown that the HIV-1 protease inhibitors indinavir, saquinavir, and ritonavir, which are all frequently used in the cocktail of HAART, impair insulin signaling at more proximal steps, that is, at the level of insulin receptor binding and IRS-1 and IRS-2 phosphorylation (16,17,44). In vitro observations suggest that indinavir and ritonavir inhibit GLUT4 activation but not insulin-stimulated GLUT4 assimilation to the plasma membrane (45,46), whereas impaired insulin-stimulated GLUT4 recruitment seems to be a mechanism explaining in part the induction of insulin resistance by nelfinavir in vitro (43).

HIV-infected patients with lipodystrophy have been found to exhibit increased rate of lipolysis (47). Clamp FFAs and glycerol, which may serve as surrogate parameters for rate of lipolysis, and non-HDL cholesterol and triglycerides correlated inversely with clamp glycogen synthase activity. This suggests that the rate of lipolysis and dyslipidemia may play roles for glycogen synthase activity and NOGM in HIV lipodystrophy, as suggested previously for HIV-negative subjects (34). Moreover, Akt phosphorylation and fat distribution correlated with clamp FFAs and glycerol, which may support the concept that fat redistribution may correlate with Akt through enhanced rate of lipolysis. In that respect, a similar defect in the transmission of the insulin signal between PI 3-kinase and Akt, which, in turn, reduced the stimulation of glycogen synthesis, has been observed in palmitate-treated myotubes (48).

Here, we emphasize the importance of glycogen synthase activity on NOGM. However, studies from Shulman and colleagues (49) have recently challenged this concept by presenting data that suggest that the rate-limiting step in glycogen synthesis may not be glycogen synthase activity but glucose uptake over the cell membrane. The significant correlations between insulin-stimulated glycogen synthase activity and NOGM in the present study may therefore also have been explained by a defective glucose transport over the cell membrane, given the fact that both glycogen synthase and GLUT4 activation lie downstream of Akt in the insulin-signaling pathway.

A limitation of the present study might be that a healthy control group was not included. However, the level of insulin-stimulated NOGM in NONLIPO patients was similar to that demonstrated in a large cohort of HIV-negative healthy control subjects (50), which probably render NONLIPO patients a valid normal control group in respect to insulin signaling. Recently, in vitro data have suggested that one of the HIV-1 accessory proteins, the viral protein R, may impair insulin signaling involving Foxo activity (51). Research is therefore warranted to determine whether HIV-1 per se, eventually mediated through viral protein R, influences the insulin signal transduction cascade in vivo, although it has been known for a long time that HIV infection itself in nonwasting pre-HAART HIV-infected patients does not impair whole-body glucose disposal (52).

Based on these results and current evidence, we hypothesize that patients with HIV lipodystrophy may exhibit defects in insulin signaling attributable to factors related to the fat redistribution itself, and in addition, such patients are subject to acquired defects in insulin signaling induced acutely by circulating protease inhibitors. This “dual detrimental action” upon insulin signaling, of which lipodystrophy could account primarily for defects in signaling downstream from IRS-1–associated PI 3-kinase activity and protease inhibitors for more proximal defects, including impaired GLUT4 activation, might greatly enhance the risk of type 2 diabetes in HIV-infected patients developing fat redistribution.

The study was supported by grants from the Danish Medical Research Council, the Danish AIDS Foundation, the Novo Nordisk Foundation, the Danish Diabetes Association, the Copenhagen Muscle Research Centre, an Integrated Project Funded by the European Union (LSHM-CT-2004), the Copenhagen Hospital Cooperation Foundation, Hvidovre University Hospital Research Foundation, and the A.P. Møller Foundation for the Advancement of Medical Science. J.F.P.W. has received support from a Hallas Møller fellowship from the Novo Nordisk Foundation.

We thank Susanne Reimer and Lena Hansen at Hvidovre University Hospital and Betina Bolmgren and Jesper Birk at Copenhagen Muscle Research Centre for skilled technical assistance. Bo Falk Hansen, (Novo Nordisk), D. Grahame Hardie (Dundee University, Dundee, U.K.), Oluf Pedersen, (Steno Diabetes Center, Copenhagen, Denmark) and Kenneth Siddle (Cambridge University, Cambridge, U.K.) are acknowledged for their donation of antibodies.