Insulin Suppresses Endotoxin-Induced Oxidative, Nitrosative, and Inflammatory Stress in Humans

Nineteen normal-weight (BMI 20–25 kg/m²) healthy male subjects aged between 20 and 33 years (mean age 26 ± 3 years) were recruited for this study. After an overnight fast, nine subjects were injected intravenously with 2 ng/kg of LPS prepared from Escherichia coli along with saline at 100 ml/h. The other 10 were infused with insulin (2 units/h) 1 h before the LPS injection along with 100 ml/h of 5% dextrose coinfused with insulin to maintain normoglycemia. Insulin/dextrose or saline infusions were continued for 6 h after the LPS injection while subjects were in the fasting state to avoid the potential proinflammatory effect of a meal (13). They were then provided with a 900-calorie meal at 6:00 p.m. after which they ate nothing till the next morning. Subjects were monitored for vital signs (temperature, pulse, blood pressure, headaches, body aches, and chills) for 24 h after the LPS injection. Blood samples were collected 1 h before the LPS injection and at 0, 1, 2, 4, 6, and 24 h after the injection. The protocol was approved by the internal review board of the State University of New York at Buffalo, and written consent was obtained from all subjects.

Blood samples were collected and polymorphonuclear leukocytes (PMNLs) were isolated and ROS generation was measured as described previously (14). The intra-assay coefficient of variation (CV) for ROS generation is 8%.

Plasma concentrations of glucose were measured by a 2300 STAT Plus glucose analyzer (YSI, Yellow Springs, OH). Insulin concentrations were measured from plasma samples using an ELISA kit (Diagnostics Systems Laboratories, Webster, TX). FFA concentrations were measured using the Half-Micro calorimetric kit from Roche Diagnostic (Indianapolis, IN). NOM (NO₂/NO₃) were assayed by the Griess reaction (R&D Systems, Minneapolis, MN), and thiobarbituric acid–reacting substances (TBARS) were assayed by spectrofluorometry with a kit from Zeptometrix (Buffalo, NY). The CVs for these assays ranged from 2 to 7 and 4 to 11% for intra- and interassay variations, respectively. Homeostasis model assessment of insulin resistance (HOMA-IR) was calculated according to the formula: (fasting insulin [microunits per milliliter) × fasting glucose [millimoles per liter])/22.5.

Commercially available ELISAs were used to measure concentrations of circulating cytokine concentrations (R&D Systems), CRP (American Diagnostica, Stamford, CT), visfatin (Phoenix Pharmaceuticals, Belmont, CA), myoglobin (Life Diagnostics, West Chester, PA), HMG-B1 (IBL Transatlantic), and LBP (Cell Sciences, Canton, MA). The CVs for these assays ranged from 3 to 8 and 6 to 11% for inter- and intra-assay variations, respectively.

Statistical analysis was conducted using SigmaStat software (SPSS, Chicago, IL). All data are means ± SEM. Statistical analysis from baselines was preformed using Holm-Sidak one-way repeated-measures analysis of variance (RMANOVA). Dunnett's two-factor RMANOVA method was used for multiple comparisons between different groups. Paired t tests were used to compare changes from baseline at 24 h.

LPS injection induced an increase in temperature from 97.7 ± 0.5°F to a peak of 100.9 ± 0.9°F at 4 h, systolic blood pressure increased from 113 ± 11 mmHg to a peak of 129 ± 13 mm Hg at 2 h, and pulse rate increased from 63 ± 7/min to a peak of 97 ± 9/min at 6 h and was not affected by insulin (Table 1). Headaches, chills, and body ache scores increased after LPS injection with peaks at 1 to 2 h with insulin infusion reducing body ache score significantly (Table 1) without affecting headaches and chills.

After LPS, total leukocyte count increased from a baseline of 4,300 ± 900 at −1 h to a peak of 11,800 ± 1,200 cells/mm³ at 6 h and was still elevated (7,000 ± 1,100 cells/mm³) at 24 h, mainly attributable to polymorphonuclear leukocytosis (Table 1). Monocytes and lymphocytes fell rapidly from 6 ± 0.4 to 0.5 ± 0.1% and from 39 ± 5 to 3 ± 0.8%, respectively. Insulin infusion did not alter this pattern. The marked reduction in monocytes and lymphocytes prevented us from examining ROS generation and other cellular markers/mediators in the mononuclear cell fraction.

In the group receiving insulin, insulin concentrations increased by up to fourfold (P < 0.001) (Table 1). In the control group, insulin concentrations did not change in the first 6 h but were significantly higher than baseline at 24 h (P < 0.05) (Table 1). There was no significant change in glucose concentrations. Consistent with that finding, HOMA-IR increased significantly at 24 h from 1.22 ± 0.24 vs. 2.36 ± 0.39 in the control group (Table 1). After LPS injection, there was a significant increase in plasma FFA concentration. Insulin infusion prevented this increase (Table 1). Serum triglyceride concentration fell significantly in both groups (Table 1). LDL cholesterol, VLDL cholesterol, and HDL cholesterol concentrations did not change.

LPS injection induced an increase in ROS generation by PMNLs of 200 ± 42% over the basal with a peak at 1 h and another peak at 4–6 h. Insulin infusion reduced (Fig. 1 A) ROS generation throughout the infusion period (P < 0.05 by two-factor RMANOVA).

LPS injection induced a rapid increase in TBARS concentration from 1.29 ± 0.29 to 2.15 ± 0.41 μmol/l (P < 0.01) at 1 h, with a return to baseline at 2 h (Fig. 1 B). This pattern was observed in each of the LPS-injected subjects. Insulin infusion totally prevented this increase.

Plasma NOM concentration increased rapidly after the injection of LPS at 1 h, peaked to 75 ± 24% over the baseline at 2 h (from 29.4 ± 2.6 to 47.7 ± 5.4 μmol/l), and declined to the baseline by 4 h. There was a secondary rise in NOM concentration at 6 h (Fig. 1 C). With insulin infusion, the LPS-induced increase in NOM was totally prevented and in fact there was a small but significant decrease by 16 ± 10% below the baseline in plasma NOM concentrations (from 31.7 ± 2.8 to 26.9 ± 2.6 μmol/l).

Plasma concentrations of MIF increased significantly after LPS injection at 1 h with a secondary increase at 4–6 h from 727 ± 111 to 1,345 ± 145 ng/ml at 6 h (Fig. 2,A). Insulin infusion with LPS prevented the LPS-induced increase of MIF during the initial increase and suppressed it significantly during the secondary increase (from 700 ± 137 to 1,080 ± 87 ng/ml at 6 h). The plasma concentration of TNF-α increased at 1 h, peaked between 1 and 2 h (P < 0.001, Table 1), and declined thereafter, reaching near the baseline at 24 h. IL-6 increased at 1 h, peaked between 2 and 4 h (P < 0.001) (Table 1), and declined toward the baseline by 24 h. CRP concentration increased at 6 h and was still elevated at 24 h (from 1.2 ± 0.2 to 15.2 ± 5.7 mg/l, P < 0.001) (Table 1). Insulin infusion did not alter the LPS-induced increases in TNF-α, IL-6, or CRP concentrations.

Plasma visfatin concentrations increased significantly after LPS injection starting at 4 h, peaked at 6 h (87 ± 37% above baseline, from 11.4 ± 1.2 to 19.2 ± 2.4 ng/ml, P < 0.001) (Fig. 2,B), and was maintained at that level for 24 h. When insulin was infused before LPS injection, visfatin concentrations fell significantly from 11.9 ± 1.5 to 7.7 ± 1.1.0 ng/ml (P < 0.001) (Fig. 2,B) at 4 h and were significantly different from that in the control group. LPS injection also caused a significant increase in resistin concentrations, which started at 2 h after the injection, peaked at 6 h (223 ± 25% above baseline, from 8.1 ± 1.1 to 24.7 ± 4.7 ng/ml P < 0.001) (Table 1) and continued to be higher than baseline at 24 h. Plasma LBP concentrations increased gradually after LPS injection and were higher by 48 ± 19% above the baseline at 24 h (P < 0.05) (Table 1 ). Insulin did not cause any significant change in the LPS-induced increases in resistin or LBP concentrations.

The plasma concentration of myoglobin also increased significantly from 22.3 ± 4.1 to 32.8 ± 5.3 ng/ml at 4 h and to 39 ± 5.8 ng/ml (P < 0.05) at 24 h after LPS injection (Fig. 3,A). Insulin prevented the increase in myoglobin concentrations. HMG-B1 concentrations in plasma increased after LPS injection starting at 1 h and peaked at 6 h (Fig. 3 B), whereas insulin infusion had no effect on the LPS-induced increase in HMG-B1concentrations.

Our data indicate several novel observations on the effects of LPS and insulin. They demonstrate for the first time the increase in ROS generation, TBARS, NOM, resistin, visfatin, myoglobin, and HMG-B1 concentrations in humans in vivo after an LPS injection. They also show for the first time that insulin infusion reduces or totally prevents the LPS-induced increases in ROS generation and the concentrations of TBARS, NOM, MIF, visfatin, and myoglobin. The relevance of each of these novel effects is discussed below.

The LPS-induced increase in ROS generation by PMNL and TBARS concentration is evidence of marked oxidative stress. Insulin suppressed the increase in ROS generation significantly while eliminating the increase in TBARS altogether. After LPS, NOM peaked at 2 h and returned to the baseline by 4 h. The insulin infusion eliminated the increase in the NOM concentration. These actions of insulin were independent of any change in glucose concentrations.

Elevated plasma NOM concentrations and inducible nitric oxide synthase expression in the liver have previously been shown to be suppressed by insulin infusions in patients in ICUs (15). In this study, the plasma concentrations of NOM in the highest quartile were associated with seven times greater mortality than those in the lowest quartile. Thus, the NOM concentration could be an important predictor of morbidity and mortality in the ICU setting (16). Whether this effect on mortality is directly related to an excess of nitric oxide generation or whether the increased NOM levels are merely markers of the intensity of systemic inflammation is not clear. Either way, the rapid induction of an increase in NOM by LPS and its total prevention by insulin are important and relevant observations.

The biphasic increase in MIF after LPS was reduced by insulin infusion. On the other hand, insulin infusion did not prevent LPS-induced increases in plasma TNF-α, IL-6, intercellular adhesion molecule-1, and MCP-1 concentrations. This observation is in contrast to our previous observation in patients with obesity and type 2 diabetes in whom insulin suppressed these mediators. CRP concentrations began to increase at 6 h after the LPS injection at which time the infusion of insulin ended. Clearly, further studies with higher doses and for longer periods of insulin infusion are required.

LPS also induced an increase in plasma FFA concentrations within 1 h, which continued for 24 h, consistent with a potent lipolytic effect of LPS. This increase was totally inhibited by insulin. The suppressive effect of insulin FFA is important because FFAs may induce oxidative and inflammatory stress (17).

Our data also show for the first time that LPS injection in the human induced an increase in plasma concentrations of resistin and visfatin. Insulin infusion resulted in the suppression of the increase in visfatin but not resistin. The LPS-induced increase in visfatin and the prevention of this increase with insulin are of interest in terms not only of the acute LPS-induced inflammation but also of the chronic inflammation in atherosclerotic plaques because such plaques are known to contain LPS- and TLR-4-expressing macrophages (18). Such plaques also express visfatin, which may be secreted locally in response to the LPS-TLR-4 interaction (10). It is, therefore, of interest that insulin suppresses the LPS-induced increase in visfatin and has previously been shown to suppress TLR-4 expression (5). Resistin is also known to stimulate the secretion of proinflammatory cytokines, and the evidence that its concentration increases after LPS in the human in vivo establishes it as a proinflammatory mediator (19).

Plasma concentrations of LBP also increased after LPS injection, demonstrated for the first time. The increase started late, at 6 h, like that of CRP and the previously described increase in procalcitonin (20) and continued overnight at 24 h. The increase in the concentration of LBP after LPS is important because LBP facilitates the binding of LPS to its receptor, TLR-4. As with CRP, the increase in LBP was not affected by insulin, possibly because both increased at 6 h and the insulin infusion was stopped at that time.

It is of interest that although plasma glucose concentrations did not alter significantly, insulin concentrations and HOMA-IR increased significantly 24 h after LPS injection in concert with the induction of profound inflammation. This result is consistent with the recent observation that the injection of LPS (3 ng/kg) in normal subjects induced insulin resistance as measured by frequently sampled intravenous glucose tolerance and HOMA-IR (21).

Our data also show for the first time that LPS induces an increase in HMG-B1 concentration in humans. Insulin did not alter this increase. HMG-B1 is a nuclear protein that binds to histones to promote proinflammatory gene transcription. It can be released from damaged, necrotic tissues. Circulating HMG-B1 acts like a proinflammatory cytokine through its binding to the receptor for advanced glycation end products (22).

The increase in plasma myoglobin concentrations after LPS and its inhibition by insulin is important because it signifies damage to the skeletal muscle and possibly the myocardium. Consistent with this observation, we have previously shown a reduction in the increase in myoglobin concentrations in patients with myocardial infarction treated with intravenous insulin infusions. This finding is suggestive of a cytoprotective effect of insulin.

The mechanisms underlying the effects of insulin observed in this report are probably related to several of our previous observations. Insulin suppresses the expression of TLR-4, the receptor for LPS, to reduce the activity of the major proinflammatory transcription factor, NFκB (3,5). Insulin has also been shown to suppress ROS generation and the expression of the p47 subunit of NADPH oxidase. Insulin has been shown previously to suppress inducible nitric oxide synthase expression and the plasma NOM concentration (15).

Consistent with our observations are those of Jeschke et al. (23), who demonstrated that compared with control subjects, patients with severe burns given insulin had lower MIF and other proinflammatory cytokines and CRP concentrations with a tendency toward higher IL-10 concentrations. Insulin infusion with maintenance of normoglycemia has been shown to reduce mortality and morbidity in patients in a surgical ICU (24) and in patients in a medical ICU (25), whose stay in the ICU is for longer than 3 days.

In summary, the injection of LPS in the human induces an immediate increase in ROS generation by PMNLs and in plasma concentrations of TBARS, NOM, FFA, MIF, resistin, visfatin, LBP, and myoglobin. The concomitant infusion of insulin induces a significant reduction in ROS generation and the total prevention of the increase in TBARS, NOM, and FFA concentrations. These actions were associated with a significant reduction in the magnitude of increase in MIF, myoglobin, and visfatin concentrations independently of any change in plasma glucose concentrations. On the other hand, insulin was not able to prevent or reduce the magnitude of increase in plasma concentrations of proinflammatory cytokines like TNF-α, IL-6, or MCP-1. Clearly, the effect of more prolonged infusions and higher doses of insulin needs to be investigated.

P.D. is supported by the National Institutes of Health (grants R01-DK-069805 and R01-DK-075877) and the American Diabetes Association (grant 08-CR-13) and by Merck, sanofi-aventis, Amylin, GlaxoSmithKline, and Solvay Pharmaceuticals.

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

P.D. developed the hypothesis, analyzed and interpreted data, and wrote the manuscript. H.G. analyzed samples, performed statistical analysis of laboratory data, provided the diagrammatic presentation of data, and wrote the manuscript. A.B. developed the hypothesis and performed clinical experiments and patient care. K.K. performed laboratory measurements and analyzed laboratory data. C.L.S. analyzed laboratory data. S.D. provided clinical care and analyzed data. A.C. provided clinical care, analyzed data, and wrote the manuscript.