|Home | About | Journals | Submit | Contact Us | Français|
Acute exposure to lipopolysaccharide (LPS) can cause hypoglycemia and insulin resistance; the underlying mechanisms however, are unclear. We set out to determine whether insulin resistance is linked to hypoglycemia through TLR4, MyD88 and NFκB, a cell signaling pathway that mediates LPS induction of the proinflammatory cytokine TNFα. LPS induction of hypoglycemia was blocked in TLR4−/− and MyD88−/− mice but not in TNFα−/− mice. Both glucose production and glucose utilization were decreased during hypoglycemia. Hypoglycemia was associated with the activation of NFκB in the liver. LPS inhibition of glucose production was blocked in hepatocytes isolated from TLR4−/− and MyD88−/− mice and hepatoma cells expressing an IκB mutant that interferes with NFκB activation. Thus, LPS-induced hypoglycemia was mediated by the inhibition of glucose production from the liver through the TLR4, MyD88, NFκB pathway, independent of LPS induced TNFα. LPS inhibition of glucose production was not blocked by pharmacologic inhibition of the insulin signaling intermediate PI3K in hepatoma cells. Insulin injection caused a similar reduction of circulating glucose in TLR4−/− and TLR4+/+ mice. These two results suggest that LPS and insulin inhibit glucose production by separate pathways. Recovery from LPS induced hypoglycemia was linked to glucose intolerance and hyperinsulinemia in TLR4+/+ mice, but not in TLR4−/− mice.
Insulin resistance is linked to the inhibition of glucose production by the TLR4, MyD88 and NFκB pathway.
Lipopolysaccharide (LPS) from Gram-negative bacterial infection can cause hypoglycemia and insulin resistance in both humans and mice; however, the underlying mechanisms are unclear (1-5). Low levels of LPS exposure through gastrointestinal tract and air-borne particles can also lead to insulin resistance (6-8). Thus, insulin resistance is decreased in mice by antibiotic treatment or housing in germ-free facilities (9, 10). Insulin resistance is also decreased by gene deletion of Toll-like receptor-4 (TLR4) or a cluster of differentiation 14 (CD14), a glycoprotein that binds to the extracellular portion of TLR4 (11-13). TLR4 is a plasma membrane protein that mediates LPS induction of inflammatory cytokines such as tumor necrosis factor alpha (TNFα) and interleukin-1 beta (IL-1β) (14). The main goals of this study were to (a) determine whether TLR4 and its downstream signaling targets mediate the induction of hypoglycemia by LPS and, if so, (b) determine whether the induction of hypoglycemia by TLR4 is linked to the development of insulin resistance.
LPS was postulated to cause hypoglycemia through the induction of the cytokines TNFα and IL-1β (15). Given that LPS does not induce hypoglycemia in IL-1α−/− and IL-1β−/− “double-knockout” mice, LPS induction of IL-1β is essential (16). Whether LPS induction of TNFα plays an essential role is unclear. LPS induction of TNFα causes hypoglycemia on the basis that (a) hypoglycemia coincides with LPS induction of TNFα, (b) the time course of hypoglycemia is similar when mice are injected with LPS or recombinant TNFα, and (c) LPS does not trigger hypoglycemia in C3H/HeJ mice, an inbred strain in which LPS injection does not induce TNFα based on a loss of function mutation in the TLR4 gene (15, 17). However, neutralizing antibodies against TNFα do not impair LPS induction of hypoglycemia, suggesting that TNFα does not play a role (18, 19).
LPS induction of IL-1 and TNFα is triggered by the interaction of Toll/interleukin-1 receptor (TIR) domains on the intracellular portion of TLR4 and TIR domains on myeloid differentiation factor 88 (MyD88) (20). However, TLR4 can also mediate its effects by interacting with TIR domains on other adaptors. LPS induction of interferon beta, for example, is mediated by the TIR domain-containing adaptor TRIF (21). Whether LPS induction of hypoglycemia is mediated by TLR4 and MyD88 is currently not known.
LPS may cause hypoglycemia by inhibiting glucose production through nuclear factor kappa B (NFκB), a downstream target of MyD88. NFκB is a heterodimer of two DNA-binding subunits, p50 and p65 (22). Under basal conditions, NFκB is retained in the cytoplasm by the inhibitor of NFκB (IκB). The translocation of NFκB to the nucleus is induced by phosphorylation of IκB. MyD88 activates NFκB by stimulating IL-1R-associated kinase (IRAK) which recruits the tumor necrosis factor receptor-associated factor (TRAF6) (21). TRAF6 phosphorylates IκB kinase beta (IKKβ) which leads to ubiquitination and degradation of IκB by SCF-b-TrCP and the 26S proteasome, respectively (23).
The role of NFκB in LPS induction of hypoglycemia is unclear. NFκB inhibits glucose production induced by glucocorticoid and glucagon (24). Acute LPS exposure lowers the expression and activity of rate limiting enzymes in glucose production, glucose-6-phosphatase (G6Pase) and phosphoenolpyruvate carboxykinase (PEPCK) (25, 26). IL-1 and TNFα suppress the gene expression of G6Pase and PEPCK by activating NFκB (27, 28). Activation of NFκB can also inhibit glucose production by stimulating the production of nitric oxide (NO) (26). NFκB increases NO production in hepatocytes by stimulating inducible nitric oxide synthase (iNOS) expression (29). However, it is not known whether LPS induction of hypoglycemia is mediated by NFκB.
Insulin is the most potent physiologic inhibitor of glucose production. LPS does not stimulate insulin secretion, but it is unknown whether LPS causes hypoglycemia by engaging the insulin signaling pathway (30). Insulin inhibits glucose production by suppressing G6Pase expression through phosphatidylinositol 3-kinase (PI3K) (31). Insulin activation of PI3K causes a transient increase of phosphatidylinositol-3,4,5-phosphate (PI3P) in cell membranes, which activates protein kinase B (PKB), also known as Akt, causing the phosphorylation and nuclear exclusion of Foxo1, a member of the Forkhead/winged helix family of transcription factors (32). LPS exposure leads to PI3K activation in neutrophils (33). It is unknown whether LPS causes hypoglycemia by inhibiting glucose production through PI3K.
Three to six month old male C57Bl/6 mice were used. TLR4−/− and MyD88−/− mice were obtained from Osaka University (14, 34). TNFα−/− mice were obtained from the Jackson Laboratory (35). All mice were fed a standard chow diet (73% carbohydrate, 18% protein, 4% fat, 5% ash) and housed in ventilated isolator cage systems in pathogen-free barrier facilities at the University of North Carolina. The colony was maintained on a 12-hour light/dark cycle at 23°C and 55% humidity. Experimental procedures were approved by the Institutional Animal Care and Use Committees (UNC) in accordance with criteria outlined in the Guide for the Care and Use of Laboratory Animals (NIH).
Food was removed from cages at 7 AM. Basal plasma glucose levels were measured at 10 AM; blood was sampled from the tail tip using a cut below the vertebrae. Glucose was measured by a hand-held glucose meter (One-Touch Ultra; Johnson and Johnson). Mice were weighed to determine the dose of LPS. LPS (E. coli serotype 0111:B4; Sigma) was given by a single intra-peritoneal injection. Blood was sampled at indicated times. Anesthesia and mouse restraining devices were not used for these procedures. TNFα levels were determined by enzyme-linked immune assay (B&D Biosciences).
Food was removed at 7 AM. Blood was sampled from the tail tip to determine basal glucose levels at 10 AM. Mice were weighed to determine the dose of insulin (Humalin R; Eli Lilly). Insulin was given by intra-peritoneal injection, and plasma glucose was measured at indicated time points after the injection.
Mice received LPS or saline by intraperitoneal injection. After complete recovery from LPS-induced hypoglycemia (48 hours), mice received an intraperitoneal injection of glucose (2.5 mg/kg), and plasma glucose was measured at indicated time points. To determine whether LPS exposure caused insulin resistance, mice were injected with 0.5, 1, 2 or 4 mg of glucose per gram body weight, and plasma insulin was measured 15 minutes after injection by radio-immune assay (LINCO).
At 8 hours after LPS or saline injection, while circulating levels of glucose were stable, trace amounts of 3H-glucose (D-3-3H-glucose; Amersham Biosciences) in 50-100μl saline were injected directly into the circulation of conscious, free moving mice through the ophthalmic plexus, as previously described (36). Blood samples (10-20 μl) were collected from the tail tip at 0, 3, 6, 9, 12, 15, 18 and 21 minutes after injection of 3H-glucose for glucose and radioactivity measurements. The disappearance of radioactivity in dried plasma measured by scintillation counting was similar to the disappearance of 3H-glucose measured by HPLC analysis on an Agilent 1100 HPLC System (Fig. 2D-E). The rate of glucose utilization was calculated by multiplying the fractional elimination constant, by the total pool of circulating glucose and dividing by body weight (37).
Nuclear extracts and electro mobility shift assays (EMSAs) were carried out as previously described (38). Snap frozen livers, isolated 0, 4 and 8 hours after LPS injection, were crushed in liquid nitrogen and Dounce homogenized in a 0.3 M sucrose buffer. Nuclei were isolated by layering and pelleting through 0.88 M and 1 M sucrose solutions. Nuclear extracts were incubated with a 32P-labeled oligonucleotide (5′-CAGGGCTGGGGATTCCCATCTCCCACAGTTTCACTTC-3′) corresponding to the NFκB binding site on the H2-κB gene promoter (38). Reaction mixes were loaded on nondenaturing 5% polyacrylamide gels in tris-glycine buffer (25 mM Tris, 190 mM glycine). Gels were dried under vacuum onto filter paper before being exposed against Kodak X-Omat film at −70 C with an intensifying screen. In competition assays, a 100-fold molar excess of unlabeled, double-stranded probe was added to the reaction mix before the addition of the probe. A competitor control was performed. Unlabeled oligonucleotide in 100-fold molar excess was added to the reaction mix before addition of the antiserum.
Liver was collected from 4-5 mice injected with LPS or saline 4 hours earlier. Reverse transcriptase II and Plat Quantitative PCR SuperMix-UDG (Invitrogen) and MyiQ Software were used on a Biorad Thermocycler. G6Pase forward and reverse primers were ATCTTCTGTTCCACGGAGAGG and CAGAGTGCTCAGGATGTTAAGG, respectively. Standards, in relative units of RNA, were 10−10, 10−11, 10−12, 10−13 g of liver RNA. R2= 0.818 where y = −2.112x+39.338; Tm was 82-83 C.
Glucose production was measured in isolated hepatocytes and hepatoma cells as previously described (36). Freshly isolated rat hepatocytes were plated at 105 cells per well in 12-well culture plates coated with rat tail collagen I (BD Biosciences). Cells were cultured in 25 mM glucose Dulbecco’s Modified Eagle Medium (DMEM; Caisson Laboratories) supplemented with 10% horse serum (HS). After 4 hours, the medium was changed to 25 mM glucose DMEM supplemented with 2.5% Fetal Bovine Serum (FBS) and 2.5% HS. After 8 hours, the medium was changed to 5 mM glucose/DMEM/5% HS with or without LPS (2.5 μg/mL) dissolved in dimethylsulphoxide (DMSO). After 8 hours, the media was changed to glucose-free and serum-free DMEM containing 5 mM of alanine, glycine, valine, sodium pyruvate and lactate (Sigma). Cells were incubated in the glucose-free medium for 8 hours. Glucose secreted into the medium was measured by colorimetric assay (Autokit Glucose CII). Cells were collected in RIPA lysis buffer and analyzed for total protein (BCA Assay Kit; Pierce). RIPA lysis buffer consisted of 50 mM Tris-HCl pH 7.5; 150 mM NaCl; 1% Triton-X 100/NP40; 0.5% sodium deoxycholic acid; 0.1% SDS; 1% PMSF; 1 mM Na3VO4; 0.1% Aproinin; and 0.1% phosphatase inhibitor (Sigma). Rat hepatoma cells (McArdle RH7777) were stored in liquid nitrogen at 106 cells/ml. Hepatoma cells were plated at 2.5 to 5.0 × 104 per well. Liver cells were exposed to LPS (2.5 μg/mL), human recombinant (100 nM from 100 U/ml stock where 28.85 U/mg, Elli Lily), LY294002 (33 μM; Calbiochem) or Lipid A (39).
Reverse transcriptase II and Plat Quantitative PCR SuperMix-UDG (Invitrogen) and MyiQ Software were used on a Biorad Thermocycler. G6Pase forward and reverse primers were ATCTTCTGTTCCACGGAGAGG and CAGAGTGCTCAGGATGTTAAGG, respectively The standards, in relative units of RNA, were 10−10, 10−11, 10−12, 10−13 g of liver RNA. Liver was collected from 4-5 mice injected with LPS or saline 8 hours earlier. R2= 0.818; Y=−2.112X+39.338; Tm was 82-83 C.
The dominant-negative IκB-SR mutant construct was used to generate an IκB-SR fragment flanked by EcoRI and ApaI sites (40). This was subcloned into pML2(EGFP-N1) to produce the IκB-SR-pML2(EGFP-N1) retroviral construct. HEK293 cells were transfected using a 3 to 1 ratio of IκB-SR-pML2(EGFP-N1) to pCL-ECO and the Fugene transfection reagent. HEK293 cells were incubated in DMEM/10% FBS at 37 C for 48 hours. The medium was then changed to DMEM/2.5% FBS and 2.5% HS, and cells were incubated at 32 C for 48 hours to allow retroviral production. The viral-containing medium (6 ml) was removed from the cells and mixed with 2.67 μg polybrene. The solution was then centrifuged at 3500 rpm for 5 min at room temperature. The viral supernatant (2 ml/well) was introduced to rat hepatoma cells (6-well plate) for 12-18 hours. The pML2(EGFP-N1) construct was used as a control for the effects of transfection. DNA and RNA were isolated (DNeasy and RNeasy Kits; Qiagen), and RNA was used to prepare cDNA using random hexamers (Superscript; Invitrogen). Integration of the retroviral IκB-SR construct into DNA and expression of IκB-SR mRNA were tested by PCR analysis.
Significant differences between the groups were examined using a two-tailed, two-sample unequal-variance Student’s t-test or Tukey-Kramer test. Significance was set at p < 0.05.
The role of TLR4 in LPS induced hypoglycemia was evaluated in TLR4−/− mice (Fig. 1A). TLR4+/+ and TLR4−/− mice were injected with various doses of LPS and plasma glucose was measured 4 hours after injection. Plasma glucose was reduced in TLR4+/+ mice but unchanged in TLR4−/− mice at 0.005, 0.05 and 0.5 mg LPS per kg body weight. The highest dose of LPS we tried, 5 mg/kg, was lethal. Thus, mice in subsequent studies were injected with 0.5 mg/kg LPS. The role of MyD88 in the LPS induced hypoglycemia was examined in MyD88−/− mice (Fig. 1B). MyD88+/+ and MyD88−/− mice were injected with LPS (0.5 mg/kg). Glucose was lower than preinjection values at 2, 3 and 4 hours after LPS injection in MyD88+/+ mice, but not in MyD88−/− mice. These results indicate that LPS induced hypoglycemia is mediated by TLR4 and MyD88.
The role of TNFα in LPS induced hypoglycemia was evaluated in TNFα−/− mice (Fig. 1C). TNFα+/+ and TNFα−/− mice were injected with LPS (0.5 mg/kg). TNFα was elevated by LPS in TNFα+/+ mice, but not in TNFα−/− mice. TNFα+/+ and TNFα−/− mice showed similar reductions in plasma glucose after LPS injection. Therefore, LPS induced hypoglycemia is not mediated by LPS induction of TNFα.
The acute effects of LPS on glucose production and utilization were examined in wild-type mice using the intravenous injection of 3H-glucose (Fig. 2A). LPS lowered the rate of glucose utilization to 9.2 + 1.5 mg/kg/min, compared to 19.1 + 0.9 mg/kg/min in controls (Fig. 2B). Intravenous injection of 3H-glucose and the sampling of tail blood to determine the rate of glucose utilization did not alter the concentration of circulating glucose over the 21-minutes used to measure glucose production and utilization (Fig. 2C). The disappearance of radioactivity in plasma, measured by scintillation counting, showed single phase (Fig. 2D) and was similar to the disappearance of 3H-glucose measured by HPLC analysis (Fig. 2E). Since circulating levels of glucose were stable during the measurement of glucose utilization, the rate of glucose utilization also reflected the rate of glucose production. Thus, LPS induced hypoglycemia reduced the rates of glucose utilization and glucose production.
In order to determine whether LPS inhibited glucose production in the liver directly, we measured glucose production in isolated rat hepatocytes exposed to 2.5 mg/ml LPS (Fig. 2F). Glucose production was measured over 8 hours of LPS exposure. LPS suppressed glucose production in isolated hepatocytes by 50%. LPS did not affect glucose production in hepatocytes isolated from TLR4−/− and MyD88−/− mice (data not shown). These results suggest that LPS-induced hypoglycemia is mediated by direct inhibition of glucose production by TLR4 and MyD88.
We postulated the involvement of NFκB in the signaling pathway that inhibits glucose production. To test this hypothesis, mice livers were isolated 4 hours after injection of LPS (0.5 mg/kg) or saline and nuclear translocation of NFκB was measured by electromobility shift assay (EMSA). A double stranded 32P-labeled oligonucleotide probe corresponding to the binding element of NFκB in the promoter region of the H2-κB gene was used (38). EMSA revealed an increase in nuclear binding of NFκB after LPS injection injection (Fig. 3A). The elevation of NFkB activation 4 hours after LPS injection was also associated with a 5- to 10-fold reduction in G6Pase mRNA by RT-PCR (p<0.05).
The rat hepatoma cell line, McArdle RH7777, was used to determine whether LPS inhibition of glucose production depended on NFκB. Hepatoma cells were transduced by a retroviral vector encoding green florescent protein (GFP) or GFP and a degradation-resistant IκB mutant (IκB-SR) that blocks nuclear translocation of NFκB (40). DNA integration and gene expression of the IκB-SR construct were confirmed by PCR (Fig. 3B). Glucose production was measured over 8 hours of LPS (2.5 μg/ml) exposure. LPS inhibition of glucose production was blocked by IκB-SR (Fig. 3C). These results support that LPS inhibition of glucose production depends on the activation of NFκB. LPS did not inhibit glucose production by damaging or shutting down hepatocytes and hepatoma cells. In our experience, adherent hepatocytes and hepatoma cells undergoing necrosis or apoptosis by inhibition of mTOR, for example, show reduced protein content. We did not observe a significant change in protein content after 8-16 hours of LPS exposure (2.5 μg/ml). Furthermore, Hoescht and trypan blue staining did not reveal any aberrant decrease the health status of the cells under the conditions described.
Two experiments were performed to determine whether LPS and insulin inhibit glucose production by different pathways. Rat hepatoma cells were either untreated or treated with insulin or insulin and LY294002 and exposed to LPS (Fig. 4A). LY294002 is an inhibitor of the insulin signaling intermediate PI3K. LY294002 did not interfere with LPS inhibition of glucose. Furthermore, the combination of insulin and LPS showed additive effects. TLR4+/+ and TLR4−/− mice were injected with insulin (0.75 U/kg), and plasma glucose production was measured after 15, 30, 45, 60, 90 and 120 minutes (Fig. 4B). The absence of TLR4 did not alter the acute effect of insulin injection on glucose. Therefore, LPS and insulin may inhibit glucose production by different pathways.
TLR4+/+ and TLR4−/− mice received an intraperitoneal injection of glucose 48 hours after injection of LPS. This period (48 hours) allowed complete recovery from LPS-induced hypoglycemia in TLR4+/+ mice. Plasma glucose was measured at 0 and 15, 30, 60, 90 and 120 minutes after the injection of glucose (Fig. 5A). TLR4+/+ and LPS-induced hypoglycemia was associated with glucose intolerance at 15 and 30 minutes after glucose injection (2.5 g/kg). Glucose intolerance is often associated with insulin resistance. Therefore, to determine whether insulin resistance is linked to LPS-induced hypoglycemia, insulin was measured 15 minutes after the injection of glucose (0.5, 1.0, 2.0 or 4.0 g/kg). Hyperinsulinemia, a hallmark of insulin resistance, was evident in TLR4+/+ mice previously exposed to LPS at each glucose dose (Fig. 5B). The absence of glucose intolerance and hyperinsulinemia in LPS-injected TLR4−/− mice suggests that insulin resistance is a secondary effect of LPS-induced hypoglycemia through the TLR4, MyD88, NFκB pathway.
This study provides the first evidence that LPS-induced hypoglycemia is mediated by the inhibition of glucose production through the TLR4, MyD88, NFκB pathway. The inability of LPS to induce hypoglycemia in TLR4−/− and MyD88−/− mice clearly shows that LPS induction of hypoglycemia is mediated by TLR4 and MyD88. LPS induction of hypoglycemia was not impaired in TNFα−/− mice, suggesting that IL-1 may be the only essential cytokine for LPS induction of hypoglycemia (16). MyD88 may be essential for the induction of hypoglycemia for both TLR4 and the IL-1 receptor as the cytosolic regions of TLR4 and the IL-1 receptor contain TIR domains that can interact with TIR domains on MyD88 (14, 20, 34, 41).
LPS injection transiently decreased both glucose production and glucose utilization. Although numerous tracer studies have shown that acute exposure to LPS inhibits glucose production, there are conflicting reports about the effect of LPS on glucose uptake (42, 43). In the present study, the effect of LPS on glucose production and utilization was measured during LPS induced hypoglycemia. The importance of circulating glucose levels for glucose uptake was previously demonstrated; glucose uptake was reduced under LPS-induced hypoglycemia but increased when euglycemia was restored by glucose infusion (44). Therefore, under physiologic conditions in vivo, LPS induced hypoglycemia is associated with the reduction of glucose production and utilization.
TLR4 and MyD88 inhibition of glucose production in vivo may depend on NFκB since we showed that (a) nuclear translocation of NFκB in the liver was increased by LPS injection, and (b) suppression of NFκB activity in hepatoma cells, through an IκB mutant, blocked LPS inhibition of glucose production. Two other targets of TLR4 and MyD88, p38 mitogen-activated protein kinase (p38 MAPK) and c-Jun NH2-terminal kinase (JNK), are linked to high glucose production and hyperglycemia (45, 46). Therefore, NFκB is a downstream target of TLR4 capable inducing hypoglycemia by inhibiting liver glucose production.
LPS inhibition of glucose production was not influenced by PI3K activity. PI3K inhibition in hepatoma cells has also failed to interfere with the suppression of G6Pase expression by TNFα (27). These observations suggest that LPS, IL-1 and TNFα regulate glucose metabolism in the liver by a different signaling mechanism than insulin. Crosstalk between the LPS and insulin signaling pathways may occur at the level of gene expression by NFκB and Foxo-1 (32). Crosstalk between the TLR4 and insulin signaling pathways has been reported in myeloid cells (47). LPS induced decreases in blood flow can also contribute to low glucose utilization after LPS injection (48).
Insulin resistance was evident after LPS exposure in vivo. Insulin resistance may be part of a systemic counter-regulatory response to LPS-induced hypoglycemia aimed at restoring circulating glucose. The mechanism may be linked to the elevation of proinflammatory cytokines as well as counter-regulatory hormones such as catecholamines, glucagon and glucocorticoids (8, 49-51). This is consistent with the observation that insulin infusion increases the cytokine and endocrine responses to LPS in humans (52). It should be noted that the induction of hypoglycemia does not always result in insulin resistance. For example, injection of adiponectin, a hormone produced by adipocytes, lowers circulating glucose by enhancing insulin sensitivity (53, 54). Along these arguments, we did not measure the activity of AMPK or any of its down stream targets. The reduction of glucose production through AMPK is mediated by an increase in insulin sensitivity (36, 55-57).It is unlikely that LPS invokes AMPK to trigger hypoglycemia because LPS invokes insulin resistance.
LPS mediates its effects by similar pathways in mice and humans. For example, TLR4 mutations lower LPS responsiveness in both (14, 58). The suppression of the TLR4, MyD88, NFκB pathway in humans offers a therapeutic strategy for the management of circulating glucose levels and insulin resistance. This is consistent with the elevation of TLR4 activity in obese type 2 diabetics and the anti-diabetic effects of aspirin and salicylate through inhibition of IKKβ (6, 59, 60).
The lipid moiety of LPS, called as Lipid A, induces hypoglycemia similar to LPS (61). In the present study, Lipid A inhibited glucose production in hepatocytes and hepatoma cells (data not shown). These observations suggest that derivatives of Lipid A that block TLR4 can have therapeutic potential.
In summary, the present study shows that insulin resistance is a secondary effect of LPS-induced hypoglycemia through the TLR4, MyD88, NFκB pathway.