| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Vascular Biology |
From the Vascular Biology Unit (N.C., M.M.B., X.H., C.S., A.I., R.A.C.), Diabetes and Metabolism Unit (Y.I.), and Myocardial Biology Unit Evans Department of Medicine (D.P.), Whitaker Cardiovascular Institute, Boston University School of Medicine, Boston, Mass.
Correspondence to Richard A. Cohen, Director, Vascular Biology Unit, 650 Albany Street, X704, Boston, MA 02118. E-mail racohen{at}bu.edu
| Abstract |
|---|
|
|
|---|
Methods and Results OxLDL transiently increased phosphorylation of Erk and Akt within 5 minutes, but 60 minutes later, resulted in decreased insulin-induced Akt phosphorylation. OxLDL promoted a 2- to 5-fold increase in oxidant generation as measured by dihydrorhodamine or dihydroethidium oxidation that was ascribed to peroxynitrite. Exogenous peroxynitrite (25 to 100 µmol/L) or oxidized glutathione mimicked the effects of oxLDL. OxLDL increased the S-glutathiolation of p21ras, and adenoviral transfection with either a mutant p21ras (C118S) lacking the predominant site of S-glutathiolation or a dominant-negative mutant restored insulin-induced Akt phosphorylation. The requirement for oxidant-mediated S-glutathiolation and activation of p21ras in mediating insulin resistance was further implicated by showing that insulin signaling was restored by Mek inhibitors or by overexpression of glutaredoxin-1. Furthermore, oxLDL increased Erk-dependent phosphorylation of insulin receptor substrate-1 serine-616 that was prevented by inhibiting oxidant generation, Erk activation, or by the p21ras C118S mutant.
Conclusions This study provides direct evidence for a novel molecular mechanism by which oxidants can induce insulin resistance via S-glutathiolation of p21ras and Erk-dependent inhibition of insulin signaling.
Bovine aortic endothelial cells were exposed to oxLDL. OxLDL induced peroxynitrite production and triggered redox-activation of p21ras via S-glutathiolation of its cysteine-118. Oxidant activation of p21ras was responsible for Erk-dependent phosphorylation of IRS-1 serine-616, which promoted endothelial insulin resistance.
Key Words: p21ras peroxynitrite S-glutathiolation glutathione insulin resistance oxidized LDL
| Introduction |
|---|
|
|
|---|
Hyperlipidemia,13 hyperglycemia,14,15 oxLDL,6 lysophosphatidyl choline,16 and other factors that are elevated in metabolic syndrome increase the generation of oxidants in vascular cells from multiple sources including mitochondria,17 NADPH oxidase,1820 and nitric oxide synthase.15,21,22 Supporting a potential direct role of oxidants in inhibition of insulin signaling, low concentrations of H2O2 were shown to rapidly inhibit insulin-induced Akt phosphorylation in smooth muscle cells.23 Angiotensin II, which increases oxidants via NADPH oxidase, increases Erk- and Jun kinase-dependent phosphorylation of IRS-1 serine-616 and -312, respectively, and inhibits insulin-induced Akt in endothelial cells.8 However, a molecular mechanism by which oxidants can cause insulin resistance is unknown.
We recently demonstrated that increased levels of oxidants in both endothelial and smooth muscle cells can directly activate p21ras by S-glutathiolation of a reactive thiol on cysteine-118 and trigger downstream phosphorylation of Erk and Akt.24,25 Therefore in the current study, we investigated if oxLDL could regulate insulin signaling as a direct result of increasing oxidant-mediated p21ras activation. We demonstrate that peroxynitrite arising from oxLDL exposure S-glutathiolates and activates p21ras, and that the resulting Mek-dependent Erk activation promotes phosphorylation of IRS-1 serine-616 and interferes with insulin-induced Akt phosphorylation. These results provide a molecular mechanism directly linking endogenous oxidant generation and insulin resistance.
| Experimental Procedures |
|---|
|
|
|---|
| Results |
|---|
|
|
|---|
|
To test the specificity of oxidant-induced inhibition of insulin signaling, the impact of oxLDL on vascular endothelial growth factor (VEGF)-induced Akt phosphorylation was tested. Although, oxLDL inhibited insulin-induced Akt phosphorylation, it did not affect Akt phosphorylation by VEGF (50 ng/mL, 15 minutes; supplemental Figure II). Thus, the mechanism of inhibition of the response to insulin involves specific components of the insulin signaling pathway.
OxLDL-Induced Insulin Resistance Is Mimicked by Peroxynitrite
We and others previously demonstrated that oxLDL exerts an important part of its cellular effects including activation of p21ras through generation of oxidants, especially peroxynitrite,17,24 and we previously showed that oxLDL-induced activation of Erk kinase is mediated by the acute production of peroxynitrite. To further evaluate the oxidant response to oxLDL over the longer time course of these studies, we determined peroxynitrite generation in BAEC by measuring their capacity to oxidize dihydrorhodamine to its fluorescent product, rhodamine, over 1 hour. Cells exposed to increasing concentrations of oxLDL (10 to 100 µg/mL) showed a concentration-dependent production of peroxynitrite up to a 3.4±0.3-fold increase over basal after 1 hour exposure to 100 µg/mL (supplemental Figure III). MnTBAP (50 µmol/L) or L-NG-arginine methyl ester (L-NAME) significantly decreased the oxidant generation caused by oxLDL consistent with peroxynitrite generation and oxidation of dihydrorhodamine.
To confirm the role of oxidants in the insulin resistance caused by oxLDL, BAEC were exposed to a bolus addition of authentic peroxynitrite (100 µmol/L, Figure 1B) or peroxynitrite slowly generated for 1 hour by coincubation with xanthine oxidase and spermine-NONOate (Figure 1C). These effects were compared with a bolus addition of hydrogen peroxide (10 to 100 µmol/L; Figure 1C). As presented in Figure 1C, peroxynitrite flux as low as 0.13 µmol/L/min (xanthine oxidase 2 mU/mL and spermine-NONOate 5 µmol/L)28 was sufficient to significantly inhibit insulin-induced Akt phosphorylation, but hydrogen peroxide at a concentration of 100 µmol/L failed to have any effect (Figure 1C).
Lysophosphatidylcholine and the PGE2 Agonist, Butaprost, Mimic oxLDL-Induced Insulin Resistance in BAEC
To validate oxidant-induced insulin resistance and to exclude potential nonspecific effects of oxLDL, BAEC were exposed to lysophosphatidylcholine (LPC), one of the active components of oxLDL.16,29 To measure the production of oxidants, BAEC were exposed for 1 hour to LPC (500 nmol/L) or oxLDL (100 µg/mL), rinsed twice with phosphate-buffered saline and then incubated 1 hour with dihydroethidium (5 µmol/L). After cell lysis, the product of the reaction of DHE with superoxide anion (2-OH-ethidium) was separated by high-performance liquid chromatography from the fluorescent nonspecific product, ethidium, and quantified.27 Normalized to ethidium, which did not change with the treatments, both LPC and oxLDL triggered a significant increase in superoxide anion release (supplemental Figure IV). Similar to oxLDL, LPC also significantly increased dihydrorhodamine oxidation (supplemental Figure V) that was inhibited by L-NAME or MnTBAP, consistent with peroxynitrite generation. In addition, LPC significantly decreased insulin-induced Akt phosphorylation (Figure 2A).
|
The prostaglandin (PG) E2 receptor has recently been described as a target of oxidized lipids in minimally oxLDL by showing that its effects were mimicked by the PGE2 receptor agonist, butaprost, and were prevented by the PG receptor antagonist, AH6809.30 Butaprost (1 µmol/L, 1 hour) significantly decreased insulin-induced Akt phosphorylation that was prevented by AH6809 (20 µmol/L, Figure 2B). Furthermore, pretreatment with AH6809 (20 µmol/L, 1 hour) significantly restored the inhibition of insulin signaling caused by oxLDL (Figure 2C). Together, these data support the potential role of PG receptors including those for PGE2 (AH6809 also blocks DP receptors) in mediating insulin resistance caused by oxLDL in BAEC.
Oxidant Activation of p21ras Promotes Insulin Resistance Via Erk Activation
Because the inhibition of insulin-induced phosphorylation of Akt correlated inversely with the Erk phosphorylation caused by oxLDL (Figure 1A), further studies were designed to address the hypothesis that p21ras mediates the oxidant-induced Erk activation and inhibition of insulin-induced Akt signaling by oxLDL.
To determine the role of Mek, which is essential for p21ras signaling to Erk, in the impaired insulin signaling caused by oxLDL or peroxynitrite, BAEC were treated with PD98059 (20 µmol/L, 1 hour), a specific Mek-1 inhibitor, or U0126 (20 µmol/L, 1 hour), a Mek-1 and Mek-2 inhibitor. PD98059 or U0126 inhibited Erk phosphorylation for 1 hour after peroxynitrite (Figure 3A) or oxLDL (Figure 3B), and the Mek inhibitors significantly restored insulin-induced Akt phosphorylation after peroxynitrite or oxLDL (Figure 3A and 3B).
|
Because oxLDL and peroxynitrite-induced Erk phosphorylation is dependent on S-glutathiolation of cysteine-118 of p21ras,24 the role of oxidant-activated p21ras in insulin resistance following exposure of BAEC to oxLDL was tested by adenoviral transfection of a p21ras C118S mutant. As expected, in cells transfected with LacZ, oxLDL dramatically inhibited Akt phosphorylation by insulin (Figure 3C). However, transfection with the p21ras C118S mutant significantly prevented the effect of oxLDL, restoring Akt phosphorylation to a state not significantly different from control (Figure 3C). Also, the inhibition of insulin-induced Akt phosphorylation after peroxynitrite was prevented by adenoviral transfection of an N17 dominant-negative form of p21ras (p21ras DN; supplemental Figure VI). In addition, consistent with a role for p21ras, the farnesyl transferase inhibitor, FTI 277 (5 µmol/L, 18 hour) also partially restored insulin-induced Akt phosphorylation, without affecting basal Akt phosphorylation (supplemental Figure VII).
Oxidant activation of p21ras-mediated signaling and S-glutathiolation in smooth muscle cells is prevented by overexpression of glutaredoxin-1,25 one of the enzymes known to reduce protein-glutathione mixed disulfides. In the present study, adenoviral overexpression of glutaredoxin-1 significantly prevented the impaired insulin-induced Akt phosphorylation after exposure to oxLDL (Figure 3C), implicating S-glutathiolation of p21ras cysteine-118 in the mechanism of insulin resistance. No significant effects on insulin-induced Akt phosphorylation were seen of either the p21ras C118S mutant or GRX-1 transfection alone, indicating that they specifically reversed the effects of oxLDL.
S-glutathiolation of p21ras Mediates Inhibition of Insulin Signaling
S-glutathiolation of cysteine-118, as demonstrated with a biotin-labeled GSH, is associated with an increase in activity of both recombinant and cellular p21ras.24 To determine whether oxLDL mediated S-glutathiolation of p21ras persists for up to 1 hour, p21ras was immuno-precipitated and then immunoblotted with a monoclonal antibody against GSH protein adducts. After 1 hour, oxLDL caused >2-fold increase in p21ras S-glutathiolation (Figure 4A and 4B). The immunoreactivity of p21ras with the anti-glutathione adduct antibody was prevented if the immunoprecipitate was first incubated with dithiothreitol (DTT, 10 mmol/L), indicating that the antibody reacts with the reducible proteinglutathione adduct (Figure 4A). In addition, adenoviral transfection with the p21ras C118S mutant or glutaredoxin-1 prevented p21ras S-glutathiolation (Figure 4B). This result is consistent with persistent S-glutathiolation of p21ras during the time course of these experiments.
|
MALDI-TOF mass spectrometry was also used to identify S-glutathiolated cysteines in a tryptic digest of p21ras immunoprecipitated from BAEC treated with oxLDL. After SDS PAGE, the Coomassie blue stained 21-kDa band (Figure 4C, left, lane 2), which corresponded to the band stained with the anti-GSH antibody (Figure 4C, left, lane 3), was digested and extracted from the gel. A peptide containing cysteine-118 in its S-glutathiolated form was identified by the increase of 305 Da in its mass accounted for by the GSH adduct (Figure 4C and supplemental Table I). This modification was not seen on other p21ras cysteines, although they were detected to be alkylated with iodoacetamide, consistent with them being in the reduced form when cells were lysed. Also, when the sample was treated with DTT (10 mmol/L), the peak for the S-glutathiolated cysteine disappeared, confirming the reducible nature of the modification (Figure 4C).
Oxidized GSH Mimics the Effect of oxLDL on Insulin Signaling
GSSG is increased in cells under conditions of oxidative stress and is implicated in causing S-glutathiolation of p21ras.24,31 BAEC were treated with cell-permeable GSSG ethyl ester (25 µmol/L or 250 µmol/L, 1 hour) before addition of insulin to evaluate the potential effect on insulin signaling. GSSG concentration-dependently decreased both basal and insulin-induced Akt phosphorylation (Figure 4D). GSSG also increased Erk phosphorylation (Figure 4D), invoking the same mechanism by which peroxynitrite or oxLDL inhibits insulin signaling. Under the same conditions, reduced GSH ethyl ester had no effect on either Akt or Erk phosphorylation (data not shown). In addition, BAEC were exposed to biotin-labeled GSH ethyl ester (250 µmol/L, 1 hour) as a control or biotinylated oxidized GSH ethyl ester (biotin GSSG, 250 µmol/L) for various times, and S-glutathiolated proteins were pulled-down with streptavidin-Sepharose beads and immuno-blotted with anti-p21ras antibody. In the control, S-glutathiolation of p21ras was detected 1 hour after biotinylated GSH ester, but was stimulated significantly to levels >2-fold greater when assayed between 5 and 60 minutes after addition of biotinylated GSSG (supplemental Figure VIII). These data indicate that GSSG can directly increase S-glutathiolation and activation of p21ras in BAEC as has been previously demonstrated for recombinant p21ras,24,31 and that the oxidized glutathione adduct can mediate insulin resistance via p21ras activation.
OxLDL Increases Phosphorylation of IRS-1 Serine-616 Via Oxidant-Mediated p21ras Activation
IRS-1 serine phosphorylation has been implicated in disrupting insulin signaling to its downstream targets, PI3 kinase and Akt. Because IRS-1 serine-616 is in an Erk phosphorylation consensus sequence, and because Erk was identified as a key component in the inhibition of Akt phosphorylation, IRS-1 phosphorylation was assessed. Optimal results were obtained by immunoprecipitation with antibody against phosphorylated IRS-1 serine-616, followed by immunoblotting for IRS-1 in the immunoprecipitate. BAEC treated for 1 hour with oxLDL (100 µg/mL) or peroxynitrite (100 µmol/L) showed increased phosphorylation of IRS-1 serine-616 (Figure 5A and 5B). PD98059 (20 µmol/L) prevented the increased phosphorylation of IRS-1 serine-616 caused by oxLDL or exogenous peroxynitrite (Figure 5A), indicating that Erk-dependent phosphorylation of serine-616 accompanies the decreased insulin-induced Akt phosphorylation caused by peroxynitrite or oxLDL. In addition, the peroxynitrite scavenger, MnTBAP, as well as the nitric oxide synthase inhibitor, L-NAME, prevented serine-616 phosphorylation after exposure to oxLDL (supplemental Figure IX), confirming a role of the nitric oxide- and superoxide anion-derived oxidant in the Erk-dependent induction of IRS-1 serine-616 phosphorylation. Finally, adenoviral transfection with the p21ras C118S mutant also prevented the increased phosphorylation of IRS-1 serine-616 caused by oxLDL or peroxynitrite in BAEC (Figure 5B), confirming the role of oxidant-activated p21ras. Furthermore, in testing this mechanism in another cell type, we found that oxLDL increased IRS-1 serine-616 phosphorylation in human HepG2 cells, and that this was prevented when these cells were transfected with p21ras C118S mutant or treated with the Mek inhibitor, U0126 (supplemental Figure X). This provides further evidence that oxidant activation of p21ras mediates IRS-1 serine phosphorylation, and that this mechanism may be common to other cells.
|
| Discussion |
|---|
|
|
|---|
|
Highly oxidized LDL used in these studies might itself be a source of oxidants and is a heterogeneous preparation. However, we found that LPC, a key active component of oxLDL, also increased peroxynitrite in BAEC and induced insulin resistance. Also, we provide evidence that the effects of oxLDL on Akt signaling are mediated, at least in part, through PG receptors, stimulation of which mimic its effect, as was recently demonstrated for minimally oxidized LDL.30 These results suggest that oxLDL is not nonspecific, but mediates oxidant production and insulin resistance through specific signaling pathways.
It should be noted that inhibition of insulin-induced Akt phosphorylation caused by oxLDL occurs slowly; indeed, it follows a transient increase and return to baseline in Akt phosphorylation that occurs over the first 30 minutes after oxLDL.24 The inhibition of insulin-induced Akt phosphorylation by oxLDL is nevertheless specific, as VEGF-induced Akt phosphorylation is unaffected. This result suggests a specific interaction with the insulin signaling pathway after exposure of cells to oxLDL. MAP kinase-dependent serine phosphorylation of IRS-1 is a well-recognized mechanism by which insulin signaling can be specifically inhibited. Noting the reciprocal relationship between Akt and Erk phosphorylation after exposure of BAEC to oxLDL or peroxynitrite suggested that Erk-dependent serine phosphorylation of IRS-1 might be implicated. This was demonstrated by showing that specific Mek inhibitors prevented both Erk phosphorylation and IRS-1 serine-616 phosphorylation, and also restored insulin-induced Akt phosphorylation. Several other MAP kinases can induce serine phosphorylation of IRS-1, and notably Jun kinase has been implicated in insulin resistance.8 Although p21ras can activate Jun kinase in BAEC, we found that a Jun kinase inhibitor provided no added benefit beyond blocking Mek to insulin-induced Akt phosphorylation (data not shown).
Because the S-glutathiolation and activation of p21ras occurs relatively slowly over a prolonged time course (10 to 15 minutes),24 it is unlikely that peroxynitrite is directly causing S-glutathiolation of p21ras. This is consistent with the fact that treating the cells with a bolus of peroxynitrite, which disappears rapidly, does not itself significantly accelerate signaling compared with that caused by oxLDL. Moreover, peroxynitrite produced after exposure to oxLDL and detected with dihydrorhodamine or dihydroethidium increases quickly within minutes and thus does not appear to be the limiting kinetic factor. In this study we also show that continuous generation of peroxynitrite at low concentrations produced by a nitric oxide donor and a superoxide anion generator has effects on insulin signaling similar to bolus addition of higher concentrations.
Because the GSSG/GSH ratio increases in endothelial cells treated with oxLDL,33 we previously speculated that peroxynitrite-induced formation of GSSG could mediate the relatively slow S-glutathiolation of p21ras and downstream signaling. The peroxynitrite formed by oxLDL stimulation likely decomposes in the presence of CO2 to carbonate anion radicals (CO3·) and nitrogen dioxide radical (NO2·).34 Because of the high intracellular concentration of free GSH, these very reactive radicals would promote glutathione radical formation (GS·) with respective rate constants of 5x106 and 2x107,34 and subsequently yield oxidized glutathione (GSSG) that is capable of S-glutathiolation of proteins. The relative slowness of the p21ras S-glutathiolation and the signaling activated by it makes the formation of GSSG more likely than that of glutathione sulfenic acid or glutathiyl radical.24 Another potential species that could be formed by peroxynitrite and glutathione is S-nitrosoglutathione that is also capable of exchange reactions to form protein-glutathione mixed disulfides with p21ras (data not shown and reference 35). Immunoprecipitation of p21ras from BAEC exposed to oxLDL and immunoblotting with an anti-glutathione protein adduct antibody demonstrated that p21ras was more S-glutathiolated in the presence of oxLDL, and this could be reversed by overexpressing glutaredoxin-1 or the p21ras C118S mutant. This mutant lacks the predominant site of S-glutathiolation.24,31 Therefore, GSSG may act as a signaling molecule, triggering p21ras activation and downstream signaling. Indeed, we showed previously that recombinant p21ras was activated by GSSG.24 In this study we showed that GSSG ethyl ester resulted in prolonged p21ras S-glutathiolation and impaired insulin-stimulated Akt phosphorylation. In addition, in this study we demonstrate by mass spectrometry S-glutathiolation of cysteine-118 in native p21ras immunopurified from BAEC stimulated with oxLDL. These studies therefore suggest that GSSG, the accumulation of which accompanies many states of oxidant stress that cause insulin resistance,32 can stimulate p21ras-dependent insulin resistance.
In summary, these studies provide a novel molecular mechanism by which oxidants generated by oxLDL or potentially other stimuli can mediate p21ras S-glutathiolation and activation, and Erk-dependent inhibition of insulin signaling. Although the studies here were performed in endothelial cells, there is no reason why one would not expect oxidants to activate p21ras and inhibit insulin signaling by this mechanism in other cell types including smooth muscle,25 hepatocytes, fat, or skeletal muscle. Indeed, we found that a similar mechanism applied to human HepG2 cells. As another example, angiotensin II causes hydrogen peroxide production by NADPH oxidase and activates p21ras in vascular smooth muscle, and this could also initiate insulin resistance. In addition to impaired insulin signaling, several other characteristics of endothelial cells in insulin-resistant states could be explained by p21ras activation including increased adhesion molecule expression, impaired nitric oxide-induced vasodilation, and altered anticoagulant status. Realizing that GSSG levels are generally elevated in oxidant stress, activation of p21ras may be a key initiating factor that translates oxidant stress into the abnormal phenotype of cells affected by it.
| Acknowledgments |
|---|
Source of Funding
This work was supported by National Institute of Health grants P01 HL081738 (R.A.C.), HL55620 (R.A.C.), HL 68758 (R.A.C.), AG27080 (R.A.C.), the National Institutes of Health Boston University Cardiovascular Proteomics Center N01-HV-28178, and a Fellowship from the Institut de Recherches Servier (N.C.).
Disclosures
None.
| Footnotes |
|---|
| References |
|---|
|
|
|---|
This article has been cited by other articles:
![]() |
C. N. White, E. J. Hamilton, A. Garcia, D. Wang, K. K. M. Chia, G. A. Figtree, and H. H. Rasmussen Opposing effects of coupled and uncoupled NOS activity on the Na+-K+ pump in cardiac myocytes Am J Physiol Cell Physiol, February 1, 2008; 294(2): C572 - C578. [Abstract] [Full Text] [PDF] |
||||
![]() |
D. Namgaladze, A. Kollas, and B. Brune Oxidized LDL attenuates apoptosis in monocytic cells by activating ERK signaling J. Lipid Res., January 1, 2008; 49(1): 58 - 65. [Abstract] [Full Text] [PDF] |
||||
![]() |
M. C. Sykes, A. L. Mowbray, and H. Jo Reversible Glutathiolation of Caspase-3 by Glutaredoxin as a Novel Redox Signaling Mechanism in Tumor Necrosis Factor-{alpha}-Induced Cell Death Circ. Res., February 2, 2007; 100(2): 152 - 154. [Full Text] [PDF] |
||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
ATVB Home | Subscriptions | Archives | Feedback | Authors | Help | AHA Journals Home | Search Copyright © 2006 American Heart Association, Inc. All rights reserved. Unauthorized use prohibited. |