FALDH Reverses the Deleterious Action of Oxidative Stress Induced by Lipid Peroxidation Product 4-Hydroxynonenal on Insulin Signaling in 3T3-L1 Adipocytes

Cell culture solutions, reagents for SDS-PAGE, and protein A-Sepharose were from Life Technologies (Carlsbad, CA). Recombinant human insulin was from Novo Nordisk (Copenhagen, Denmark); HNE was from Calbiochem (La Jolla, CA); l-α-phosphatidylinositol was from Sigma (St. Louis, MO); thin-layer chromatography silica plates were from Merck (Darmstadt, Germany); Crosstide was from Upstate Biotechnology (Lake Placid, NY); polyvinyldiene difluoride membranes were from Millipore (Bedford, MA); and ECL reagents 2-deoxy-d-[³H]-glucose, [γ-³²P]ATP, and d-[2-³H]-glucose were from Amersham Pharmacia Biotech (Uppsala, Sweden). Rosiglitazone was from AG Scientific (Paris).

Antibodies to protein kinase B (PKB)-α/β, phospho-PKB (serine 473 and threonine 308), phospho-serine 307, and phospho-MAPK (phospho-MAPK p42/p44) were from Cell Signaling Technologies (Beverly, MA); antibodies to IRS-1 and phospho-tyrosine (clone 4G10) were from Upstate Biotechnology (Lake Placid, NY). Antibody to IRS-1 and IRS-2, used for immunoprecipitation, was produced in our laboratory. Antibodies to insulin receptor β (IRβ), IRS-2, p42/p44 MAPK, and Myc were from Santa Cruz Biotechnology (Santa Cruz, CA); antibody to HNE-Michael adduct was from Calbiochem and R&D Systems (Minneapolis, MN); and secondary anti-mouse or anti-rabbit antibodies conjugated to horseradish peroxidase were from The Jackson Laboratories (Copenhagen).

3T3-L1 fibroblasts, from the American Type Culture Collection (Rockville, MD), were grown and differentiated to adipocytes as described previously (14). HNE cytotoxicity was determined measuring cell viability with sodium 3′-(1-(phenylaminocarbonyl)-3,4-tetrazolium)-bis (4-methoxy-6-nitro) benzene sulfonic acid hydrate (XTT) assay (Roche Applied Science, Mannheim, Germany) and lactate dehydrogenase (LDH) activity (28).

Adenoviruses expressing FALDH were generated as described previously (14,29). Empty and FALDH adenoviruses titer of stocks was >10⁸ plaque-forming units/ml.

3T3-L1 adipocytes were washed and solubilized as described previously (14). Lysates were kept on ice for 20 min, and insoluble material was removed by centrifugation at 14,000g for 20 min. Protein concentration was determined by colorimetric assay (Bio-Rad). For immunoprecipitation and immunoblots, antibodies were used as described previously (31). Immunoreactive proteins were detected using horseradish peroxidase–linked secondary antibodies and enhanced chemiluminescence according to the manufacturer's instructions (Amersham Biosciences). Signal intensities were measured with ImageQuant.

3T3-L1 adipocytes were incubated in serum-free Dulbecco's modified Eagle's medium (DMEM) supplemented with 0.2% (wt/vol) BSA for 12 h. Cells were incubated or not with HNE and with or without insulin for 10 min. Cells were solubilized and supernatants (300 μg) were immunoprecipitated with antibody to IRS-1 coupled to protein A–sepharose beads for 3 h at 4°C. Immune pellets were washed, and lipid kinase assays performed as described previously (31). Phosphorylated lipids were detected by autoradiography on Kodak X-Omat films.

3T3-L1 adipocytes, infected or not with adenoviruses, were incubated in serum-free DMEM containing 0.2% (wt/vol) BSA for 12 h and treated or not with HNE. For glucose uptake and lipogenesis, cells were incubated and treated as described previously (30). Radioactivity was counted, and samples were normalized to protein concentration.

Extracted total RNA was treated with DNase (Ambion, Austin, TX), and 1 μg was reverse transcribed using Reverse Transcription System kit (Promega, Charbonnières, France) with random primers and oligo(dT)₁₅. Quantitative PCR was performed as described using oligonucleotides to quantify FALDH mRNA and 36B4 mRNA as reference gene (14).

Results are presented as means ± SE with experiment numbers indicated in the figure legends. Statistical significance was assessed using Student's t test.

Because insulin resistance has been associated with increased oxidative stress, we examined the HNE effect on insulin signaling in 3T3-L1 adipocytes. Exposure to HNE results in a dose- and time-dependent decrease in IRS-1 protein (Fig. 1A). Quantification of total IRS-1 indicates that the maximal decrease is 80%. Western blotting reveals that HNE induces a decrease in IRS-2 protein (Fig. 1B) with a time and dose dependency similar to that associated with IRS-1. In contrast, HNE has no detectable effect on p42/p44 MAPK proteins. Next, we searched for a possible action of HNE on molecules of the MAPK pathway. We found that the content of the three Shc isoforms and of growth factor receptor bound 2 (Grb2) remained unchanged in cells exposed to HNE (Fig. 1C). Moreover, HNE has no effect on insulin-induced threonine and tyrosine phosphorylation of p42/p44 MAPK (Fig. 1D). Finally, we measured cell viability in HNE-treated cells using XTT reagent (Fig. 1E). Viability was only marginally decreased at the highest HNE concentrations used. Moreover, we indirectly quantified cell viability measuring LDH activity in the incubation medium, which reflects the presence of necrotic cells. We found no significant difference in cell viability after incubation with HNE. In summary, exogenous HNE at nontoxic concentrations does not appear to impinge on the Shc-MAPK signaling module in 3T3-L1 adipocytes but leads to an important reduction in IRS-1/-2 proteins. This decrease is not due to a cytotoxic HNE effect on cell viability.

To investigate whether HNE treatment impairs insulin-induced tyrosine phosphorylation of its receptor β subunit (IRβ) and its proximal substrates, IRS, we incubated 3T3-L1 adipocytes with increasing HNE concentrations. HNE addition to the medium before exposure to insulin inhibits IRβ and IRS tyrosine phosphorylation in a time- and dose-dependent manner compared with control cells (Fig. 2A and B). The reduced IRS tyrosine phosphorylation can likely be accounted for by the reduction in IRS-1/-2 proteins and by reduced IRβ kinase activity. Importantly, the IRβ protein level remained virtually unchanged during HNE treatment. We next examined IRS-1 phosphorylation on serine 307 in presence of HNE (Fig. 2C). HNE transiently enhanced serine 307 phosphorylation between 1 and 4 h of HNE treatment.

HNE is known to induce formation of HNE-Michael adducts by reaction of the HNE aldehyde moiety with cysteine, histidine, or lysine residues in proteins (16). HNE induced formation of several aldehyde-protein adducts in 3T3-L1 adipocytes, as shown by Western blot using antibody to HNE-Michael adducts (Fig. 3A). To determine whether IRS-1 and -2 could form adducts with HNE, we incubated 3T3-L1 adipocytes with HNE and analyzed immunoprecipitated IRS-1 and IRS-2 by Western blot with antibody to HNE-Michael adducts (Fig. 3B). As expected, IRS-1 and -2 protein are decreased. Concomitantly, HNE-Michael adducts accumulate in a time-dependent manner. Thus, HNE is able to directly interact with IRS-1 and -2, leading to generation of HNE-IRS adducts. Taken together, our results suggest that, under oxidative stress conditions, formation of HNE-modified IRS adducts is increased in 3T3-L1 adipocytes and probably leads to IRS degradation. Moreover, pretreatment with the lysosomal inhibitor, chloroquine, partially prevents HNE-induced IRS-1 degradation (Fig. 3C). By contrast, pretreatment with MG132 or lactacystin, two proteasomal inhibitors, had no effect on IRS-1 level. To sum up, HNE-IRS-1 appears to be directed preferentially to the lysosomal pathway for degradation.

We evaluated the ability of insulin to activate PI 3-kinase and PKB after HNE exposure. As expected, insulin induces a fivefold increase in PI 3-kinase activity associated with IRS-1 in control cells. In contrast, HNE reduces basal PI 3-kinase activity by 50% and insulin-stimulated PI 3-kinase activity by 72% (Fig. 4A). It is likely that the decrease in PI 3-kinase activity is linked to decreased IRS-1 tyrosine phosphorylation in HNE-treated cells. To assess whether this defect in PI 3-kinase activity is associated with blunted PKB activation, we evaluated insulin-induced PKB phosphorylation after HNE treatment. First, we observed a dose-dependent decrease in PKB protein in presence of HNE (Fig. 4B). Second, we found that HNE treatment leads to a time-dependent decrease in insulin-stimulated PKB phosphorylation on serine 473 and threonine 308 (Fig. 4C). The reduction in PKB phosphorylation is correlated with the decrease in protein, suggesting that reduced PKB activation can be accounted for by a decrease in both decreased PKB protein and insulin receptor signaling.

As signaling through PI 3-kinase was altered upon HNE treatment, we investigated the HNE effect on insulin-induced metabolic responses. As expected, insulin induces a threefold increase in glucose uptake in control cells, whereas within 4 h, both basal (twofold relative to control) and insulin-stimulated glucose transport (fourfold relative to control) are decreased in cells incubated with HNE (Fig. 5A). Similarly, we found, in cells incubated with HNE, a time-dependent decrease in basal and insulin-stimulated lipogenesis compared with nontreated cells (eightfold decrease within 4 h) (Fig. 5B).

We previously showed that FALDH plays an important role in the decrease in ROS production in 3T3-L1 adipocytes exposed to HNE (14). Therefore, we hypothesized that FALDH could prevent the deleterious HNE action on insulin signaling. As expected, in control cells infected with empty adenovirus, HNE induces a robust decrease (70%) in insulin-induced IRS-1 tyrosine phosphorylation (Fig. 6A, lanes 4 and 5) as compared to nontreated cells (Fig. 6A, lanes 2 and 3). In contrast, in cells infected with FALDH adenovirus, we observed only a slight decrease (25%) in IRS-1 tyrosine phosphorylation in presence of HNE (Fig. 6A, lanes 9 and 10) compared with nontreated cells (Fig. 6A, lanes 7 and 8, and B). Looking at IRS-1 levels, we found that ectopic FALDH expression partially restores IRS-1 levels in presence of HNE (Fig. 6A, lanes 9 and 10) in contrast to control cells (Fig. 6A, lanes 4 and 5), which explains the improvement of IRS-1 tyrosine phosphorylation. Moreover, cells infected with FALDH adenovirus show a decrease in formation of HNE adducts compared with control cells (Fig. 6C). Finally, we found a reduction in HNE–IRS-1 adducts in cells expressing ectopic FALDH compared with control cells (Fig. 6D). Interestingly, these results indicate that by reducing the generation of HNE–IRS-1 adducts, increased FALDH expression allows partial restoration of the decrease in insulin-induced IRS tyrosine phosphorylation caused by HNE. To summarize, FALDH protects 3T3-L1 adipocytes against HNE action and hence permits the maintenance of an unperturbed insulin-induced activation of IRS.

To strengthen the idea that increased FALDH expression prevents HNE damage on signaling events induced by insulin, we investigated the consequence of ectopic FALDH expression on glucose uptake and lipogenesis in 3T3-L1 adipocytes incubated with HNE. As expected, HNE treatment of control cells induces a robust decrease (60%) in insulin-induced glucose transport (Fig. 7A, compare lanes 1 and 2). In contrast, in cells ectopically expressing FALDH, HNE reduces insulin-stimulated glucose transport only by 30% (Fig. 7A, compare lanes 3 and 4). Moreover, infection with empty or FALDH adenovirus had no effect on basal glucose transport (data not shown). Concerning lipogenesis, the impairment in triglycerides synthesis observed when control cells were treated with HNE (Fig. 7B, compare lanes 1 and 2) is more important than in cells ectopically expressing FALDH (Fig. 7B, compare lanes 3 and 4). In control cells exposed to HNE, lipogenesis is decreased by 78%, whereas in FALDH-infected cells, the HNE-induced decrease amounts to 44%. Moreover, infection with empty or FALDH adenovirus had no effect on basal lipogenesis (data not shown). Note that insulin stimulation of glucose uptake or of lipogenesis is slightly lower in infected cells compared with noninfected cells (Fig. 5), probably due to some noxious effect of adenovirus infection.

To sum up, expression of FALDH can partially prevent the deleterious HNE action on metabolic responses induced by insulin, demonstrating the importance of detoxifying enzymes to preserve insulin signaling under oxidative stress.

Because we previously showed that FALDH gene expression is increased by insulin, we hypothetized that rosiglitazone may protect against oxidative damage by increasing FALDH gene expression. Activation of peroxisome proliferator–activated receptors (PPARs), which are implicated in regulation of gene expression in response to insulin or to the antidiabetic drugs glitazones, has been reported to reduce cellular damage due to advanced glycated end product or oxidative stress (30,32,33). To determine whether rosiglitazone induces FALDH gene expression, we measured the amount of FALDH mRNA after rosiglitazone treatment (Fig. 8A). Thus, rosiglitazone induces a threefold increase in FALDH gene expression compared with nontreated cells. Next, we looked at the effect of preincubation of 3T3-L1 adipocytes with rosiglitazone on HNE-induced IRS-1 downregulation (Fig. 8B). As expected, IRS-1 protein is decreased ∼50% in control cells, whereas cells incubated with rosiglitazone showed only a 10% decrease, indicating that rosiglitazone protects cells against HNE-induced oxidative stress.

Insulin is a master regulator of several key functions in metabolism control. Defects in these control points due to inhibition of insulin signaling contribute decisively to the development of insulin resistance and type 2 diabetes. Oxidative stress has emerged as a causative factor for insulin resistance and as participating in the disease process leading to diabetes and its complications (8,9,34). During oxidative stress, several reactive products are generated, which include ROS and aldehydes such as HNE, through a lipid peroxidation process (11,17). Elevated ROS or HNE levels are correlated with hyperglycemia and type 2 diabetes, whereas antioxidants are known to reverse, at least in part, insulin resistance in rodents (35,36). HNE is the major peroxidation product of polyunsaturated fatty acids and is the most reactive one. HNE can be produced in relatively large amounts reaching up to millimolar concentrations both in vitro and in vivo in response to oxidative insults (16,37,38). Moreover, increased lipid peroxidation and HNE concentrations are found in diabetic subjects and the HNE concentrations used in our study are within the range of those reported in pathophysiological conditions, including diabetes (24). HNE accumulation has been implicated in several deleterious processes, such as inhibition of enzymes by chemical modification at low nontoxic concentrations, and by doing so, hampers their functions (16,17,39). In addition, HNE affects insulin secretion and signaling. HNE restrains glucose-induced insulin secretion (24) and in β-cells of Goto-Kakizaki rats, a model of nonobese type 2 diabetes, hyperglycemia might be responsible for the increase in HNE-modified proteins (25). Both increased lipid peroxidation and altered removal by detoxifying enzymes occur in diabetic rats and are possible causes of HNE accumulation (22,23). To sum up, growing evidence implicates HNE in functional alterations in diverse disease processes, including alterations in glucose homeostasis. Because insulin resistance inevitably precedes development of type 2 diabetes, we hypothesized that an increased level of toxic aldehydes, such as HNE, could be involved in dysfunctioning of signaling molecules and hence could hamper insulin action. We chose 3T3-L1 adipocytes as model systems for insulin action to analyze the effect of oxidative stress induced by HNE on insulin signaling. Augmented adipose mass due to nutrient excess is associated with increased lipid-derived metabolites that impair insulin signaling and generate insulin resistance (40,41).

We show a time- and dose-dependent decrease in IRS-1/-2 proteins in 3T3-L1 adipocytes exposed to nontoxic HNE concentrations, which is associated with a reduction in IRS tyrosine phosphorylation and an increase in IRS-1 phosphorylation on serine 307. In addition, we found increased JNK activation by HNE, which could explain this IRS-1 serine phosphorylation (data not shown). In addition, HNE reduces IRβ activation by insulin without significantly decreasing IRβ protein. We favor the idea that HNE acts by generating structural modifications in IRβ, leading to its impaired function. Our findings agree with a report showing a decrease in platelet-derived growth factor receptor β (PDGFRβ) activation in the presence of HNE without reducing total PDGFRβ number (42). Moreover, in our experiments, HNE seems to have no effect on MAPK signaling. Shc, Grb2, and MAPK protein levels were not affected by HNE. The small decrease in IRβ does not perturb MAPK activation in our experimental conditions, which is very likely due to the intact Shc/Grb2 module.

HNE interacts spontaneously with amino acids of proteins, leading to generation of Michael adducts (11,38). Therefore, we investigated whether the decrease in IRS proteins is associated with formation of HNE-modified IRS-1 and -2 adducts. Remarkably, we found in cells exposed to HNE a time-dependent accumulation of IRS-1 and -2 tagged with HNE. Importantly, no increase in HNE-modified MAPK protein adduct was detected, which could explain the absence of decrease in MAPK protein in presence of HNE (data not shown). Although it is known that proteins tagged with HNE are more sensitive to degradation than native proteins, the mechanisms underlying degradation of HNE adducts are not well defined (43–45). We show that IRS-1 protein level in presence of HNE is partially restored by pretreatment of cells with the lysosomal inhibitor chloroquine. However, inhibition of lysosomes is not sufficient to totally prevent HNE-induced decrease in IRS-1, suggesting the implication of other pathways.

Because IRS proteins play a crucial role in insulin signaling and because HNE leads to their downregulation, we looked at the HNE effect on PI 3-kinase and PKB. As expected, we found that HNE treatment of 3T3-L1 adipocytes results in decreased insulin-stimulated PI 3-kinase activity. Moreover, when cells are exposed to HNE, PKB phosphorylation on the two sites necessary for optimal activity is reduced, reflecting decreased PKB activation. Next, we were interested in possible consequences of oxidative stress on key insulin metabolic responses. We show that nontoxic HNE concentrations decrease glucose transport and lipogenesis. The decrease in glucose transport agrees with other studies reporting altered glucose transport in oxidative stress conditions induced by H₂O₂ (10,46).

Our demonstration that HNE impairs insulin signaling would indicate that this aldehyde plays an important role in insulin resistance development and therefore could foster progression to type 2 diabetes. We hypothesized that an increased antioxidant potential of cells and tissues is likely to improve insulin resistance. Antioxidant treatment, based on α-lipoic acid addition or antioxidant enzyme overexpression, improves insulin sensitivity (34,35). Concerning antioxidation machineries, we previously reported that FALDH decreases ROS production induced by HNE treatment (14). Moreover, we found that FALDH mRNA expression is increased by insulin and decreased in insulin-resistant animal models. Therefore, for testing whether an enzyme involved in detoxification can improve insulin signaling in cells exposed to oxidative stress, FALDH appeared to us as an appropriate candidate. Remarkably, 3T3-L1 adipocytes infected with an adenoviral construct encoding FALDH present stronger IRS-1 tyrosine phosphorylation in response to insulin than control cells, and this is associated with partial restoration of IRS-1 protein. Consistent with this, FALDH partially prevents formation of HNE adducts with several proteins, including IRS-1, which confirms that HNE binding to IRS favors their degradation. Finally, increased FALDH expression improves insulin metabolic responses, such as glucose transport and lipogenesis, in HNE-treated cells. To further document the potential role of FALDH protection against oxidative stress of diabetes, we looked at the effect of the antidiabetic drug rosiglitazone. PPAR agonists are known to modulate expression of antioxidant genes, such as NADP(H) oxidase or Cu²⁺/Zn²⁺ superoxide dismutase (30,47). Interestingly, we show that FALDH gene expression is increased by rosiglitazone. At the concentration used, rosiglitazone is expected to activate both PPARα and -γ. Therefore, our result agrees with studies showing an increase in FALDH gene expression in response to clofibrate, an activator of PPARα, or a decrease in PPARα-null mice (48,49). Moreover, we found that IRS-1 protein is virtually unchanged by HNE in 3T3-L1 adipocytes pretreated by rosiglitazone compared with nontreated cells. The antioxidant effect of rosiglitazone is likely to be, at least in part, due to increased endogenous FALDH, which detoxifies HNE. However, because rosiglitazone modulates several cellular functions, we cannot exclude participation of molecules other than FALDH. Our finding showing an antioxidative action of rosiglitazone through increased FALDH gene expression fits with studies showing that rosiglitazone and troglitazone improve atherosclerosis by decreasing ROS generation and lipid peroxidation (32,50). The control exerted by the antidiabetic drug rosiglitazone on expression of genes, such as FALDH, involved in protection against oxidative stress, might reflect the existence of cellular defense mechanisms against damage of oxidative stress on insulin action.

To conclude, the key findings of our study are 1) 3T3-L1 adipocytes treatment with HNE impairs insulin action, and 2) increased FALDH expression protects 3T3-L1 adipocytes against HNE-induced oxidative stress. Because we previously showed that in insulin-resistant animals the stimulatory action of insulin on FALDH gene expression is lost and because HNE induces insulin resistance, we suggest the occurrence of a vicious circle between HNE accumulation and insulin resistance. Considering our findings on the ability of FALDH to reverse the harmful impact of HNE on insulin action, it is likely that the means leading to increased detoxification of aldehydes represent potential novel candidates to combat insulin resistance.

D.D. has received support from Association pour la Recherche contre le Cancer. Our research was supported by INSERM, by the Programme National de Recherche sur le Diabète of INSERM (2004), by Conseil Général des Alpes-Maritimes et Conseil Régional PACA, by the University of Nice Sophia-Antipolis, and by a grant from the European Community (FP6 Eugene2 [LSHM-CT-2004-512013]).

We thank Sophie Grillo for assistance in the initial part of the study and Jean Giudicelli for LDH activity assay.