Lipid Peroxidation Product 4-Hydroxy-trans-2-nonenal Causes Endothelial Activation by Inducing Endoplasmic Reticulum Stress*

Background: Oxidized lipids cause endothelial activation. Results: Endothelial activation by the lipid peroxidation product, 4-hydroxy-trans-2-nonenal, was associated with ER stress and was prevented by chaperones of protein folding. Conclusion: ER stress regulates endothelial activation by oxidized lipids. Significance: Vascular inflammation caused by oxidized lipids could be attenuated by decreasing ER stress. Lipid peroxidation products, such as 4-hydroxy-trans-2-nonenal (HNE), cause endothelial activation, and they increase the adhesion of the endothelium to circulating leukocytes. Nevertheless, the mechanisms underlying these effects remain unclear. We observed that in HNE-treated human umbilical vein endothelial cells, some of the protein-HNE adducts colocalize with the endoplasmic reticulum (ER) and that HNE forms covalent adducts with several ER chaperones that assist in protein folding. We also found that at concentrations that did not induce apoptosis or necrosis, HNE activated the unfolded protein response, leading to an increase in XBP-1 splicing, phosphorylation of protein kinase-like ER kinase and eukaryotic translation initiation factor 2α, and the induction of ATF3 and ATF4. This increase in eukaryotic translation initiation factor 2α phosphorylation was prevented by transfection with protein kinase-like ER kinase siRNA. Treatment with HNE increased the expression of the ER chaperones, GRP78 and HERP. Exposure to HNE led to a depletion of reduced glutathione and an increase in the production of reactive oxygen species (ROS); however, glutathione depletion and ROS production by tert-butyl-hydroperoxide did not trigger the unfolded protein response. Pretreatment with a chemical chaperone, phenylbutyric acid, or adenoviral transfection with ATF6 attenuated HNE-induced monocyte adhesion and IL-8 induction. Moreover, phenylbutyric acid and taurine-conjugated ursodeoxycholic acid attenuated HNE-induced leukocyte rolling and their firm adhesion to the endothelium in rat cremaster muscle. These data suggest that endothelial activation by HNE is mediated in part by ER stress, induced by mechanisms independent of ROS production or glutathione depletion. The induction of ER stress may be a significant cause of vascular inflammation induced by products of oxidized lipids.

Lipid peroxidation products, such as 4-hydroxy-trans-2nonenal (HNE), cause endothelial activation, and they increase the adhesion of the endothelium to circulating leukocytes. Nevertheless, the mechanisms underlying these effects remain unclear. We observed that in HNE-treated human umbilical vein endothelial cells, some of the protein-HNE adducts colocalize with the endoplasmic reticulum (ER) and that HNE forms covalent adducts with several ER chaperones that assist in protein folding. We also found that at concentrations that did not induce apoptosis or necrosis, HNE activated the unfolded protein response, leading to an increase in XBP-1 splicing, phosphorylation of protein kinase-like ER kinase and eukaryotic translation initiation factor 2␣, and the induction of ATF3 and ATF4. This increase in eukaryotic translation initiation factor 2␣ phosphorylation was prevented by transfection with protein kinase-like ER kinase siRNA. Treatment with HNE increased the expression of the ER chaperones, GRP78 and HERP. Exposure to HNE led to a depletion of reduced glutathione and an increase in the production of reactive oxygen species (ROS); however, glutathione depletion and ROS production by tertbutyl-hydroperoxide did not trigger the unfolded protein response. Pretreatment with a chemical chaperone, phenylbutyric acid, or adenoviral transfection with ATF6 attenuated HNE-induced monocyte adhesion and IL-8 induction. Moreover, phenylbutyric acid and taurine-conjugated ursodeoxycholic acid attenuated HNE-induced leukocyte rolling and their firm adhesion to the endothelium in rat cremaster muscle. These data suggest that endothelial activation by HNE is mediated in part by ER stress, induced by mechanisms independent of ROS production or glutathione depletion. The

induction of ER stress may be a significant cause of vascular inflammation induced by products of oxidized lipids.
Extensive experimental and clinical evidence suggests that atherosclerosis begins with the subendothelial accumulation of lipoproteins in lesion-prone sites (1). Within the acellular space of the subintima, which is devoid of antioxidants, unsaturated lipids in the lipoproteins undergo extensive oxidation and aggregation. It is currently believed that the reactive products generated within oxidized lipoproteins cause endothelial activation, leading to an increase in endothelial adhesion and cytokine production. Activation of the endothelium is one of the earliest events in atherogenesis that results in the recruitment of monocytes to sites of lesion initiation associated with the accumulation of oxidized lipoproteins. However, the mechanisms by which lipoprotein oxidation products cause endothelial activation remain unclear.
Oxidation of lipids by reactive oxygen species (ROS) 3 generates several reactive molecules of which aldehydes are major end products (2). These aldehydes are more stable than ROS, and therefore they can diffuse from their site of formation to propagate and amplify oxidative injury. Among the products generated in oxidized lipids, the C9 unsaturated aldehyde 4-hydroxy-trans-2-nonenal (HNE), is the most abundant unsaturated aldehyde. It is generated by the oxidation of -6 polyunsaturated (arachidonate, ␥-linolenate, linoleate) fatty acids and under some conditions constitutes a majority of the unsaturated aldehydes present in oxidized lipids (3). HNE is a highly reactive aldehyde that avidly attacks nucleophilic centers in proteins, phospholipids, and nucleotides (2). Several studies have shown that HNE-modified proteins and DNA accumulate in diseased and injured tissues (4). Significantly, high levels of protein-HNE adducts are also generated in oxidized LDL (5), and HNE-modified proteins have been detected in atherosclerotic lesions (6) and inflammation (7). Exposure to HNE has been shown to increase monocyte adhesion (8) and to stimulate cytokine production (9) in endothelial cells. Nonetheless, the mechanisms by which HNE causes endothelial activation remain unclear. Therefore, the current study was designed to identify the molecular and cellular mechanisms that contribute to HNE-induced endothelial activation. Our results show that exposure to HNE leads to the modification of ER-resident proteins and that HNE-induced endothelial activation is mediated in part by the unfolded protein response (UPR) to ER stress.

EXPERIMENTAL PROCEDURES
Chemical Reagents and Antibodies-[ 3 H]HNE and HNE were synthesized as described (10). The polyclonal antibody against keyhole limpet hemocyanin-protein-HNE was raised as described before (15). Phycoerythrin-conjugated rat antimouse CD31 was obtained from BD Pharmingen (San Diego, CA) and Serotec (Raleigh, NC), respectively. Monoclonal anti-KDEL and ATF6 (activation transcription factor 6) antibodies were obtained from Stressgen (Ann Arbor, MI) and IMGENEX (San Diego, CA), respectively. Polyclonal anti-actin antibody was purchased from Sigma-Aldrich. Polyclonal antibodies against ATF3 and phospho-PERK (Thr 981 ) were obtained from Santa Cruz Biotechnology, Inc. (Santa Cruz). Anti-eIF2␣ and phospho-eIF2␣ (Ser 51 ), GRP94 and protein-disulfide isomerase (PDI) antibodies, and HRP-linked secondary goat anti-rabbit or goat anti-mouse antibodies were purchased from Cell Signaling Technology (Danvers, MA). Anti-ICAM-1 monoclonal antibody was purchased from BD Biosciences. Alexa Fluor 488labeled anti-rabbit IgG and Texas Red-labeled anti-mouse IgG antibodies and the Slow Fade Gold antifade reagent were purchased from Molecular Probes/Invitrogen. Monochlorobimane was purchased from Molecular Probes/Invitrogen. Phenyl butyric acid (PBA) and taurine-conjugated ursodeoxycholic acid (TUDCA) were purchased from Pfaltz and Bauer Inc. (Waterbury, CT) and Calbiochem, respectively. Electrophoresis and Western blot supplies were purchased from Bio-Rad, and primers for PCR were obtained from Integrated DNA Technologies (Coralville, IA).
Cell Culture and Treatment-Human umbilical vein endothelial cells (HUVEC) were obtained from Lonza (Walkersville, MD) and cultured as described before (11). Subconfluent (ϳ80%) cells were incubated with the indicated concentrations of HNE for up to 2 h in Hanks' balanced salt solution (HBSS). Where indicated, HNE containing HBSS was removed after 2 h and replaced with HNE-free cell culture medium, and cells were maintained in this medium for the indicated time. To examine the effect of the chemical chaperones, the cells were preincubated with 10 mM PBA (Pfaltz and Bauer) as described before (12) followed by incubation with HNE in HBSS containing PBA. Treatment with [ 3 H]HNE-HUVEC cultured in 6-well dishes were incubated with 5-25 nmol of [ 3 H]HNE containing (2 ϫ 10 6 cpm/well) for 30 min in HBSS. Incubation medium was aspirated, and any adherent radioactivity to the cells was carefully removed by repetitive washing. Once no radioactivity was observed in the wash, cells were lysed, and proteins were precipitated by PCA, and the pellet was rinsed with acetone to remove any residual adherent radioactivity. The cell pellet was resolubilized in 0.1 M Tris, pH 8.0, containing 1% SDS. Protein concentration was measured by Lowry's method, and proteinbound radioactivity was counted in an aliquot of the cell pellet.
Immunocytochemistry-Cells cultured on coverslips were incubated with HNE for 30 min, rinsed with ice-cold phosphate-buffered saline (PBS), and fixed in formalin for 10 min. Cells were then permeabilized in PBS containing 0.1% Triton X-100 for 3 min. Subsequently, samples were incubated in PBS containing 1% BSA for 15 min at room temperature and then with the primary antibody for 16 h at 4°C. Samples were incubated with the fluorescent secondary antibodies for 1 h at room temperature in dark. Nuclei were stained with DAPI, and cells were mounted with Slow Fade Gold antifade reagent. Images were acquired on a fluorescent microscope.
Liquid Chromatography-Mass Spectrometric Analysis of HNE-modified Proteins-HUVEC cultured in T75 flasks to ϳ80% confluence were incubated without or with HNE for 30 min in HBSS at 37°C and frozen at Ϫ80°C until used for proteomic analyses. Cells were thawed on ice, lysed in a hypotonic buffer (10 mM HEPES, pH 7.4, 1.5 mM MgCl 2 , 0.1% Nonidet P-40, 1ϫ Roche Phosphostop phosphatase inhibitor mixture, and 1ϫ Roche Complete mini protease inhibitor mixture), transferred to a 1.5-ml microcentrifuge tube, and gently triturated using a pipette tip several times. The lysates were stored on ice for 10 min and then spun at 600 ϫ g for 3 min to pellet nuclei. The cytosolic fraction was transferred to a fresh tube and stored on ice prior to protein assay.
The cytoplasmic cell fraction (200 l) was loaded onto a size exclusion chromatography (SEC) column (BioSep-SEC-S4000, Phenomenex (Torrance, CA)) and eluted isocratically in 0.1 M phosphate buffer at a flow rate of 1 ml/min using an Ultimate TM 3000 (Dionex, Sunnyvale, CA) HPLC. The elution was monitored with an Ultimate 3000 dual wavelength UV-visible detector operating at 280 nm. Fractions (1 ml) were collected and digested overnight at 37°C with shaking, using mass spectrometry grade trypsin (without reduction or alkylation) following buffer adjustment to pH 8.0. The digested protein sample was trap-cleaned (MicroTrap, Michrom Bioresources, Auburn, CA), concentrated, and, following lyophilization, redissolved in 2% acetonitrile, 0.05% formic acid.
The digested samples (1-5 g) of five SEC fractions were analyzed using one-dimensional reversed phase HPLC-MS/MS as described (13). The sample fractions containing stress response proteins, including HSP90, HSP70, and protein-disulfide isomerases were further characterized using two-dimensional liquid chromatography (LC) of strong cation exchange followed with reverse phase prior to MS/MS analysis using a Thermo Scientific LTQ ion trap mass spectrometer interfaced to the Ultimate TM 3000 (Dionex) using a nanospray source operated in a data-dependent manner. Tandem mass spectra were extracted without charge state deconvolution or deisotoping. All MS/MS samples were analyzed using Sequest (Thermo Fisher Scientific (San Jose, CA), version 27, rev. 11). Sequest was set up to search the human_refseq_20110926 data base (unknown version, 32,972 entries) assuming the digestion enzyme trypsin (maximum missed cleavages of 2). Sequest was searched with a fragment ion mass tolerance of 1.00 Da and a parent ion tolerance of 1.2 Da. Modification of cysteine, histidine, and lysine by 4-hydroxynonenal (ϩ156, ϩ78, ϩ52 for ϩ1, ϩ2, or ϩ3 charge states) or dehydrated HNE (ϩ138, ϩ69, ϩ46 for ϩ1, ϩ2, or ϩ3 charge states) on cysteine, histidine, and lysine were specified in Sequest as variable modifications. Scaffold (version Scaffold_3.1.4.1, Proteome Software Inc. (Portland, OR)) was used to validate MS/MS-based peptide and protein identifications. Peptide identifications were accepted if they could be established at Ͼ95.0% probability as specified by the Peptide Prophet algorithm. Protein identification was accepted if it could be established at Ͼ95.0% probability and contained one or more identified peptides below the 1% false discovery rate. Protein probabilities were assigned by the Protein Prophet algorithm. Proteins that contained similar peptides and could not be differentiated based on MS/MS analysis alone were grouped to satisfy the principles of parsimony.
Apoptosis-Apoptosis in HNE-treated cells was examined by fluorescence-assisted cell sorting (FACS) using the Annexin V apoptosis detection kit FITC from eBioscience (San Diego, CA) as per the manufacturer's instructions, on an LSRII flow cytometer. Data were analyzed using the FACSDiva software (BD Biosciences).
Western Blot Analysis-Western blotting was performed in total cell lysates as described (14). To detect protein-HNE adducts, DTT was omitted from the lysis and sample buffers.
Genetic Ablation Using Small Interfering RNA (siRNA)-IRE-1 (inositol-requiring ER-to-nucleus protein-1), eIF2␣, and PERK mRNA in HUVEC cells were knocked down using an RNAi method following an siRNA transfection protocol provided by Invitrogen. The siRNAs were purchased from Qiagen (Valencia, CA), each as a pool of four target-specific 20 -25nucleotide siRNAs. Scrambled siRNA, purchased from Qiagen, contained non-targeting 20 -25-nucleotide RNA and was applied to control cells as a negative control. Briefly, after culturing the cells in antibiotic-free growth medium at 37°C in a humidified atmosphere of 5% CO 2 for 24 h, siRNA diluted in Opti-MEM (Invitrogen) and preincubated with the transfection agent oligofectamine (Invitrogen) was added. After transfection with scrambled or target-specific siRNA for 24 h (for RNA) or 48 h (for protein), the medium was replaced with HBSS and treated with HNE as indicated. RNA was isolated using the RNeasy minikit (Qiagen), and quantitative RT-PCR was performed to measure specific mRNA expression.
Surface Expression of ICAM-1-HUVEC were incubated without or with HNE in HBSS for 2 h at 37°C followed by incubation in endothelial cell culture medium for 4 h at 37°C. Cells were harvested in 5 mM EDTA and incubated with anti-ICAM-1 antibody for 1 h on ice. After incubation, the cells were pelleted, resuspended in PBS containing an Alexa488 conjugated anti-mouse IgG, and incubated for an additional 30 min. After washing, the fluorescence properties of these cells were analyzed on an LSR II flow cytometer (BD Biosciences). Cells stained with an isotype control antibody were used to detect background staining, and the mean fluorescence intensity of these samples was subtracted from the mean fluorescence intensity of anti-ICAM-1-stained cells.
Adhesion Assay-HUVEC (5 ϫ 10 4 cells) cultured for 24 h in 96-well plates were incubated with HNE in HBSS for 2 h at 37°C. Cells were then incubated in endothelial cell culture medium for 18 h. THP-1 cells were labeled with calcein acetoxymethyl ester (Invitrogen) (10 M; 30 min at 37°C) and allowed to adhere to the endothelial monolayer (50,000 THP-1 cells/well; 1 h at 37°C in HBSS with 0.1% glucose). The unadhered THP-1 cells were washed four times with PBS, and the adhesion was measured using a florescent plate reader (16).
Adenoviral Transfection-The adenovirus expressing ATF6 was kindly provided by Dr. Ron Prywes (Columbia University). HUVEC were transfected with the indicated amount of adenovirus in 0.5 ml of serum-free cell culture medium for 30 min in 12-well plates (18). Subsequently, culture medium was aspirated, and fresh cell culture medium with serum was added to the cells. Cells were incubated for 24 h. Transfected cells were then incubated without or with HNE for the indicated time in HBSS, and mRNA expression and protein abundance were measured by quantitative real-time PCR and Western blotting, respectively.
Intravital Microscopy-Sprague-Dawley rats (200 -250 g) were administered PBS (vehicle; 250 l, intravenously in tail vein) or HNE (1 mg/kg in 250 l of PBS, intravenously in tail vein). After 30 min, the concentration of HNE in the plasma was measured by gas chromatography-mass spectrometry as described (6). To examine the effect of HNE on endothelial function, the rats were anesthetized with sodium pentobarbital (50 mg/kg intraperitoneally) 18 h after treatment with HNE. The trachea and the carotid artery were cannulated to maintain airway patency and to directly monitor blood pressure, respectively. The cremaster muscle was prepared for in vivo microcirculatory observation as described previously (19) and positioned over an optic port in a specially designed plexiglass bath. The bath was filled with 60 ml of modified Kreb's solution that was maintained at a temperature of 35 Ϯ 0.5°C and a pH of 7.4 Ϯ 0.5. The rat or the tissue bath was placed on the modified stage of a Nikon MM-11 microscope so that the microcirculation could be observed by transillumination of the cremaster muscle. Closed circuit television microscopy was used to observe and quantify the diameters of single, unbranched third order venules that had basal diameters of 25-35 m. The video system was calibrated with a stage micrometer, and the vessel diameters were measured with a video caliper. The adhesion interactions between leukocytes and the vascular endothelium were measured by quantifying the number of transiently (rolling) and firmly (sticking) adherent leukocytes. Leukocyte rolling was determined by counting the number of leukocytes passing an arbitrary reference point, and leukocyte adhesion was quantified by counting, for each vessel, the number of adherent leukocytes in a 100-m length. To study the effect of chemical chaperones on leukocyte rolling and adhesion to cremaster endothelium in vivo, rats were administered PBA (0.25 g/kg in PBS, by gavage) or TUDCA (0.25g/kg in PBS intraperitoneally) three times at 12-h intervals prior to HNE treatment.
Statistical Analysis-Data are mean Ϯ S.E. Student's t test was used for comparison between two groups. One-way analysis of variance was used for comparison between multiple groups. Significance was set at p Ͻ 0.05.

Protein-HNE Adducts
Colocalize with ER-To study the effects of HNE on endothelial activation, we first determined the concentrations at which HNE does not induce cell death. HNE is highly toxic and has been shown to trigger apoptosis and necrosis in a variety of cell types (20,21). Hence, to identify a concentration range for studying the non-lethal effects of HNE, HUVEC were preincubated with 0 -100 M HNE for 2 or 6 h and then labeled with Annexin V and 7-aminoactinomycin D to measure apoptosis and necrosis by flow cytometry. Treatment with 5-25 M HNE for 2 h induced cell death in ϳ2% of the cells. Even with 50 -100 M HNE, Ͼ90% cells were alive after 2 h. Apoptosis in 10 -15% cells was observed after treatment with 50 -100 M HNE; however, ϳ2% apoptosis was observed in cells treated with 25 M HNE for 18 h (data not shown). Therefore, for all subsequent experiments, Ͻ25 M HNE was used. Although high concentrations of HNE are generated in oxidized LDL (2), HNE is rapidly metabolized, and therefore steady-state concentration of HNE in normal plasma and tissue is between 0.1 and 6 M (2, 22). However, between 2 and 40 M HNE has been detected in diseased tissue under oxidative stress (2). Hence, HNE concentrations of Ͻ25 M are within the range expected under oxidative stress in vivo, and the effects observed at these concentrations are unlikely to be overwhelmed by the activation of cell death pathways.
Because HNE and related aldehydes form covalent adducts with proteins, and several of the effects of HNE could be related to protein modification (2), we studied the modification of endothelial proteins by HNE. Incubation of HUVEC with [ 3 H]HNE (5 nmol in 1.0 ml of HBSS; 30 min at 37°C) resulted in the binding of 43 Ϯ 2 pmol of HNE with cell proteins (Fig. 1A, i). Incubation with a higher concentration of HNE (10 -25 M) increased protein-HNE adduct formation in a concentrationdependent manner (Fig. 1A, i). Western blot analysis showed that incubation with HNE (25 M; 30 min at 37°C in HBSS) resulted in the appearance of several proteins that displayed positive reactivity with anti-protein HNE antibodies (Fig. 1A, ii), indicating that HNE causes extensive modification of endothelial proteins.
To examine the cellular localization of HNE-modified proteins, the cells were incubated with HNE, and protein-HNE adducts were visualized by immunocytochemical analysis and confocal microscopy. Only sparse staining was observed in untreated cells. Treatment with HNE resulted in a large increase in antibody staining of the cell. To examine whether treatment with HNE also modifies ER proteins, the cells were stained with the anti-KDEL antibody, which binds to ER-resident proteins. The difference in staining from one cell to the next reflected the extent of ER in different cells.
After comparing several such cells, we did not see a significant difference in the ER staining between untreated and HNE-treated cells. Confocal overlay images showed significant co-localization of the protein-HNE adducts with the ER (Fig. 1B). These data suggest that HNE forms covalent adducts with several endothelial proteins and that some of these adducts co-localize with the ER.
To identify which proteins were modified by HNE, we analyzed proteins extracted from HNE-treated HUVEC by onedimensional LC-MS/MS. Analysis of SEC-HPLC fractions collected from the separation of HNE-treated cell cytoplasm using one-dimensional LC-MS/MS proteomic methods identified 101 proteins by one peptide or more at the 0.3% protein and peptide false discovery rate. Molecular chaperones, including HSP90, HSP70, and PDI along with structural proteins (actin and vimentin) comprised the five most abundant proteins (by a label-free spectral counting approach). These chaperones were localized to two SEC-HPLC fractions (retention times of 4 and 5 min). Further analysis of these two fractions using a two-dimensional LC-MS/MS approach, used to achieve a greater sensitivity to low abundance post-translational modifications, identified a total of 243 proteins by one peptide or more at the 0.2% peptide and 0.3% protein false discovery rate. HNE and dehydrated HNE modifications of lysine, histidine, and cysteine side chains were detected on 58 peptides derived from 27 proteins, including six ER-associated proteins and 21 non-ER-associated proteins ( Table 1). As shown in Fig. 1C, HNE formed a Michael adduct with the lysine residue of the peptide NQLTSNPENTVFDAK* (ϩ156) of ER chaperone GRP78. Similar Michael adducts were also formed with ER-associ-ated proteins procollagen lysine, 2-oxoglutarate 5-deoxygenase, and ribosome-binding protein-1 (Table 1). For ribosome-binding protein-1, one peptide was identified to have multiple HNE modifications. Additionally, dehydrated HNE adducts (consistent with Schiff base formation) were observed with lysine and histidine residues of several ER-associated proteins, including GP96 (endoplasmin; Table 1). Modification by HNE was also detected in other non-ERassociated proteins, including caspase-8 isoform C, heat shock cognate 71, protein-glutamine ␥-glutamyltransferase 2, and E3 ubiquitin ligase proteins (Table 1). Together, these data suggest that HNE modifies several proteins in endothelial cells, including ER-resident chaperones.
HNE Triggers Unfolded Protein Response-Given the observations that protein-HNE adducts co-localize with ER proteins and that several ER proteins were modified by HNE, we examined the possibility that HNE induces ER stress. Indeed, microarray analysis of RKO cells treated with HNE was shown to increase in ER stress genes, such as HERP, ATF3, and GADD34 (23), and treatment with HNE has been reported to increase the phosphorylation of IRE1 and eIF2␣ in human endothelial cells (24). Nevertheless, the impact of HNE on the UPR in its entirety has not been assessed, and the role of ER stress in mediating the biological response to HNE has not been studied. Hence, we examined the effect of HNE on each of the three canonical arms of the UPR: XBP-1 splicing, the activation of the PERK-eIF2a and ATF6, and the transcription factors ATF3 and ATF4, which are induced by ER stress (25).
To examine whether HNE increases XBP-1 splicing, the cells were incubated with 5-25 M HNE for 2 h, and changes in the splicing of XBP-1 mRNA were measured. As shown in Fig. 2A, treatment with HNE led to a marked increase in the intensity of the spliced form. In untreated cells, there was no increase in the spliced form of XBP-1, which was robustly increased upon treatment with the known UPR inducer thapsigargin. These data suggest that HNE stimulates the IRE-1 pathway of the UPR.
To examine how HNE affects the PERK-eIF2␣ pathway, the cells were incubated with HNE for 30 min and stained with anti-phospho-PERK antibody. Treatment with HNE led to a 2-fold increase in PERK phosphorylation (Fig. 2B, i). HNE also led to a concentration-dependent increase in eIF2␣ phosphorylation (Fig. 2B, ii). These results show that HNE-induced ER stress leads to the activation of the PERK-eIF2␣ pathway of the UPR. However, under conditions of nutrient deprivation and viral infection (26), PERK could also be activated by a PERKindependent mechanism. Therefore, to determine the relationship between PERK and eIF2␣ phosphorylation, we examined the effect of PERK gene silencing on HNE-induced phosphorylation of eIF2␣. As shown in Fig. 2B, iii, transfection of HUVEC with PERK siRNA decreased the abundance of PERK protein by ϳ70% and inhibited HNE-induced eIF2␣ phosphorylation by The MS2 spectrum collected included extensive peptide fragmentation data allowing for Ͼ93% sequence coverage, including the ϩ2 b15-NH3-NH3 ion and sequence identification after database searching corresponding to the ER-associated protein GRP78. Similar data are provided for assignment of other HNE-modified ER proteins listed in Table 1. ϳ70%, whereas cells transfected with the oligofectamine (vehicle) or non-targeting siRNA (NT-siRNA) were not affected. These data indicate that HNE stimulates eIF2␣ phosphorylation via PERK.
Incubation of the cells with HNE led to a 4-fold increase in ATF4 protein (data not shown). Moreover, as shown before (23), treatment with HNE led to a 4-fold increase in ATF3 protein (data not shown). Transfection of HUVEC with PERK or eIF2␣ siRNA attenuated HNE-induced ATF3 up-regulation (data not shown), suggesting that HNE-induced increase in ATF3 is via the PERK/eIF2␣ pathway. Treatment with HNE also led to a significant increase in CHOP (data not shown). Collectively, these observations suggest that HNE activates the PERK/eIF2␣ pathway, leading to an increase in transcription factors that are activated upon ER stress.
To examine the effect of HNE on the adaptive phase of the UPR, we examined changes in ATF6. Whereas control cells had minimal ATF6 in the nuclei (Fig. 2C, i), incubation of the cells with HNE resulted in a 6-fold increase in the nuclear translocation of ATF6 as compared with untreated cells, suggesting that HNE activates ATF6. To examine whether activation of ATF6 leads to the induction of ER-resident chaperones, which ameliorate ER stress and attenuate the UPR, cells were incubated without or with HNE, and expression of mRNA of ER chaperones was examined by quantitative RT-PCR. As shown in Fig.  2C, ii, treatment with HNE led to a significant increase in the expression of GRP78 and HERP mRNA. HNE-induced increase in GRP78 mRNA was accompanied by a Ͼ2-fold increase in GRP78 protein in cells treated with HNE (data not shown). These data suggest that in addition to triggering the alarm phase of the UPR, HNE also activates adaptive UPR responses. Taken together, these results indicate that HNE activates all three of the canonical pathways of the UPR.
Depletion of Reduced Glutathione Is Not Sufficient to Cause ER Stress-Having found that HNE induces ER stress, we next examined the effect of HNE on ROS formation, activation of ROS-dependent phase II detoxification enzyme, heme oxygenase-1, [Ca 2ϩ ] i , and glutathione levels in endothelial cells. Treatment of HUVEC with HNE significantly increased the formation of ROS and heme oxygenase-1, but it did not affect

Modification of proteins by HNE in HUVEC
Site-specific modification of cysteine (C), histidine (H), or lysine (K) amino acid side chains by HNE forming a stable Michael adduct has been noted with lower-case letters (c, h, or k). Site-specific modification of C, H, or K amino acid side chains by HNE with subsequent loss of water to form the Schiff-base (HNE-⌬H2O) has been noted with lower-case and bolded letters (c, h, or k). Amino acids that appear prior to and following the peptide are denoted using parentheses.  (27). Therefore, some of the cellular effects of HNE could be secondary to GSH depletion. To examine this possibility, HUVEC were incubated with HNE, and GSH content was measured by FACS using MCB as a fluorescent probe for detecting reduced thiols. Incubation with HNE led to a steady decline in MCB fluorescence such that after 60 min, Ͼ30% of MCB fluorescence was lost (Fig.  3A). Similar to HNE, t-BHP (25 M), which oxidizes GSH to GSSG (28), also caused a large decrease in MCB fluorescence. These observations suggest that both HNE and t-BHP lead to a decrease in intracellular thiols, mainly GSH. To examine whether thiol depletion is sufficient to cause ER stress, we measured the splicing of XBP-1 and phosphorylation of eIF2␣. However, although t-BHP caused GSH depletion (Fig. 3A), it did not increase XBP-1 splicing (Fig. 3B) or eIF2␣ phosphorylation (Fig. 3C). Treatment of cells with the saturated aldehyde nonanal did not cause GSH depletion (Fig. 3A), XBP-1 splicing (Fig. 3B), or eIF2␣ phosphorylation (Fig. 3C). Together, these observations suggest that although the electrophilic nature of HNE is required for inducing ER stress, depletion of GSH or an increase in ROS formation is not sufficient to trigger the UPR by HNE. HNE Causes Endothelial Activation-To elucidate the role of ER stress in HNE-induced endothelial activation, we tested whether inhibition of ER stress by chemical chaperones would prevent endothelial activation by HNE. Treatment with HNE led to a concentration-dependent increase in the surface expression of ICAM-1 and the adhesion of THP-1 cells to HUVEC (Fig. 4A, i and ii). HNE also increased the transmigration of THP-1 cells (Fig. 4A, iii) and the production of proinflammatory cytokines (Fig. 4B). Preincubation with the chemical chaperone PBA prevented HNE-induced increase in ATF3 (Fig. 5A), ATF4 and heme oxygenase-1 protein (data not shown), suggesting that an increase in protein folding capacity diminishes HNE-induced ER stress. Importantly, preincubation with PBA prevented HNE-induced increase in the adhesion of THP-1 cells to HUVEC (Fig. 5B). It also prevented the increase in the production of IL-8 in HNE-treated cells (Fig. 5C). These observations are consistent with the view that HNE-induced endothelial activation is mediated in part by ER stress.

ER-associated proteins
Although PBA is an effective chaperone, its nonspecific effects cannot be entirely excluded. Therefore, for evaluating the role of ER stress more specifically, we examined the effects of ATF6 overexpression. Transfection with ATF6 virus increased ATF6 mRNA (Fig. 6A) and led to a concentration-dependent increase in GRP78 (Fig. 6B), GRP94, and PDI protein (data not shown), suggesting that overexpression of ATF6 increases the overall protein folding capacity of the cell. To examine how this would modify the effects of HNE, we compared cell adhesion and IL-8 synthesis in cells infected with the ATF6 virus with those transduced by the empty vector alone. HNE treatment of cells transduced with the null vector resulted  in an increase in adhesion of THP-1 cells; however, this was significantly less in cells infected with the ATF6 virus (Fig. 6C). Similarly, HNE increased the transcription of IL-8 gene by Ͼ20-fold in cells infected with empty vector, whereas infection with the ATF6 virus significantly decreased the induction of IL-8 by HNE (Fig. 6D).
To further assess the specificity of the UPR in mediating HNE-induced endothelial activation, we examined the role of IRE-1. For this, the cells were treated with IRE-1 siRNA. Transfection with the siRNA inhibited the expression of IRE-1 mRNA by 90% (data not shown) and attenuated HNE-induced IL-8 production (data not shown). These results suggest that the IRE-1 pathway of the UPR is a critical regulator of HNEinduced inflammation.
To determine whether HNE activates the endothelium in vivo, 1 mg/kg HNE was injected in the tail vein of adult male Sprague-Dawley rats. Thirty minutes after injection, the plasma HNE concentration was ϳ0.1 M. In vehicle-treated rats, the plasma concentration of HNE was 0.015 M, indicating that the injection led to a marked increase in plasma HNE concentration. Eighteen hours after treatment, ER stress was appreciably increased in the cremasta of HNE-treated rats versus rats that were treated with the vehicle alone as evident by the increased staining for KDEL (Fig. 7A). Moreover, both the rolling of leukocytes (Fig. 7B) and their firm adhesion to the endothelium (Fig. 7C) was increased by HNE. Pretreatment with PBA (0.25 g/kg) or TUDCA (0.25 g/kg) attenuated both ER stress and leukocyte rolling (Fig. 7B) as well as leukocyte adhesion (Fig.  7C) to the cremaster endothelium. Taken together, these data support the notion that HNE increases the adhesiveness of the intact endothelium and that this is prevented by decreasing ER stress.

DISCUSSION
The findings of this study suggest that the lipid peroxidation product HNE induces ER stress, which results in the activation of both the alarm and the adaptive phases of the UPR. We found that treatment of endothelial cells with HNE stimulates all three of the canonical pathways of the UPR, leading to the activation of several stress-sensitive transcription factors, such as ATF3, ATF4, and CHOP. Although treatment with HNE led to the modification of several proteins, changes in the ER appear to be particularly significant because inhibition of ER stress or an increase in the protein folding capacity of the cell prevented HNE-induced increases in endothelial adhesion and cytokine production. Collectively, these observations reveal a new mechanism of action of lipid peroxidation products, and they indicate a novel link between ER stress and endothelial activation caused by lipid peroxidation products, such as HNE.
ER stress and the UPR are increasingly recognized as important regulators of cell function (26). Conditions under which the influx of nascent, unfolded proteins exceeds the folding capacity of the ER trigger the UPR to restore homeostasis. To match demand to capacity, protein translation and transcription of secreted proteins are transiently inhibited, whereas the degradation of unfolded proteins is increased. Changes in the protein folding status are transmitted to the nucleus by the UPR, which results in ER expansion and an increase in the protein folding capacity of the cell. Conditions under which homeostasis cannot be restored lead to apoptosis. Multiple conditions can trigger the UPR, including viral infections, glucose deprivation, and changes in ATP or calcium levels. Our work adds to the growing list of these conditions, by demonstrating that electrophilic injury by lipid peroxidation products like HNE is a potent, robust, and complete initiator of the UPR.
Although redox status has been considered to be an important regulator of the UPR, our results show that treatment with t-BHP did not trigger the UPR, although it depleted GSH and increased ROS production. Therefore, GSH depletion and ROS production in HNE-treated cells are unlikely to be important mediators of the UPR. Instead, it seems that the ability of HNE to cause ER stress may be related to the modification of multiple proteins that could overwhelm the total proteolytic capacity of the cell and thereby trigger the UPR. Previous studies have shown that inhibition of the proteosome can be propagated to the ER to cause ER stress and activate the UPR (26). However, the observation that HNE stimulates the UPR suggests that ER itself may be a direct target of electrophilic injury. The cytosol, the mitochondria, and the nuclei maintain a high GSH/GSSG ratio (ϳ100:1), which prevents electrophilic protein modification and oxidative protein damage. In contrast, the ER maintains a highly oxidized state (GSH/GSSG ratio 1:1-3:1) (29), and therefore proteins in the ER may be particularly vulnerable to oxidative damage or modification by electrophiles, such as HNE. Moreover, the cytosol contains several aldehyde-metabolizing enzymes (aldehyde dehydrogenases and reductases and GSTs) and cofactors (NAD, NADPH, and GSH) required to metabolize and detoxify aldehydes. These defenses are not present in the ER; therefore, even low levels of lipid peroxidation products could readily diffuse into the ER and modify proteins without encountering significant defense.
That the ER is a direct target of HNE is consistent with our immunohistochemical data indicating a significant overlap between protein-HNE and ER staining and with our mass spectrometric data showing HNE adduction of several ER proteins. The vulnerability of the ER to HNE is further supported by the robust activation of all three of the pathways of the UPR (IRE-1, PERK, and ATF6) in HNE-treated cells. These signaling events elicit the adaptive, alarm, and apoptotic responses in an attempt to restore ER function. Because XBP-1 splicing occurs exclusively by IRE-1 activation, the formation of the spliced form of XBP-1 is considered an essential mediator of ER stress (26). Our data showing that HNE causes the cleavage of the XBP-1 mRNA suggest that HNE-induced ER stress activates the IRE-1 pathway of the UPR. Knockdown of IRE-1 completely prevented the induction of IL-8 in HNE-treated cells, indicating that the activation of IRE-1 upon ER stress is an obligatory requirement for IL-8 production in HNE-treated cells.
In addition to IRE-1, the PERK/eIF2␣ pathway of the UPR was also activated by HNE. Under most conditions, PERK is activated exclusively by ER stress; however, its downstream effector kinase eIF2␣ could also be phosphorylated by other kinases, including PKR, GCN2, and HRI (26). Our data show that transfection with PERK siRNA attenuated the HNE-induced activation of eIF2␣, suggesting that HNE causes the phosphorylation of eIF2␣ via PERK. Whereas eIF2␣ phosphorylation suppresses global protein synthesis to decrease ER load, it selectively up-regulates ATF4. This in turn can activate the transcription factor ATF3, which is a negative regulator of LPSinduced cytokine production (30). We observed that HNE upregulates both ATF4 and ATF3. These observations are in agreement with an earlier report that HNE induces ATF3 in RKO human colorectal carcinoma cells, albeit at a much higher concentration (23). A comparison of these data suggests that endothelial cells might be more sensitive to HNE-induced ATF3 induction than cancer cells. Moreover, because HNEinduced up-regulation of ATF3 was attenuated by PBA and knockdown of PERK and eIF2␣, it appears that HNE induces ATF3 by activating the PERK/eIF2␣ pathway of the UPR.
ATF6 is the transducer of the protective pathway of the UPR. It increases the transcription of genes encoding molecular chaperones and those involved in protein folding. ATF6 also induces the transcription of XBP-1 (31). Our data show that HNE activates ATF6 as evinced by the nuclear translocation of ATF6 and the increase in the transcription of HERP and GRP78 genes in HNE-treated endothelial cells. Significantly, we observed that adenoviral infection of endothelial cells with ATF6 the up-regulated molecular chaperones GRP78, GRP94, and PDI and diminished HNE-induced IL-8 expression, suggesting that activation of this protective pathway of the UPR plays a critical role in preventing aldehyde-induced endothelial activation.
Our observation that HNE causes endothelial activation by inducing ER stress suggests that conditions that are associated with an increase in lipid peroxidation could lead to vascular inflammation via ER stress. Notably, the accumulation of proteins modified by lipid peroxidation products in atherosclerotic lesions is associated with an increase in the expression of ATF3 and ATF4 specifically in areas that colocalize with aldehydemodified proteins (32). Although a causative role of ER stress in atherosclerotic lesion formation remains to be established, the findings of this study support the possibility that induction of the UPR in the endothelium by products of lipid peroxidation may be an early initiating event in atherogenesis.
In addition to atherogenesis, several other disease conditions, such as heart failure, Parkinson and Alzheimer diseases (33), obesity, and insulin resistance (34) are associated with tissue accumulation of HNE-modified proteins, indicating on-going oxidative damage. Significantly, these pathological states are also associated with ER stress (35). That ER stress contributes to the development of these conditions is suggested by the observation that treatment with chemical chaperones inhibits atherosclerosis (36) and prevents insulin resistance and restores glucose levels in ob/ob mice (12). Similarly, ER stress has also been implicated in the development of neurodegenerative disorders, such as Alzheimer and Parkinson diseases (37). Because these disease states induce oxidative injury, it is likely that aldehydes generated by oxidized lipid may be important triggers of ER stress during the manifestation/progression of these diseases.