Dietary oxidized n-3 PUFA induce oxidative stress and inflammation: role of intestinal absorption of 4-HHE and reactivity in intestinal cells^[S]

Materials

4-HHE, 4-HNE, and trideuterated compounds were synthesized according to Soulère et al. (17). Omegavie® Tuna oil 25 DHA-flavorless as a source of triacylglycerol rich in long-chain n-3 fatty acids (TG-DHA, 26% of DHA), lecithin rich in DHA (PL-DHA, 41% of DHA), and kiwi seed oil were provided by Polaris (Pleuven, France). Lard was supplied by Celys (Rezé, France), sunflower oil was from Lesieur® (Asnières-sur-Seine, France), and oleic sunflower oil from Olvéa (Marseille, France). Vegetable lecithin rich in linoleic acid 18:2 n-6 (PL-LA) was from Lipoid (Frigenstrasse, Germany).

Preparation of unoxidized and oxidized lipid blends and mice diets

Four lipid blends were prepared at the labscale to obtain similar fatty acid composition and quantities in the four diets and similar amounts of phospholipids and triacylglyerols. In these blends, DHA was supplied either in the form of triacylglycerols (TG diet) or phospholipids (PL diet); i.e., the name chosen for the diets reflects the type of molecules that carry long-chain n-3 PUFA in the diet. The oxidized lipid blends (TG-ox and PL-ox, respectively) were prepared as follows. Primary lipid mixtures for PL and TG groups were prepared with a small proportion of lard to maintain oxidability of PUFA. These preliminary oil mixtures were dispersed in aqueous phase (mineral water; Evian) to prepare 30% w/w oil-in-water emulsions. The emulsions were then kept at 50°C in the dark with continuous shaking until the oxidation level was considered sufficient according to our previous experiments (estimated α-tocopherol contents of the blends decreased by 50%). Oxidized emulsions were then lyophilized, and the resulting oxidized lipid mixtures completed with the necessary quantity of lard to reach the required final composition of lipid blends. The composition of the four diets is reported in Table 1.

Animals and diets

Male C57BL/6 mice (8 wk, 20g) were from Janvier SA (Le Genest Saint-Isle, France) and were housed in a temperature-controlled room (22°C) with a 12 h light/12 h dark cycles. After 2 weeks of chow diet, mice were randomly divided into four groups of 12 mice and fed one of the 4 high-fat diets containing n-3 PUFAs: unoxidized groups: PL, TG; oxidized groups: PL-ox, TG-ox. Animal experiments were performed under the authorization n°69-266-0501 (Direction Départementale des Services Vétérinaires du Rhône). All experiments were carried out according to the guidelines laid down by the French Ministry of Agriculture (n° 87-848) and the E.U. Council Directive for the Care and Use of Laboratory Animals of November 24th, 1986 (86/609/EEC), in conformity with the Public Health Service (PHS) Policy on Humane Care and Use of Laboratory Animals. COS holds a special license (n° 69266257) to experiment on living vertebrates issued by the French Ministry of Agriculture and Veterinary Service Department. Body weight was measured twice weekly and food intake was measured weekly. After 8 weeks, mice were euthanized by intraperitoneal (IP) injection of sodium pentobarbital (60 mg/kg). Plasma, liver, white adipose tissue (epididymal, retroperitoneal, subcutaneous) and small intestine mucosa were collected.

Caco-2/TC7 cell culture and treatment

Caco-2/TC7 cells, which are the widely used in vitro model of human origin to test intestinal absorption of lipids, were provided by Monique Rousset (Centre de Recherche des Cordeliers, Paris, France) and used between passages 35 and 45. Cells were seeded in 75 cm² flasks (Falcon; Becton Dickinson) until 80–90% confluence. They were grown in complete DMEM (Gibco) supplemented with 20% heat-inactivated FBS (Gibco), 1% nonessential amino acids (Gibco), and 1% antibiotics (penicillin/streptomycin; Gibco) and then maintained under a 10% CO₂ atmosphere at 37°C. For experiments, cells were seeded at a density of 25 × 10⁴ cells per filter on microporous (0.4 μm pore size) polyester filters (Transwell; Corning, USA) and grown to confluence in complete medium, which was routinely reached 7 days after seeding. The cells were used 21 days after seeding. Monolayers were incubated with 4-HHE or 4-HNE in the apical compartment (1–100 µM brought in DMSO at 0.5% in the final medium), and the basolateral compartment received serum-free DMEM. After incubation, basolateral media and cells were collected.

Animal treatment with 4-HHE

After an overnight fast, three groups of four male C57/BL6 mice (8 weeks old, 22 g) were given a single application of 4-HHE diluted in water via 0.5% DMSO by oral gavage at a dosage of 10 mg/kg body weight and then were euthanized 1, 2, and 4 h after gavage. A fourth group of mice was euthanized immediately after gavage for the baseline control. For euthanizing, mice were anesthetized by IP injection of pentobarbital (35 mg/kg), and blood was collected by cardiac puncture with heparinized syringes. Plasma and small intestine mucosa (duodenum and jejunum) were collected.

4-Hydroxy-2-alkenals: derivatization, analysis, and quantification

4-hydroxy-2-alkenals were derivatized from 400 µl of basolateral media of Caco-2/TC7 or from 300 µl of plasma, under argon, and in lipid blends as described previously (18). Deuterated 4-HNE and 4-HHE (20 ng), used as internal standards, were added to the samples. Briefly, they were treated with O-2,3,4,5,6-pentafluorobenzyl hydroxylamine hydrochloride. After acidification with H₂SO₄, pentafluorobenzyloxime derivatives were extracted with methanol and hexane. The hydroxyl group was then converted into trimethylsilylether after an overnight treatment with N,O-bis(trimethylsilyl)trifluoroacetamide at room temperature. The pentafluorobenzyloxime trimethylsilylether derivatives of 4-HHE (O-PFB-TMS-4-HHE) and 4-HNE (O-PFB-TMS-4-HNE) were then analyzed by GC-MS using negative ion chemical ionization (NICI) mode on a Hewlett-Packard quadripole mass spectrometer interfaced (18) with a Hewlett-Packard gas chromatograph (Les Ullis, France).

IL-6 and MCP-1 measurements

IL-6 and MCP-1 (Clinisciences, Nanterre, France) were assayed by ELISA kit according to the manufacturer's instructions.

Plasma triacylglycerols and NEFA measurements

Plasma triacyglycerols (TAG) were measured with the triglyceride PAP kit (BioMérieux, Marcy l'Etoile, France) as previously described (19). Plasma TAG concentration was calculated by subtracting the free glycerol in plasma measured with the glycerol UV-method (R-Biopharm/Boehringer, Mannheim, Germany). Plasma NEFA was measured using NEFA-C kit (Wako Chemicals, Neuss, Germany) (19).

Fatty acid analysis

Total lipids were extracted from 35 µl of plasma as described previously (19). The organic phase was evaporated under N₂, and total fatty acids were transesterified using boron trifluoride in methanol (BF₃/methanol) (19). The FA methyl esters were then analyzed by GC using a DELSI instrument model DI 200 equipped with a fused silica capillary SP-2380 column (60 m × 0.22 mm). Heptadecanoic acid (C17:0; Sigma, France) was used as an internal standard.

Quantification of free malondialdehyde and hydroperoxides in oil mixtures

Thiobarbituric acid-malondialdehyde (MDA) adducts were separated using a method adapted from different authors (20, 21) by HPLC and measured by fluorimetry using an external calibration curve (excitation 535 nm, emission 555 nm). Hydroperoxides were quantified in lipid blends according to a method adapted from Nourooz-Zahed et al. (22).

Quantitative PCR analysis

Total RNA was extracted from Caco-2/TC7, duodenum, jejunum, and ileum of mice using the NucleoSpin^® RNA/Protein kit (Macherey Nagel, Duren, Germany). cDNAs were synthesized from 1 µg of total RNA in the presence of 100 units of Superscript II (Invitrogen) with a mixture of random hexamers and oligo (dt) primers (Promega, Charbonnières, France). The amount of target mRNAs was measured by RT, followed by real-time PCR, using a Rotor-Gene Q (Qiagen, France). Primer sequence and RT-quantitative PCR conditions are available upon request (cyrille.debard@univ-lyon1.fr). Hypoxanthine guanine phosphoribosyl transferase (HPRT) mRNA level and TATA-box binding protein (TBP) mRNA were used to normalize data of duodenum, jejunum, ileum, and Caco-2/TC7 cells, respectively.

Western blot analysis

Total proteins from Caco-2/TC7 and jejunum of mice were extracted with the NucleoSpin^® RNA/Protein kit (Macherey Nagel). A total of 40 μg of protein from each sample were subjected to 10% SDS-PAGE and transferred to a polyvinylidine fluoride (PVDF) membrane. Membranes were blocked with kit reagent and incubated overnight with antibodies according to the manufacturer's recommendations. Antibodies against total NF-κB P65 and phospho-NF-κB P65 (Ser 529) were obtained from Signalways Antibody (Clinisciences, France); phospho-IκBα (phosphoS32+S36) was from Abcam (Cambridge, UK); and β-actine was from Sigma, France. After incubation with secondary antibody, blots were developed with a commercial kit (WesternBreeze Chemiluminescent, Invitrogen, France). Quantification was performed by densitometry analysis of specific bands on immunoblots with Quantity One software (Bio-Rad, Marne-la-Coquette, France).

Immunohistochemistry

Duodenal tissues were removed, fixed in 90% ethanol for 24 h at −20°C, and processed into paraffin. Serial paraffin sections (4 µm) were rehydrated, and endogenous peroxidase activity was quenched with 20 min incubation in 5% H₂O₂/PBS. After incubation in 2.5% normal horse blocking serum (Impress, Vector), sections were incubated for 60 min at room temperature with anti-lysozyme (1/100; Zymed Laboratories) antibody diluted in blocking solution. The immune reactions were then detected by incubation with a ready-to-use peroxidase-labeled secondary reagent, ImmPRESS (MP-7401 for rabbit antibodies; Abcys, Paris, France) (30 min, room temperature). Control experiments were performed simultaneously omitting the primary antibody or incubating with preimmune rabbit serum. The sections were then counterstained and mounted.

Quantitative analysis for Paneth cells

The Paneth cell lineage was analyzed by assessing the percentage of crypt cross-sections with Paneth cells (per 80 crypts). For this purpose, four to six sections were analyzed per mouse. A crypt was considered when it was cut along or nearly along the length of the crypt lumen (at least two thirds of the length of the crypt). All slides were analyzed by a single investigator who was blinded to the treatment groups.

Protein carbonyl content

Protein carbonyl content was determined with an Oxyblot Oxidized Protein Detection Kit from Chemicon (Hampshire, UK). The carbonyl groups in the protein chains were derivatized into dinitro-phenyl-hydrazone by reaction with dinitro-phenyl-hydrazine (DNPH) according to the manufacturer's instructions. After the derivatization of the protein sample (20 µg), one-dimensional electrophoresis was carried out on a 10% SDS-PAGE gel, and proteins were transferred to PVDF membranes. After incubation with anti-DNP antibody, the blot was developed with a chemiluminescence detection system. The intensity of each line was measured and normalized by mean control levels.

Dot blot analysis

Fifty milligrams of total proteins from Caco-2/TC7 cells or from duodenum and jejunum of mice were spotted onto nitrocellulose membrane. Blocked membrane was incubated overnight with anti-HNE-adducts (Calbiochem 393207, San Diego, CA, USA) or anti-HHE-adducts (CosmoBio N213730-EX, Tokyo, Japan) antibodies. Blots were developed with a commercial kit (WesternBreeze Chemiluminescent, Invitrogen, France). The quantification was performed by densitometric analysis of specific spots on immunoblots using Quantity One software.

Statistical analysis

All data are presented as means ± SEM and were analyzed using Statview 5.0 software (Abacus Concept, Berkeley, CA). One-way ANOVA followed by Fisher PLSD was used for i) the dietary study to compare PL, PL-ox, TG, and TG-ox groups, ii) the gavage study to compare plasma alkenal concentrations as a function of time, and iii) the Caco-2 cell studies to compare treatment effects. Two-way ANOVA followed by Fisher PLSD was used to compare oxidized versus unoxidized groups globally in the dietary study (mice of oxidized groups versus mice of unoxidized groups). Differences were considered significant at the P < 0.05 level.

Diet compositions and biometric data

Table 2 shows that we succeeded in producing lipid mixtures containing different amounts of oxidation products, especially 4-HHE, which was significantly higher in oxidized versus unoxidized lipid blends. MDA and hydroperoxides in oxidized oils were in a range considered as acceptable for human consumption. Noticeably, oxidation did not impact the fatty acid profile (Table 2); n-3 PUFA content and n-6/n-3 ratio were consistent with dietary recommendations. Mice in oxidized and unoxidized groups did not differ in final body weight gain, liver weight, white adipose tissue, plasma TAG, or plasma NEFA (Table 3).

Aldehyde stress and inflammatory markers in plasma of mice fed oxidized n-3 diets

Because consuming oxidized n-3 diets could contribute to circulating biomarkers of lipid peroxidation such as 4-HHE and 4-HNE, we measured these 4-hydroxy-2-alkenals in plasma. Fig. 1A shows a 4-fold increase of 4-HHE concentration in the plasma of oxidized groups (427–508 nM) versus unoxidized groups (89–128 nM) (P < 0.01). However, the concentration of 4-HNE (from n-6 PUFA) did not differ among groups (from 6 to 9 nM; Fig. 1B).

FA composition of total plasma lipids was characterized to test whether chronic ingestion of oxidized diet would affect PUFA metabolism. Plasma FA profile (see supplementary Table I) actually reflected the composition of ingested dietary fats. The proportions of n-3 PUFA (EPA, DHA, α-linolenic acid) and n-6 PUFA (arachidonic acid, 20:4 n-6; linoleic acid, 18:2 n-6) were similar in plasma of each oxidized and corresponding unoxidized groups.

Regarding metabolic inflammation, there were higher concentrations in plasma of the proinflammatory cytokine IL-6 (Fig. 1C) and the chemokine MCP-1 (Fig. 1D) in oxidized groups.

Altogether, high-fat diets containing realistically oxidized n-3 PUFA induced the highest amounts of 4-HHE and inflammatory markers in plasma. We therefore questioned whether oxidized oils would affect the small intestine, which is the primary defense line of the host regarding ingested products.

Effects of oxidized diets on markers of stress and inflammation in the small intestine

We analyzed the small intestine mucosa regarding the expression of genes known to be specifically involved in cell defense against oxidative stress and detoxification. The expression of GPx2, a gastrointestinal glutathione peroxidase mainly expressed in the intestine, was increased specifically in the jejunum of oxidized groups (Fig. 2D; P < 0.05). Moreover, we examined the gene expression of two markers that are linked to endoplasmic reticulum (ER) stress. Mice fed oxidized diets exhibited a significantly higher expression of prosurvival factor glucose-regulated protein 78 (GRP78), both in the duodenum and in the jejunum (Fig. 2B, E; P < 0.05). The proapoptotic C/EBP homologous protein (CHOP) was significantly higher in the jejunum of oxidized groups (Fig. 2F). However, on the whole, no effects of n-3 PUFA oxidation were observed in the ileum. Regarding inflammation, results show that the jejunum of mice fed oxidized diet exhibited an increased phosphorylation of NF-κB P65 (Fig. 3A; P < 0.0001), a major transcription factor involved in inflammation, that was associated with a significant increase of phosphorylated IkappaB α (IκBα), an inhibitor NF-κB protein (Fig. 3B; P < 0.0001). The activation of NF-κB requires the phosphorylation and degradation of the IκB protein that prevents the nuclear translocation of active NF-κB. We also analyzed Paneth cells in the duodenum, which are involved in innate immunity by sensing bacteria and by discharging antimicrobial peptides including α-defensins. Paneth cell number was significantly decreased after oxidized versus unoxidized diets (Fig. 3C). Moreover, the gene expression of lysozyme 1 (Lyz-1) produced by Paneth cells was decreased in the duodenum in the PL-ox group compared with the PL group (93 ± 20 versus 315 ± 161; values normalized to the levels of HPRT mRNA); no differences were observed lower in the intestine in the jejunum (15–17 ± 2 mRNA/HPRT regardless of oxidation) nor in the ileum (∼250 ± 100 mRNA/HPRT regardless of oxidation).

Kinetics of intestinal absorption of 4-HHE in mice and protein modifications in the small intestine

To test the hypothesis that the increased plasma level of 4-HHE following consumption of oxidized diet could be partly explained by the absorption of 4-HHE in the small intestine, we measured the levels of 4-HHE in the plasma of fasted mice after gavage with 4-HHE. Fig. 4A shows that 4-HHE intragastrically fed to mice at 88 µmol/kg body weight was partly absorbed after 1 h. Peak plasma concentration of 184 nM was reached 2 h after treatment (P < 0.001 versus baseline), corresponding to ∼0.1% of the fed dose. Importantly, no concomitant differences were observed in the plasma levels of 4-HNE (Fig. 4B).

We further tested the formation of Michael adducts of proteins in the duodenum and jejunum. As shown in Fig. 4C and D, there were significant increases of 4-HHE protein adducts that peaked at 2 h postgavage both in the duodenum (P < 0.001 versus baseline) and in the jejunum (P < 0.05 versus baseline). The observed decrease in adduction at 4 h reveals the capacity of intestinal mucosa to cope with carbonyl stress. We also observed an activation of GPx2 mRNA in the duodenum 4 h after gavage with 4-HHE: 80 ± 32 versus 38 ± 12 at baseline (mRNA/HPRT, P = 0.013); this effect was no longer observed in the jejunum (10 ± 3 at 4 h versus 9 ± 6 at baseline).

Absorption of 4-hydroxy-2-alkenals in vitro by Caco-2/TC7 cells

To further evaluate the capacity of enterocytes to absorb 4-hydroxy-2-alkenals, we incubated Caco-2/TC7 cells for 24 h with increasing concentrations of 4-HHE or 4-HNE. 4-Hydroxy-2-alkenals were devoid of cytotoxic effects as checked by the absence of leakage of lactate dehydrogenase and did not alter transepithelial electrical resistance (TEER), meaning that cell monolayer maintained its integrity (data not shown).

Fig. 5 shows that the basolateral medium (BLM) concentration of 4-HHE was significantly increased in a dose-dependent manner (P < 0.01 versus control). Meanwhile, 4-HNE was increased from 50 µM compared with untreated cells. Detected 4-hydroxy-2-alkenals concentrations in BLM corresponded to ∼0.2–1% of incubated amount. Noticeably, using similar concentrations, 4-HHE was more abundant in the BLM than 4-HNE (P < 0.0001).

Protein modifications by 4-hydroxy-2-alkenals in Caco-2/TC7 cells

We quantified protein carbonylation and the formation of aldehyde-protein adducts. Cell incubated for 24 h with increasing concentrations of 4-HHE or 4-HNE exhibited an increased level of reactive carbonyl group in proteins in a dose-dependent manner from 50 µM (P < 0.0001) versus untreated cells (Fig. 6A, B). Fig. 6C also reveals a significant increase of HNE-protein adducts after incubating cells for 2 h with 4-HNE from 10 µM in a dose-dependent manner versus untreated cells. Similar treatment using 4-HHE resulted in increased amounts of HHE-protein adducts from 50 µM (P < 0.05; Fig. 6D).

GPx2 and ER stress-linked gene expression in Caco-2/TC7 cells

We then sought to determine whether some antioxidant systems were changed due to 4-hydroxy-2-alkenals. Fig. 7A and B shows that GPx2 expression tended to increase after 24 h at 50 µM of either 4-HNE or 4-HHE (P = 0.06) and significantly increased at a concentration of 100 µM (P < 0.05) compared with control. Moreover, 4-HNE and 4-HHE (100 µM) led to increased expression of GRP78 (P < 0.05) (Fig. 7C, D).