Introduction

Antioxidants are currently considered as one of the major classes of functional food components. In addition to natural antioxidants such as polyphenols or carotenoids, secondary antioxidants formed during food processing have attracted growing attention (Urquiaga and Leighton 2000; Rousseau et al. 1992; van Boekel et al. 2010). A diet rich in thermally processed food, for example, increased resistance of plasma and serum LDL against oxidative stress (Seiquer et al. 2008; Dittrich et al. 2009). In particular, Maillard reaction products (MRPs) and melanoidins, which are formed during food processing by the reaction of amino acids with sugars, show potent oxygen and radical scavenging activity (Dittrich et al. 2003; Samaras et al. 2005; Borelli et al. 2002). In addition to direct antioxidative effects, MRPs and melanoidins may also exert indirect antioxidative activity by inducing the expression of cytoprotective proteins in cells. Pre-incubation of HepG2 cells with melanoidins, for example, increased the concentration of intracellular reduced glutathione (GSH) and protected cells efficiently against oxidative stress (Goya et al. 2007; Martin et al. 2009). GSH is synthesised enzymatically by γ-glutamylcysteine synthetase and glutathione synthetase; both enzymes are regulated by the nuclear factor-erythroid-2-related factor 2 (Nrf2). This redox-sensitive transcription factor is ubiquitously expressed in tissue with particularly high levels in the intestine, lung and kidney (McMahon et al. 2001; Moi et al. 1994). In non-activated cells, Nrf2 is bound to the repressor Kelch-like ECH-associated protein 1 (Keap1) causing ubiquitination and degradation of Nrf2. Since Keap1 itself is anchored to the actin cytoskeleton, the inactive complex is restricted to the cytoplasm. Upon activation, Nrf2 is released from Keap1, translocates into the nucleus and binds to the antioxidant response element (ARE)/electrophile response element (EpRE) of certain genes. Genes that possess the respective elements in their promoter regions are thus regulated by Nrf2. Besides γ-glutamylcysteine synthetase and glutathione synthetase, phase II target genes such as NAD(P)H: quinone oxidoreductase 1, catalase and thioredoxin are up-regulated by Nrf2 (Surh et al. 2008). The proteins encoded by those genes are responsible for cellular detoxification and maintain antioxidant capacities, which may counteract unbalanced redox homoeostasis. Thus, dietary activation of Nrf2 could reverse biochemical dysfunction resulting from diseases related to oxidative stress and/or hyperglycaemia (Xue et al. 2008).

Activation of Nrf2 in the gut appeared to have several beneficial effects. Inflammatory bowel diseases, namely Crohn’s disease and ulcerative colitis, are characterised by an imbalance between reactive oxygen species (ROS) and mucosal antioxidant response (Kruidenier et al. 2003). Indeed, as shown in a mouse model, activators of Nrf2 such as prohibitin can improve colitis (Theiss et al. 2009). Enhanced intestinal Nrf2 signalling was further associated with a beneficial effect on gut barrier dysfunction (Jin et al. 2008), intestinal mucosal injury (Jin et al. 2009) and intestinal inflammation other than colitis (Jin et al. 2008). Moreover, there is growing evidence that Nrf2 is also a promising target for the prevention of colorectal cancer (Saw and Kong 2011; Pool-Zobel et al. 2005).

To date, defined phytochemicals such as sulphoraphane (broccoli), carnosol (rosemary) and curcumin (turmeric) as well as dietary plant extracts from broccoli, coffee and spices (turmeric, rosemary) have been identified as potent Nrf2 activators. They induce EpRE activity in liver HepG2 cells indicating a beneficial impact on the pathogenesis of oxidative stress associated disorders (Paur et al. 2008). Indeed, the consumption of cruciferous vegetables such as broccoli was shown to reduce the risk of cancer such as colon and prostate cancer in humans (Clarke et al. 2008).

The present study investigated the impact of MRPs on Nrf2 translocation in vitro in intestinal Caco-2 cells and macrophages, which are important intestinal cell types. Furthermore, Nrf2 activation was studied ex vivo in intact human gut tissue samples. The second part of this study focused on the involvement of ROS in Nrf2 activation. These effects were compared with investigations of coffee, a beverage rich in MRPs.

Materials and methods

Chemicals and cell culture

d-(-)-Ribose, l-lysine, catalase (from bovine liver), albumin, bovine serum albumin (BSA), lipopolysaccharide (LPS) (from Escherichia coli 055:B5), magnesium chloride, trypan blue, phenylmethylsulphonylfluoride (PMSF), dithiothreitol (DTT), Nonidet P-40 substitute (NP-40), and horseradish peroxidase (HRP)-conjugated anti-rabbit antibody (A6154) were purchased from Fluka-Sigma-Aldrich (Buchs, Switzerland). Rabbit polyclonal anti-Nrf2 antibody (AP06252 PU-N) was from Acris Antibody GmbH (Herford, Germany). Hydrogen peroxide (H2O2) (30%), 4-(2-hydroxyethyl)-1-piperazineethanesulphonic acid (HEPES), 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-tetrazolium bromide (MTT), and potassium chloride were supplied by Merck KgaA (Darmstadt, Germany). Roasted coffee beans and raw coffee beans (both 100% Coffea arabica) were obtained from a local retailer. Bio-Rad DC-Protein Assay Reagents Package was obtained from BioRad Laboratories GmbH (Munich, Germany), protease inhibitor tablet complete (PIS) from Roche Diagnostics GmbH (Mannheim, Germany) and enhanced chemiluminescence reagent (ECL) Western blotting detection reagents from Amersham Biosciences (Munich, Germany). Cell lines were purchased from ATCC (Wesel, Germany). HAM’s F12 (1,176 g/L NaHCO3, stable glutamine), trypsin (0.25%)/EDTA (0.02%), foetal calf serum (FCS), penicillin (10,000 U/mL)/streptomycin (10,000 μg/mL), phosphate buffered saline (PBS) Dulbecco and nonessential amino acids were supplied by Biochrom AG (Berlin, Germany) and minimum essential media (MEM) by Invitrogen (Darmstadt, Germany). Endo-paractol was obtained from Temmler Pharma GmbH & CO. KG (Marburg, Germany) and dimethylsulphoxide (DMSO) from Acros Organics (Nidderau, Germany). Nitrocellulose transfer membrane (pore size: 0.45 μm) was supplied by Schleicher & Schuell GmbH (Dassel, Germany).

Instruments

Universal grinder (Type A 10) and Ultra Turrax homogeniser were obtained from IKA®-Werke GmbH & CO. KG (Staufen, Germany). Mucosa oxygenator was purchased from Intestino-Diagnostics (Erlangen, Germany) and VersaDoc™ Imaging System from BioRad Laboratories GmbH (Munich, Germany). The counting chamber (Neubauer, improved) was from Carl Roth GmbH + Co. KG (Karlsruhe, Germany).

Preparation of Maillard reaction mixtures and a heated ribose solution

A heated Maillard reaction mixture (MRMh) was prepared from an equimolar solution of d-(-)-ribose and l-lysine (each 0.5 M), dissolved in PBS and adjusted to pH 9.4 with HCl (9.7 M). Filtered aliquots of 20 mL were heated in capped screw neck bottles for 30 min at 120°C in a drying oven. After cooling in an ice bath, the pH of MRMh was adjusted from pH 8.6 to 8.0 with HCl (9.7 M), and MRMh was filter-sterilised (0.22 μm). Aliquots were stored at −20°C. An unheated Maillard reaction mixture (MRMu) was prepared simultaneously, but without the heating step. For the heated ribose solution (ribh), the ribose was heated alone without addition of lysine as described for MRMh. Since the concentration of the reaction products in the final mixtures is not known, the given concentration reflects the initial concentration of the reactants. Assuming that about 30% of ribose and lysine were converted into Maillard products under the applied conditions, the content of MRP of 50 mM MRMh is equivalent to the concentration of melanoidins in a coffee brew (0.45 g melanoidins per 100 mL) (Fogliano and Morales 2011).

Preparation of roasted and raw coffee extract

Using a standardised preparation method, 15 g of roasted coffee beans was ground for 30 s in a universal grinder. Raw coffee beans were subjected first to three freeze/thaw cycles in liquid nitrogen followed by grinding for 60 s. A quantity of 3.75 g ground coffee powder was filter-brewed with hot tap water (Ø 85°C) to a total volume of 48 mL (coffee temperature: Ø 60°C). The coffee was cooled on ice for 1 h. Thereafter, the pH was adjusted with NaOH (2 N) from pH 5.6 (Ø) to pH 7.4. The volume of the coffee brew was filled up to 50 mL with hot water and filter-sterilised (0.22 μm). In all experiments, the coffee extract was freshly prepared and used exactly 1 h after brewing. The coffee concentration used in the experiments refers to the dry weight, which was analysed in at least three independent preparations per coffee lot.

Cell culture

Cells of various passages were cultured at 37°C in a humidified atmosphere with 5% CO2. Each cell culture medium was supplemented with penicillin (100 U/mL) and streptomycin (100 μg/mL). Rat macrophages (NR8383) were maintained in HAM’s F12 medium supplemented with 15% (v/v) heat-inactivated FCS. Human epithelial colorectal adenocarcinoma cells (Caco-2) were grown in MEM supplemented with 20% (v/v) FCS and nonessential amino acid solution (0.1 mM). NR8383 macrophages were harvested by scratching; Caco-2 cells were detached by trypsin (0.25%)/EDTA (0.02%).

Tissue culture of intact human gut tissue

The study was approved by the ethics committee of the University Medical Centre in Erlangen (Germany). Patients gave their informed consent to this local study. Mucosal gut specimens (3–15 mg) were taken from patients during gastrointestinal endoscopies because of various gastrointestinal complaints and to check up for manifestations of inflammation, neoplasm or food allergy. Samples were taken from two defined locations in the lower gastrointestinal tract, the terminal ileum of the small intestine and the ascending colon of the large intestine (Ø 6–7 biopsies per location per patient). In all patients, macroscopic as well as microscopic investigation of the biopsies did not show any chronic inflammatory gut diseases such as Crohn’s disease. The human gut tissue samples were kept in PBS transport media at 37°C and bubbled constantly with room air providing an adequate oxygen supply at a pO2 of 85–95 mm Hg (mucosa oxygenation) to avoid ischaemia (Raithel et al. 2006). The wet weight was determined prior to stimulation.

Analysis of Nrf2 translocation in cell lines—in vitro stimulation and nuclear protein extraction

Cells were stimulated with MRMs, ribose solution, coffee extracts (roasted and raw) and H2O2. In some experiments, cells were co-treated with catalase (150 U/mL) or heat-inactivated (5 min; 95°C) catalase (150 U/mL), which was added to the cells 10 min prior to stimulation. As negative control, cells were incubated with water/PBS instead of the stimulant. More specifically, NR8383 macrophages (3 × 106 cells) were grown for 4 days followed by stimulation in serum-free media for 2 h in short-term experiments and up to 24 h in the long-term experiments. To change the incubation medium, floating cells were collected by centrifugation (1,500 rpm, 2 min) and both floating and adherent cells were washed separately with serum-free media and unified thereafter. Caco-2 cells (1 × 106 cells) were grown for 5 d. Prior to stimulation, the medium was removed and adherent cells were washed with PBS and thereafter stimulated for 2 h in PBS.

The cellular nuclear extracts were prepared on ice directly after stimulation unless stated otherwise, according to Andrews and Faller (1991) with slight modifications. Briefly, adherent cells were detached by scraping. Both floating cells (macrophages) and adherent cells were collected by centrifugation (1,500 rpm, 4 min, 4°C). Floating and adherent cells were combined and washed three times with ice-cold PBS. Next, the cell pellet was re-suspended in 1 mL of ice-cold hypotonic buffer Acell, which was freshly supplemented with PIS (1%), PMSF (2 mM) and DTT (0.5 mM). After incubation for 15 min on ice, 65 μL of 10% NP-40 was added and cells were mechanically lysed by vortexing for 15 s. Cell nuclei were collected by centrifugation and washed with buffer Acell to completely remove cytoplasmic proteins. Finally, nuclear proteins were extracted with 52 μL of ice-cold high salt extraction buffer B, which was freshly supplemented with PIS (1%), PMSF (2 mM) and DTT (0.5 mM). The suspension was vortexed every 20 min. After 1 h, the supernatant containing the nuclear proteins was collected by centrifugation. Protein concentration of the nuclear extract was measured directly according to the DC-Protein protocol using the manufacturer’s recommendations (Bio-Rad). BSA, which was dissolved in supplemented buffer B, was used as standard. Nuclear cell extracts were stored at −80°C until use for Western blotting.

Analysis of Nrf2 translocation in intact human gut tissue—ex vivo stimulation and nuclear protein extraction

Human gut tissue samples were incubated and stimulated using a mucosa oxygenation method (Raithel et al. 2006). Briefly, two mucosal gut specimens (terminal ileum/ascending colon) were exposed to MRMh or roasted coffee extract for 2 h in a supplemented incubation medium (PBS, 3% albumin, 2.4% HEPES buffer and 0.0025% endo-paractol as defoaming agent). The negative control specimens were exposed to PBS or water. In some experiments, catalase (150 U/mL) was added to the human gut tissue samples 10 min prior to stimulation. During stimulation at 37°C, human gut tissue samples were continuously oxygenated (mucosa oxygenation) to prevent ischaemia and to support the diffusion of the medium dissolved stimuli into the tissue (Raithel et al. 2006).

After stimulation, the nuclear proteins were extracted according to a modified method of Thiele et al. (1999). In detail, human gut tissue samples were washed in a 50-mL tube with ice-cold PBS for 2 min while shaking at 400 rpm. To lyse the cells, the human gut tissue samples were re-suspended in ice-cold hypotonic buffer Atissue (10 mM HEPES, 1.5 mM MgCl2, 10 mM KCl) which was freshly supplemented with DTT (0.5 mM), PMSF (0.2 mM), PIS (1%) and NP-40 (0.56%), followed by three freeze–thaw cycles in liquid nitrogen. Afterwards, the specimens were mechanically disrupted with an Ultra Turrax homogeniser for 1 min on ice and allowed to stand for 30 min before the cell nuclei were collected by centrifugation. The subsequent nuclear protein extraction was carried out as described above using 25 μL instead of 52 μL of the high salt extraction buffer B including supplements.

Nrf2 analysis by Western blotting

After in vitro or ex vivo stimulation, the nuclear Nrf2 content was determined in the isolated nuclear protein extracts. The nuclear Nrf2 was detected immunochemically via Western blotting according to a modified method described by Muscat et al. (2007) for Western blot analysis of NF-κB. Briefly, nuclear proteins were denatured at 95°C for 7 min. For normalisation, an aliquot of the nuclear extract was applied to reach a constant protein load (10 μg for cell experiments; 5 μg for tissue experiments) per lane. In some experiments, equal protein load was controlled by analysing in parallel nucleoplasmic β-actin by Western blotting (RSD < 16%). The proteins were separated by sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE, 10% SDS-polyacrylamide gel) and transferred to a nitrocellulose membrane by Western blotting. After blocking, the membranes were incubated with the primary rabbit anti-Nrf2 antibody (NR8383 macrophages: diluted 1:2,000 in blocking buffer; Caco-2 cells and human gut tissue: diluted 1:1,000 in blocking buffer) overnight at 4°C. Thereafter, the membranes were exposed to the HRP-conjugated second anti-rabbit antibody (1:2,000 diluted in blocking buffer) for 1 h at room temperature. After a washing step, the membranes were incubated with an ECL reagent to visualise the protein bands. The data was analysed densitometrically using a VersaDoc™ Imaging System. The apparent size of Nrf2 was ~43 kDa. Nrf2 activation in stimulated cells was expressed as n-fold increase of nuclear Nrf2 amounts compared to untreated control cells exposed only to the solvent PBS/water.

MTT assay

The viability of NR8383 macrophages stimulated with MRMh was investigated according to a modified method of Mosmann (1983). The method is based on the ability of active cells to metabolise the MTT into dark blue insoluble formazan. Specifically, NR8383 macrophages (1 × 106 cells) were allowed to settle overnight at 37°C in a 12-well plate. Next, cells were stimulated with MRMh (10–100 mM) for 24 h at 37°C in supplement-free cell culture medium. In the case of catalase co-treatment, catalase (150 U/mL) was added to the cells 10 min prior to stimulation. Negative control cells were incubated with PBS without stimulant. As positive control, cells were treated with 10% DMSO instead. After stimulation, cells were incubated with MTT (1 mg/mL in PBS) for 3 h at 37°C. The developing dark blue formazan crystals were dissolved in isopropanol/1 N HCl (25:1) solution under agitation. Finally, the absorbance was read after 10 min at 595 nm. The absorbance of control cells was considered 100% viable cells. The cell viability of stimulated cells was expressed as percent of control. For media replacement during the assay, the cell suspension was centrifuged at 1,500 rpm for 2 min at room temperature.

Determination of cell viability by the trypan blue dye exclusion test

The cell viability of NR8383 macrophages exposed to coffee extract was further determined by the trypan blue dye exclusion test. Depending on cellular membrane integrity, which is one main characteristic of vital cells, the anionic diazo dye trypan blue is absorbed into the cell and appears blue. Briefly, NR8383 cells (1 × 106 cells) were grown for 4d in the incubator before being exposed to coffee extract (1–4 mg/mL) in PBS for up to 6 h. After stimulation, cells were treated with trypan blue dye solution (5 mg/mL PBS) and viable as well as dead cells were counted under the microscope via a counting chamber. Cell viability was calculated as the ratio between living cells to total cell number and expressed as a percentage of the values obtained for control cells, which was set as 100%.

Statistical analysis

Statistical significance of the data was calculated using Graphpad prism 5. The results were expressed as the mean ± SD from n independent experiments for n > 2 and mean ± range for n = 2. The significance of differences was calculated by a Student’s t test and levels of confidence were defined as *p < 0.05, **p < 0.01, and ***p < 0.001.

Results

Induction of nuclear translocation of Nrf2 by MRPs in different cell types and intact human gut tissue

The purpose of the study was to investigate the activity and mechanisms of MRPs to induce nuclear translocation of Nrf2 in different cell lines and in intact human gut tissue. For this purpose, MRPs were generated by heating ribose and lysine for 30 min at 120°C. Similar mixtures have been previously applied to determine physiological effects of MRPs (Muscat et al. 2007; Hegele et al. 2009). In order to investigate whether Maillard reaction products activate Nrf2 in vitro, NR8383 macrophages were incubated for 2 h (short-term incubation) with a heated MRM. The activation of Nrf2 was defined as the translocation of cytoplasmic Nrf2 into the nucleus. Nuclear Nrf2 was detected by Western blotting. MRMh significantly induced Nrf2 activation in macrophages in a concentration-dependent manner reaching a fivefold up-regulation at 100 mM MRMh (Fig. 1).

Fig. 1
figure 1

Nrf2 activation by MRMh in NR8383 macrophages, Caco-2 cells and intact human gut tissue samples during short-term incubation. Cells and tissue were incubated with MRMh (10–100 mM) for 2 h. a The intensity of nuclear Nrf2 is expressed as n-fold increase compared to PBS-treated control cells. Data are mean ± SD for n > 2; mean ± range for n = 2 (#) (macrophages: n = 2–5; Caco-2: n ≥ 5; gut tissue: n ≥ 5). *p < 0.05, **p < 0.01, ***p < 0.001. (n.d. = not determined), b Representative Western blot of Nrf2 in NR8383 macrophages, Caco-2 cells and human gut tissue

Full size image

For long-term incubation experiments, macrophages were incubated with different concentrations of MRMh for up to 24 h (Fig. 2a). The nuclear Nrf2 level increased consistently in a time- and concentration-dependent manner. No influence on Nrf2 translocation was measured over an incubation period of 24 h in the case of 10 mM MRMh. However, exposure of macrophages to ≥25 mM MRMh significantly enhanced Nrf2 translocation, reaching 50-fold higher Nrf2 levels after stimulation for 24 h with 50 mM MRMh compared to PBS-treated control cells. It was confirmed by the MTT assay that incubation with ≤25 mM MRMh for up to 24 h did not exert cytotoxic effects. Incubation with 50 mM Mrmh for 24 h resulted in a cell viability of 60 ± 21%. As a positive control, 50 μM tert-butylhydroquinone was applied under the same conditions, leading to a 1.9-fold increased nuclear Nrf2 content compared to the control. tert-Butylhydroquinone is an established inducer of nuclear Nrf2 translocation (Cheung et al. 2011).

Fig. 2
figure 2

Nrf2 activation by a a heated Maillard reaction mixture (MRMh) and c an unheated Maillard reaction mixture (MRMu) in NR8383 macrophages during long-term incubation. Cells were stimulated with MRMh (10–50 mM) or MRMu (25–100 mM) for 2–24 h. The intensity of Nrf2 is expressed as n-fold increase compared to PBS-treated control cells. Data are mean ± SD for n > 2; mean ± range for n = 2 (#) (MRMh: n = 3–5; MRMu: n = 2–3). *p < 0.05, **p < 0.01, ***p < 0.001 (b/d) Representative Western blot of Nrf2 in NR8383 macrophages incubated with b MRMh and d MRMu

Full size image

To study how MRPs induce the nuclear translocation of Nrf2, it was investigated whether long-term activation of Nrf2 depends on the continuous presence of the MRP stimulant. Thus, macrophages were incubated for 2 h with 25 mM MRMh. On removal of the MRMh after 2 h, the cells were incubated for another 22 h and nuclear Nrf2 concentration was determined after a total incubation period of 24 h. Nuclear Nrf2 levels continued to increase even in the absence of the stimulus to about eightfold higher levels compared to short-term incubation alone (Fig. 3). Cell viability was not reduced after incubation with MRMh (25 mM) for 24 h.

Fig. 3
figure 3

Nrf2 activation by MRMh in NR8383 macrophages. Cells were stimulated with the heated MRM (25 mM) for 2–24 h. Nuclear Nrf2 levels were analysed directly after stimulation or after a following post-incubation at 37°C without stimulant. a The intensity of Nrf2 is expressed as n-fold increase compared to PBS-treated control cells. Data are mean ± SD for n > 2; mean ± range for n = 2 (#) (n = 2–5). *p < 0.05, **p < 0.01, ***p < 0.001, b Representative Western blot of Nrf2 in NR8383 macrophages

Full size image

To verify whether nuclear translocation of Nrf2 was indeed induced by MRPs and not by a high concentration of amino acids and sugars, an unheated mixture of ribose and lysine (MRMu) was tested in the same way as MRMh. Incubation with a 25 mM solution of ribose/lysine (MRMu) for up to 24 h and with a 50 mM solution for up to 6 h did not lead to a detectable increase in nuclear Nrf2 signal intensity. Only after exposure to 50 mM MRMu for 12 and 24 h, the nuclear Nrf2 concentration was increased 8- and 18-fold, respectively, compared to control, which was, however, much lower than the induction by the heated mixture (MRMh) (Fig. 2c).

The role of caramelisation products in Nrf2 activation was tested. As caramelisation products are formed by sugar degradation in the absence of amines from heat-treated ribose alone, macrophages were exposed to a heated ribose solution. Even when concentrations up to 100 mM were applied, no induction of nuclear translocation of Nrf2 could be detected (data not shown). Thus, it can be concluded that the observed effect was indeed induced by MRPs and not by caramelisation products.

To investigate if MRP-induced nuclear translocation of Nrf2 is unique in macrophages, intestinal Caco-2 cells were also short-term stimulated with MRMh containing MRPs. In all tested concentrations, MRMh led to a significant increase of nuclear Nrf2 concentration (Fig. 1). However, the induction was generally lower than the effect observed in macrophages (twofold vs. fivefold compared to control).

Macrophages and epithelial Caco-2 cells represent relevant cell types of the intestine. However, it is possible that these cell lines show a different response than primary cells. Additionally, the interaction of different cell types in an intact tissue may have strong influence on cellular reactions. Therefore, ex vivo incubation of intact human gut tissue was introduced to assess these factors (Raithel et al. 2003). In the present study, intact human mucosal gut tissue samples obtained from the lower gastrointestinal tract of six patients were used. Depending on the number of biopsies available from each patient, each experiment was performed in five or six replicates. In all patients, macroscopic and microscopic investigation of the biopsies did not show any acute or chronic inflammatory gut diseases. As cell viability decreased by approximately 50% after mucosa oxygenation ex vivo for 6 h, the tissue samples were only subjected to short-term incubation for 2 h (Sauer et al. 2011). After an incubation time of 2 h, cell viability was not reduced. The tissue samples were stimulated ex vivo with different concentrations of MRMh and the nuclear translocation of Nrf2 was analysed. In the gut tissue samples, the amount of nuclear Nrf2 increased significantly after treatment with 50 mM and 100 mM MRPs (1.9-fold and 2.1-fold increase, respectively, compared to control) (Fig. 1).

Role of extracellular H2O2 in the MRP-induced activation of Nrf2

To gain some insight into the mechanisms of how MRPs induce nuclear translocation of Nrf2, it was investigated if the observed Nrf2 activation was mediated by ROS, generated by MRPs (Mueller et al. 2011; Hegele et al. 2009). For this purpose, macrophages were simultaneously incubated with MRMh and catalase or heat-inactivated catalase. Nrf2 translocation was subsequently analysed at several time points. As illustrated in Fig. 4a, co-treatment with catalase, but not with heat-inactivated catalase, significantly blocked the MRMh-induced translocation of Nrf2 in NR8383 macrophages. Cell viability was not reduced after incubation with MRMh (25 mM) ± catalase for 24 h. In the presence of NR8383 macrophages, the applied MRMh produced between 100 and 450 μM extracellular H2O2, depending on the MRMh concentration and incubation time. Co-incubation with catalase completely removed extracellular H2O2 (see (Sauer et al. 2011) for details).

Fig. 4
figure 4

Nrf2 regulation by MRMh ± (heat-inactivated) catalase in a NR8383 macrophages, c Caco-2 cells and e human gut tissue. Cells and tissue samples were stimulated with MRMh (25 mM/100 mM) for 2–24 h. In some experiments, cells were co-treated with catalase (150 U/mL) or heat-inactivated catalase (150 U/mL). The intensity of Nrf2 is expressed as n-fold increase compared to PBS-treated control cells. Data are mean ± SD for n > 2; mean ± range for n = 2 (#) (NR8383 macrophages n = 3–5; Caco-2: n = 2–6; human gut tissue: n = 3–5). *p < 0.05, **p < 0.01, ***p < 0.001, Representative Western blot of Nrf2 in b NR8383 macrophages, d Caco-2 cells and f human gut tissue

Full size image

Similarly, supplementation of catalase significantly reduced MRMh-induced Nrf2 activation in Caco-2 cells (Fig. 4c) and also showed an inhibitory trend in human gut tissue samples ex vivo (Fig. 4e). Extracellular H2O2 ranged between 50 and 100 μM, when human gut tissue was incubated with MRMh depending on its concentration and incubation time (Sauer et al. 2011). The data indicate a similar mechanism of Nrf2 activation involving H2O2 across the tested cell types and tissue. The smaller effect of unheated MRM (Mrmu) on Nrf2 translocation after long-term incubation was not blocked by co-treatment with catalase (data not shown) indicating that other mechanisms were involved.

Since extracellular H2O2 played a crucial role in the MRMh-induced Nrf2 activation, the impact of a pure H2O2 solution on Nrf2 activation was likewise investigated in macrophages. The nuclear Nrf2 level of macrophages did not significantly respond to either 100 or 500 μM H2O2 during incubation for 24 h (data not shown), suggesting that permanent generation of H2O2 from MRPs rather than the addition of a single bolus of H2O2 was important to induce nuclear Nrf2 translocation.

Induction of nuclear translocation of Nrf2 by coffee in different cell types and intact human gut tissue

Coffee is a food product rich in MRPs, which are formed during roasting. Furthermore, Nrf2 induction by coffee has been observed in AREc32 cells, in EpRE-LUC transfected HepG2 cells and in human colon carcinoma cells (Boettler et al. 2010; Paur et al. 2010; Cavin et al. 2008). Therefore, this study investigated the effect of coffee on nuclear translocation of Nrf2 in different cell types present in intestinal tissue as well as on intact human gut tissue and compared the findings to the effect of MRPs.

Short-term stimulation with coffee extract for 2 h showed a non-significant trend in activating Nrf2 in NR8383 macrophages, but not in Caco-2 cells and human gut tissue samples (Fig. 5). The maximal concentration of the tested coffee extracts (4 mg/mL) corresponds to a ~1:3 diluted (German) coffee beverage.

Fig. 5
figure 5

Nrf2 activation by short-term exposure of macrophages, Caco-2 cells and intact human gut tissue to coffee extract. Cells and tissue samples were incubated with coffee extract (1–4 mg/mL) for 2 h. The intensity of nuclear Nrf2 is expressed as n-fold increase compared to water-treated control cells. Data are mean ± SD for n > 2; mean ± range for n = 2 (#) (NR8383 macrophages: n = 4–8; Caco-2: n = 3; human gut tissue: n = 2–5)

Full size image

Additionally, Nrf2 translocation was monitored during prolonged stimulation of macrophages with coffee extract for up to 24 h. Indeed, coffee extract significantly activated Nrf2 in macrophages during prolonged stimulation (Fig. 6). The activation of Nrf2 increased gradually with sample concentration and incubation time. Whereas 1 mg/mL of coffee extract slightly elevated nuclear Nrf2 amounts twofold compared to water-treated control cells after incubation for 24 h, 4 mg/mL of coffee extract up-regulated nuclear Nrf2 levels nearly 20-fold. On the contrary, stimulation with raw coffee extract (4 mg/mL), which does not contain roasting products, did provoke only a minor Nrf2 activation in macrophages. The trypan blue dye exclusion test determined a viability exceeding 90% for all cells incubated for up to 6 h with coffee extracts (1–4 mg/mL).

Fig. 6
figure 6

Nrf2 activation by roasted and raw coffee extract in NR8383 macrophages. Cells were stimulated with coffee extract (1–4 mg/mL) or raw coffee extract (4 mg/mL) for 2–24 h. a The intensity of Nrf2 is expressed as n-fold increase compared to water-treated control cells. Data are mean ± SD (n = 3–8). *p < 0.05, **p < 0.01, ***p < 0.001 b Representative Western blot of Nrf2 in NR8383 macrophages

Full size image

The effect of catalase co-treatment on nuclear Nrf2 translocation in macrophages was monitored during stimulation with coffee extract. There was a slight trend towards reduced nuclear Nrf2 levels in the presence of catalase, which was, however, not significant (Fig. 7). H2O2 concentration in the incubation solutions of macrophages with coffee ranged between 50 and 230 μM. No H2O2 could be detected after co-incubation with catalase (Sauer et al. 2011).

Fig. 7
figure 7

Nrf2 regulation by coffee extract ± catalase in NR8383 macrophages. Cells were stimulated with coffee extract (4 mg/mL) for 2–24 h. In some experiments, cells were co-treated with catalase (150 U/mL). a The intensity of Nrf2 is expressed as n-fold increase compared to water-treated control cells. Data are mean ± SD (n = 5–6) b Representative Western blot of Nrf2 in NR8383 macrophages

Full size image

Discussion

The goal of the present study was to investigate the influence of MRPs on the nuclear translocation of the redox-sensitive Nrf2 in different cell types present in intestinal tissue as well as in intact human gut tissue. Nrf2 is an inducible transcription factor, which is ubiquitously expressed in tissue (Moi et al. 1994). Activation of Nrf2 triggers a pathway regulating the expression of detoxifying and antioxidative active proteins such as glutathione S-transferase, heme oxygenase 1 and NAD(P)H: quinone oxidoreductase 1 (Surh et al. 2008; Itoh et al. 1997).

Activation of Nrf2 is classified as a therapeutic target in various diseases such as Parkinson’s disease (Cuadrado et al. 2009), cardiovascular diseases (Li et al. 2009) and type 2 diabetes mellitus (Cheng et al. 2011). Moreover, the intake of dietary Nrf2 activators is of crucial relevance in intestinal dysfunction. Induction of Nrf2 is considered to reduce the risk of colorectal cancer (Saw and Kong 2011; Pool-Zobel et al. 2005) and has a beneficial impact on gut barrier dysfunction (Jin et al. 2008), intestinal mucosal injury (Jin et al. 2009) and intestinal inflammation (Jin et al. 2008) such as colitis (Khor et al. 2006).

During short-term stimulation (2 h), a heated MRM significantly activated Nrf2 in Caco-2 cells and particularly in macrophages in a concentration-dependent manner. Even intact human gut mucosal samples showed a significant nuclear translocation of Nrf2 within 2 h, indicating the biological relevance of this finding. Interestingly, short-term incubation of macrophages with MRMh followed by a stimulant-free post-incubation further enhanced Nrf2 activation compared to short-term incubation alone, suggesting that MRMh exhibits not only direct effects on different cell types, but possibly also long-lasting trophic and protective effects on intestinal tissue. To date, two Nrf2 activation mechanisms are proposed for the inhibition of Keap1-dependent Nrf2 degradation and subsequent translocation into the nucleus. Firstly, ROS and electrophiles, such as H2O2 and sulphoraphane, might stabilise cytoplasmic Nrf2 by modifying specific cysteine thiols of Keap1. Secondly, phosphorylation of serine and threonine residues of Nrf2 via protein kinases such as protein kinase C (PKC) and c-Jun N-terminal kinases (JNK) is also suggested to cause Nrf2 release from Keap1. Continuing activation of Nrf2 after removal of the stimulant points to Nrf2 activation via a signal transduction chain instead of direct interaction with the Nrf2-Keap1 complex (Surh et al. 2008). Alternatively, it cannot be completely ruled out that irreversible Keap1 cysteine modification is persistent over an extended period of 24 h and thus causes further up-regulation in the absence of the stimulant.

During long-term incubation of macrophages with MRMh, the Nrf2 response increased over time up to 24 h. Very recently it was shown that a heated glucose-BSA solution (AGE-BSA) increased nuclear Nrf2 amounts in bovine aortic endothelial cells. Moreover, an enhanced protein expression of Nrf2-regulated proteins such as heme oxygenase 1 and NAD(P)H: quinone oxidoreductase 1 was detected (He et al. 2011). In vitro screening raised the question of how Nrf2 responded in human intestinal tissue consisting of a wide range of cell types. The human intestine represents a primary target for food-induced inter- and intracellular reactions independent of the resorption rate and systemic metabolism. The mucosa layer of the gut consists of a variety of cell types including immune cells such as macrophages, which were found to show the highest Nrf2 response. Therefore, Nrf2 activation by MRPs was studied ex vivo in the mucosa of the gut. Indeed, MRMh significantly activated Nrf2 not only in various cell lines, but also in human gut tissue ex vivo. It can be assumed that MRP-induced Nrf2 activation contributes to up-regulation of NAD(P)H: quinone oxidoreductase, thioredoxin and γ-glutamylcysteine synthetase caused by coffee roasting products (Paur et al. 2010).

MRMh-induced Nrf2 activation was reduced or fully suppressed in macrophages and Caco-2 cells after co-treatment with catalase. This observation indicates that extracellular H2O2 is involved in MRMh-induced Nrf2 activation. Nrf2 activation by H2O2 has been previously reported in rat pulmonary microvascular endothelial cells (Ning et al. 2010). It is well known that MRPs generate H2O2 in various cell-free systems as well as in the presence of cells (Mueller et al. 2011; Hegele et al. 2009). Therefore, it can be assumed that H2O2 formation is not an artificial effect of cell culture in vitro, but occurs also in various settings such as food preparation or in the gastrointestinal tract. Also, it was previously shown that MRPs trigger nuclear NF-κB translocation via a H2O2-mediated mechanism (Muscat et al. 2007). Both NF-κB and Nrf2 are redox-sensitive transcription factors (Bellezza et al. 2010). Besides MRP-dependent H2O2 generation, NAD(P)H oxidase, a membrane-bound enzyme, can also liberate ROS into the extracellular space (Petry et al. 2010). It is hypothesised that extracellular H2O2, irrespective of its origin, was not completely detoxified by the antioxidant system of the cell including glutathione peroxidase, peroxiredoxins or catalase and numerous non-enzymatic antioxidants. Instead, extracellular H2O2 might penetrate the cell. The increasing intracellular oxidative stress level may then trigger cellular signalling such as the Nrf2 pathway. Interestingly, incubation of macrophages with a single bolus of H2O2 did not activate Nrf2. This may be due to complete cellular detoxification of a single dose of H2O2, whereas MRPs permanently generate baseline levels of H2O2. Moreover, Nrf2 activation by MRMh could also be enhanced by an interaction of H2O2 with MRPs. A similar effect was described for the DNA breaking ability of Maillard products (Iida et al. 2002). In the present study, heated ribose solution did not have any impact on Nrf2 translocation excluding any involvement of caramelisation products, which are generated in addition to MRPs during heat treatment of sugar-rich foods.

Since MRPs generated by heating an amino acid with a sugar induced nuclear Nrf2 translocation, the effect of coffee, a beverage rich in MRPs, was investigated in the same test system. During coffee roasting, amine and sugar components of the green coffee bean react extensively to give the so-called melanoidin fraction, which comprises about 25% of the dry matter of the coffee brew (Fogliano and Morales 2011). Thus, it can be assumed that MRPs in roasted coffee induce nuclear Nrf2 translocation in a similar way. Furthermore, other coffee ingredients or the more complex composition of coffee melanoidins may modulate MRP-induced Nrf2 activation likewise. After short-term incubation, coffee extracts did not influence the nuclear Nrf2 concentration either in Caco-2 cells or in human gut tissue, but showed a trend of increased Nrf2 activation in macrophages. Upon long-term incubation, however, coffee extracts significantly activated Nrf2 in macrophages in a concentration and time-dependent manner. This result was in alignment with studies on the effect of coffee on human colon carcinoma cells (Boettler et al. 2010), HepG2 cells transfected with an EpRE-LUC construct (Paur et al. 2010) and ARE-luciferase reporter cells (Cavin et al. 2008) as well as Nrf2-luciferase reporter mice (Paur et al. 2010). The incubation periods of these in vitro assays were 3, 17 and 24 h, respectively. Interestingly, the effect in human colon carcinoma cells did not show a clear relation to the coffee concentration; in contrast, the effect of 1 pg/mL coffee extract was even more pronounced than the effect of 1 or 500 μg/mL (Boettler et al. 2010).

On the contrary to roasted coffee extract, raw coffee extract did not up-regulate Nrf2 in macrophages. It was shown before that the level of Nrf2 activation was related to the degree of coffee roasting (Paur et al. 2010). Thus, the roasting products were considered to be the active compounds in coffee triggering Nrf2. Co-treatment with catalase showed only a slight trend to block the coffee-induced Nrf2 translocation indicating that coffee-induced Nrf2 activation cannot be fully explained by MRP-generated extracellular H2O2. These findings are consistent with the previous study on coffee-induced Nrf2 activation in human colon cells (Boettler et al. 2010). Besides the roasting products, several coffee compounds such as the diterpenes cafestol and kahweol, N-methylpyridinium, hydroxyhydroquinone and chlorogenic acid were identified as potential Nrf2 activators (Boettler et al. 2010; Paur et al. 2010; Higgins et al. 2008). However, none of these components could fully explain coffee-induced Nrf2 activation.

In summary, the effect of coffee extract and Maillard reaction products to activate Nrf2 in various cell lines and human gut tissue was studied. It was demonstrated that cellular effects of coffee and Maillard reaction products, which had been established in different immortalised cell lines, were also evident when intact human gut tissue was stimulated ex vivo. Moreover, this study provides growing evidence that the translocation of Nrf2 is triggered by the roasting products of the coffee extract, which are structurally similar to Maillard reaction products. The mechanism of Nrf2 activation by MRPs was dependent on their activity to generate H2O2. In contrast, coffee-induced Nrf2 translocation was only slightly reversed by catalase. Therefore, the activity of coffee can only be partially explained by MRP-induced mechanisms. The identification of MRPs as potent Nrf2 activators might explain, besides their direct antioxidative activity, how MRPs decrease the susceptibility towards oxidative stress in vitro and in vivo.