Maillard reaction products modulate gut microbiota composition in adolescents

2.1 Human experiment (expt. 1)

The selection of the subjects, the diets composition and the study design have been described elsewhere 15. Briefly, 20 male adolescents (12.4 ± 0.34, mean ± SE year of age) participated in a 2-week randomized two-period crossover trial, in which two different diets were consumed with a 40-day washout period in between. Two 7-day menus which were formulated to be similar in energy and nutrients contents, and containing the same servings per day of the different food groups were designed: a white diet (WD), free, as much as possible, of foods in which the Maillard reaction develops during cooking practices or those usually containing MRP; and a brown diet (BD), rich in processed foods with an evident development of browning and thus rich in MRP. Lunch and dinner were prepared by a local catering firm and distributed daily to the participants, and also to the researchers to enable the analysis of the MR markers. The lunch and dinner 7-day menus, as well as the composition of breakfasts and afternoon snacks were widely described in Seiquer et al. 15. Each 7-day menu was repeated once during each of the 14-day experimental periods.

The food composition of the diets was converted into energy and nutrient values using the Spanish Food Composition Tables 16, under AYS44 Diet Analysis software supplied by ASDE, SA (Valencia, Spain). No significant differences were found in energy and nutrient composition between diets. The overall daily content (in fresh matter) was as follows: energy 2530 kcal, fat 107.5 g, carbohydrate 316.9 g, protein 90.1 g, fiber 25.1 g, cholesterol 311.4 mg, sodium 1865 mg, potassium 3826 mg, calcium 1049 mg, phosphorus 1595 mg, magnesium 372 mg, iron 17.5 mg, zinc 8.9 mg. The total polyphenols content analysis indicated no significant differences between diets (2.20 ± 0.04 mg/g diet on average) 17. The analysis of Maillard reaction markers in the diets was performed as previously described 6, 18. Briefly, furosine (e-N-(2-furoyl-methyl)-L-lysine) was quantified in acid hydrolysis by isocratic ion-pair reversed phase liquid chromatography, using 5 mM sodium heptane sulphonate including 20% of acetonitrile and 0.2% of formic acid as mobile phase, a C18 column at 32°C and UV detection at 280 nm. Hydroxymethylfurfural (MHF) was measured in the aqueous extract of the samples after clarification with Carrez I (15% w/w potassium ferrocianide) and Carrez II (30% w/w zinc acetate) solutions by isocratic ion-pair reversed phase liquid chromatography, using a mobile phase of water/acetonitrile (95:5) and the same column, temperature and UV detection previously mentioned. Lastly, carboxymethyl-lysine (CML) was determined in the acid hydrolysis previously reduced by reversed phase liquid chromatography with precolumn derivatisation with o-phthaldialdehyde (OPA). An ODS-2 analytical column at 32°C was used. Elution buffers were: (A) 15 mM sodium phosphate buffer pH 7.2-acetonitrile (83:17, v/v) and (B) acetonitrile. The binary gradient was linear from 0% to 30%B in 10 min and stand for 3 minutes at 60%B. Then gradient was set back to 0%B within 5 minutes. The OPA-derivatives were detected fluorimetrically at 340 nm excitation and 455 nm emission wavelengths. Furosine, HMF and CML were quantified by the external standard method, building the corresponding calibration curve from stock solutions. Data from these analyses showed a greater development of the Maillard reaction in the BD than in the WD, according to significantly higher values of HMF and CML (HMF 0.94 ± 0.01 and 3.87 ± 0.03 mg/kg, CML 6.62 ± 0.25 and 15.72 ± 0.43 mg/100 g of protein in the WD and BD, respectively). No differences were found in furosine content between diets (6.99 ± 0.45 and 6.37 ± 0.15 mg/100 g in the WD and the BD, respectively), supporting that foods usually contain certain amounts of early Maillard reaction compounds even when non-severe thermal treatments are applied. Compliance with dietary treatments was assessed by daily records sheets, which were used to calculate daily food intake and in turn, daily intake of MR markers (Amadori compounds, HMF and CML). It was assumed that furosine represent 36% of Amadori compounds 19. Fecal samples and fasting venous blood were collected from each volunteer after each dietary treatment. Aliquots of fecal samples were lyophilized and stored at −20°C for future analysis. EDTA-containing vacutainers were used for the collection of whole-blood samples; plasma was recovered by centrifugation at 1700 × g for 15 min (4°C) and analyzed by using a haematological autoanalyzer (Hitachi 917 autoanalyzer, Hitachi, Boehringer Mannheim, Mannheim, Germany) for the determination of biochemical parameters (glucose, total cholesterol, triglycerides, HDL/LDL ratio, transaminase enzymes as GOT, GPT and GGT, and total bilirubin). Body weight and height were recorded at the end of each period, and the body mass index (BMI) was calculated.

This human trial was approved by the ethics committee of the San Cecilio University, Hospital of Granada and was performed in accordance with the Helsinki Declaration of 2002, as revised in 2004. The informed consent was obtained from the parents of all the adolescents participating in the study.

2.2 Rat experiment (expt. 2)

The diets and experimental design were as described in a previous report 20. The AIN-93G purified diet for laboratory rodents (Dyets Inc, Bethlehem, PA) was used as the control diet (CD). An MRP model system from glucose-lysine mixture heated at 150°C for 90 min (as described in Delgado-Andrade et al. 21) was added to the AIN-93G diet at a final concentration of 3% to obtain the glucose-lysine diet (GLD). The individual analysis of the diets revealed no significant differences of the overall nutrient composition between them. Nutrient content (g/kg) was as follows (mean ± SD): moisture 81.4 ± 0.8, protein 176.6 ± 3.1, fat 78.1 ± 0.9, calcium 4.79 ± 0.09, phosphorus 3.28 ± 0.03. Maillard reaction markers were analyzed according to the methods formerly described. Higher development of the MR was found in GLD compared with CD, as shown by the increased values of the MRP markers analyzed (furosine 28.8 ± 0.5 and 1787.08 ± 7.31 mg/kg, HMF 0.44 ± 0.06 and 5.15 ± 0.08 mg/kg, CML 2.20 ± 0.07 and 12.46 ± 0.72 mg/kg in the CD and the GLD, respectively).

Sixteen male weanling Wistar rats weighing 40.77 ± 0.29 g (mean ± SE) were randomly distributed into two groups (n = 8) and each group was assigned to one of the dietary treatments. The animals were individually housed in metabolic cages in an environmentally controlled room under standard conditions. The rats had ad libitum access to their diets and demineralised water (Milli-Q Ultrapure Water System, Millipore Corps., Bedford, MA, USA) and were fed the different diets for 87 days. Feed intake was monitored weekly during this period; on the basis of food intake and MRP dietary content, daily intake of MRP markers was calculated. On day 88, after an overnight fast, animals were anaesthetized with sodium pentobarbital (5 mg/100 g of body weight) (Abbott Laboratories, Granada, Spain) and terminal exsanguination was performed by cannulation of the carotid artery. Ceca were removed, immediately frozen at −20°C and then freeze-dried until analysis.

All management and experimental procedures carried out in this rat trial were in strict accordance with current European regulations (86/609 E.E.C.) regarding laboratory animals. The Bioethics Committee for Animal Experimentation at our institution (EEZ-CSIC) approved the study protocol.

2.3 q-PCR analysis of microbiota composition

Samples for q-PCR analysis were run in duplicate. Total DNA was isolated from freeze-dried fecal (children) or cecal (rat) samples (40 mg) using the QIAamp DNA stool kit (Qiagen, West Sussex, UK) by following manufacturer´s instructions. In order to increase its effectiveness, the lysis temperature was increased to 95°C and an additional step with lysozyme (10 mg/mL, 37°C, 30 min) incubation was added. Eluted DNA was treated with RNase and the DNA concentration assessed spectrophotometrically by using a NanoDrop ND-100 Spectrophotometer (NanoDrop Technologies, Wilmington, DE, USA). Purified DNA samples were stored at −20°C until use 22. Bacterial log₁₀ number of copies was determined in by using q-PCR. The 16S rRNA gene-targeted primers and PCR conditions used in this study were as in 23. The different microbial groups quantified included total bacteria, bacteroides, bactobacilli, bifidobacteria, Eubacterium rectale/Clostridium coccoides group, Clostridium leptum, enterobacteria and Escherichia/ Shigella group.

2.4 Statistical analysis

Individual humans (expt. 1) or rats (expt. 2) were considered the experimental unit. After confirming the normal distribution of data, results of MRP intake, microbiota counts, and the plasma biochemical and anthropometrics parameters in male adolescents after dietary treatments, were analyzed by one-way ANOVA followed by the test of Least Significant Difference (LSD) of Fisher to compare means that showed significant variation (p < 0.05). The relationship of MRP intake with microbial counts and biochemical and anthropometrics parameters in adolescents was evaluated by computing the relevant correlation coefficient (Pearson linear correlation) at the p < 0.05 confidence level. Analyses were performed using Stat Graphics Centurion XV, version 15.2.06 software (Stat Point Inc.).

3.1 Intake of MRP markers

The daily intake of MRP markers studied in the present assay (i.e. Amadori compounds, HMF and CML) is shown in Fig. 1, both in expt. 1 (humans, A) and in expt. 2 (rats, B). Balanced, varied diets adequate for their needs and adapted to dietary subjects´ habits were designed for the adolescents assay. Thus, the diets contained a variety of foods which were subsequently prepared in different ways with the aim of suppressing or enhancing the MR development (white or brown, WD or BD, respectively). As a consequence, a big range of MRP from different substrates could have been formed in the BD, which may thus contain polysaccharide-rich and protein-rich melanoidins.

No significant differences in total food consumption by the adolescents were observed between diets (mean ± SE: WD 1930 ± 50 g/day and BD 1802 ± 47 g/day, in fresh matter) but, as a consequence of the differences in dietary MRP contents, significant differences were found concerning the MRP intake (Fig. 1A). Amadori compounds intake in the human assay, although significantly different between diets, was quantitatively similar, supporting that any diet, even when non-severe culinary techniques are used, involves certain MRP intake, particularly of the early compounds 24. Consumption of the BD promoted mainly the intake of more advanced MRP, i.e., consumption of HMF and CML was five- and twofold higher, respectively, in BD than it was in the WD. It is known that drastic conditions of food processing, such as increasing temperature and cooking duration, favour the MR development and the presence of melanoidins 25. In fact, in addition to the already mentioned markers, the determination of the CIELAB colour parameters in the different diets 6 showed significant loss of luminosity (L*) and a progress of colour to the red zone (a*) in the BD, in accordance with the higher presence of MRP due to the more severe cooking treatment of meals. Therefore, consumption of the same food servings but prepared by different culinary techniques, may lead to different intakes of MRP, depending on the severity of the thermal processing 19.

In the rats assay, animals consuming the GLD also had higher MRP intake than those consuming the CD (Fig. 1B), without changes in their food consumption (mean ± SE: CD 15.0 ± 0.5 g/day and GLD 14.7 ± 0.7 g/day). Larger differences between groups were detected in rats than in humans, as intake of Amadori compounds, HMF and CML was approximately 60, 11 and 6 fold higher, respectively, in animals consuming the MRP-rich diet than in those fed on the control diet. Thus, the use of a model-system led to more drastic conditions in the rat assay, whereas diets in the human experiment were similar to those usually consumed by this population. The GL model-system has been widely used to study MRP implications for health, as lysine is the more sensitive amino acid to the MR and, frequently, the limiting amino acid in many food systems.

3.2 Microbiota composition of feces (adolescents) and cecal contents (rats)

A range of in vitro models, such as anaerobic batch culture fermenters or hybridation techniques, have been applied to examine the interaction between MRP and the different species of bacteria, but some of them offer conflicting results 2, 11. Previous studies have shown increases of total bacterial population after in vitro batch culture fermentation with MRP from glucose-lysine 11, melanoidins obtained from bread crust or coffee silver-skin 12, 26, or lactoglobulin glycated with galactooligosaccharides 23, thus stating the ability of microorganisms to use browning compounds as a source of nitrogen and carbon. However, consumption of whole grain breakfast cereals 27 or coffee 13 has failed to modify total fecal bacterial numbers in humans. Thus, although previous in vitro studies have concluded that human gut microbiota is affected by melanoidins, there is a need of intervention trials with human volunteers to confirm microbiological in vitro observations.

Data on fecal bacterial numbers after consumption of the different diets in experiments 1 and 2 are depicted in Table 1. No differences were found in total fecal bacterial counts after consumption of the different diets by the adolescents; contrary wise, a significant decrease was observed in the total bacterial number of copies in the cecal contents of rats fed on the GLD compared with those on the CD (Table 1). Results of the present assay showed that the intake of the measured MRP in humans was not correlated with total fecal bacteria, but negative correlations were found between MRP consumption by rats and total microorganisms (Tables 2 and 3, respectively). The differences in the effects on total microbial counts between both assays could be attributed to: i) the more drastic conditions of the rat assay (i.e. higher MRP intake), ii) significant compositional and metabolic differences in the melanoidins tested in both experiments, and/or iii) physiological and microbiota composition differences between rats and humans. It is well known that results from rat assays cannot be completely extrapolated to humans; while rodents are cecal fermenters, carbohydrate fermentation in humans primarily takes place in the proximal part of the colon, which leads to the depletion of these substrates in the distal colon 28. Also, selective effects have been shown for different melanoidins, supporting that both the starting material and the processing conditions influence the melanoidins potential to influence microbial growth 12. The decrease in total bacteria observed in rats after GLD consumption may be related with a possible toxic effect of GL mixtures, in line with the antimicrobial activity of model-system melanoidins initially reported by Einarsson et al. 29, and more recently by Rufián-Henares and Morales 30 for coffee and biscuits melanoidins of different molecular weighs.

Concerning the effects towards specific strains, significant decreases in lactobacilli, enterobacteria and the Escherichia / Shigella group were evidenced after consumption of the BD in young volunteers, whereas no changes were noticed for bifidobacteria, bacteriodes and clostridia. It must be underlined that, besides the MRP consumption, some other consequences of thermal food treatment, such as vitamin degradation or lipid oxidation, could be also implicated in the results observed. In the rat assay, a significant inhibition of the growth of cecal lactobacilli was also detected after MRP consumption, whereas no significant changes were observed in the other bacterial groups analyzed. Negative correlations were found in the human assay between fecal lactobacilli numbers and the intake of advanced MR compounds (HMF and CML), but not with Amadori compounds (Table 2). On the contrary, fecal bifidobacteria counts were negatively correlated only with the early compounds intake. In the rat assay, cecal lactobacilli were negatively correlated with the three MRP markers intake (Table 3), and no correlations were found with bifidobacteria. Thus, our results in humans suggest a possible negative effect of MRP intake on lactobacilli growth, (supported by the rat assay), which was reduced. However, negative correlations between Amadori compounds intake and bifidobacteria did not lead to significant reductions of fecal bifidobacteria counts. Disparate effects on bifidobacteria have been observed in in vitro experiments, such as a stimulating growth from bread crust melanoidins 12 and coffee 26, or an absence of effect of GL heated mixtures 11, in agreement with our results in rats. A possible explanation for the divergence of the present in vivo studies with previous in vitro work may be the fact that in human intervention studies the concentration of Maillard products is probably much lower than those used for in vitro studies. Some researchers have documented that the consumption of breakfast cereals or coffee by humans resulted in a significant increase of fecal bifidobacteria numbers 13, 27. In line with our results, Borrelli and Fogliano 12 showed that Lactobacillus spp have a poor aptitude to use bread melanoidins for growth, and several authors describe inhibition of lactobacilli growth in the presence of compounds from gluten-glucose model-systems 31 or from autoclaved lactose-amino acids mixtures 32. The pathways by which Maillard products might have an antibacterial activity are not fully understood. Limiting the availability of metal ions due to the binding capacity of MRP has been proposed as a possible mechanism to impair the gut bacterial growth 11, and it was demonstrated that coffee melanoidins chelate essential metals such as iron and Mg²⁺ 33. Moreover, the cell membrane is disrupted after incubation of Escherichia coli with coffee and biscuit melanoidins which is possibly linked to the Mg²⁺ chelating 30. On the other hand, it was postulated that specific substructures of MRP, which occur naturally in heat-treated foods such as coffee, generate reactive oxygen species including H₂O₂ 34. Given that MRP exist naturally in processed foods, it was suggested that these compounds might generate H₂O₂ in heat-treated foods, with the corresponding and well-known antimicrobial effects of this compounds.

Bifidobacteria and lactobacilli contribute to maintain the balance of microbiota populations and, therefore, the inhibition of their growth could result in an increase of other bacterial groups, including detrimental species, among which some pathogen Clostridium spp and Bacteriodes spp have been identified 14. However, no changes in these bacteria were found in the present assay after consumption of MRP-rich diets, according with bibliographic data which show limited effects or stability in clostridia and bacteroides due to the presence of browning compounds 2, 12.

The decreases of enterobacteria observed among humans after BD consumption were correlated with increased intakes of HMF and CML, whereas Escherichia / Shigella growth was negatively correlated only with HMF intake, without correlation with CML (Table 2). Helou et al. 35 observed that CML has no effect on E. coli growth, thereby concluding that these MRP are not consumed by these bacteria, which is in line with the absence of correlation found in the present experiments. Other authors 30 have found that melanoidins of food-origin have an antimicrobial activity against E. coli, by causing irreversible damages to its inner and outer membranes. Specifically, the antibacterial activity against E. coli of cocoa melanoidins was highest for molecular weight fractions <5 kDa 36. It must be highlighted that during the BD period adolescents consumed a wide variety of cocoa derivatives (cocoa powder, snakes, yogurt, chocolate, etc.), which were forbidden in the WD period and, of course, inexistent in the rat diet. Thus, the decrease in the Escherichia/ Shigella group could be at least in part attributed to cocoa melanoidins consumption.

3.3 Implications for human metabolism and health

Currently there is an increasing interest in research on the impact of gut microbiota in health and disease. Evidence support the role of gut microbiota in modulating host metabolism, affecting energy homeostasis, inflammation and development of obesity and related disorders, such as cardiovascular disease and metabolic syndromes 37. According to the European Food Safety Authority, changes in intestinal microbiota derived from prebiotics and probiotics effect should be accompanied by a beneficial physiological or clinical outcome 38. As dietary habits are one of the essential factors contributing to the microbial diversity, food components such as MRP may ultimately affect human health through microbiota modifications. Thus, in the present study, plasma biochemical parameters related with lipid and amino acid metabolisms, as well as anthropometric data, were examined in the young volunteers assay.

Ingestion of the MRP-rich diet did not impact upon the biochemical parameters measured in adolescents in the present conditions (Table 4), and no statistical differences were found compared with the MRP-poor diet. Blood lipids (TC, TG, lipoproteins) and metabolic parameters (glucose, liver enzymes) have been associated with changes in the microbiota composition in mice 39. However, significant decreases in blood lipids after prebiotic intake have usually been found in hyperlipidaemic subjects 40, but not in those normolipidaemic 41. As individuals of the present study were healthy boys and their biochemical parameters were within the normal range, results of the present assay were not surprising. Moreover, diets designed for the feeding periods were balanced and based in the adolescent's food preferences, thus, similar to those consumed usually by this population, as commented. Therefore, changes in gut microbiota found after consumptions of the diets in our experimental conditions did not lead to modifications in the measured blood parameters or anthropometric data. The intake of the three MRP analyzed was not statistically correlated with none of the biochemical variables analyzed. Assays in rats have also shown that the presence of MRP in the diets failed to lower liver and plasma cholesterol levels, and do not modify LDL or HDL levels 42.