Melanoidins from Coffee, Cocoa, and Bread Are Able to Scavenge α-Dicarbonyl Compounds under Simulated Physiological Conditions.

Free amino residues react with α-dicarbonyl compounds (DCs) contributing to the formation of advanced glycation end products (AGEs). Phenolic compounds can scavenge DCs, thus controlling the dietary carbonyl load. This study showed that high-molecular weight cocoa melanoidins (HMW-COM), HMW bread melanoidins (HMW-BM), and especially HMW coffee melanoidins (HMW-CM) are effective DC scavengers. HMW-CM (1 mg/mL) scavenged more than 40% DCs within 2 h under simulated physiological conditions, suggesting some physiological relevance. Partial acid hydrolysis of HMW-CM decreased the dicarbonyl trapping capacity, demonstrating that the ability to react with glyoxal, methylglyoxal (MGO), and diacetyl was mainly because of polyphenols bound to macromolecules. Caffeic acid (CA) and 3-caffeoylquinic acid showed a DC-scavenging kinetic profile similar to that of HMW-CM, while mass spectrometry data confirmed that hydroxyalkylation and aromatic substitution reactions led to the formation of a stable adduct between CA and MGO. These findings corroborated the idea that antioxidant-rich indigestible materials could limit carbonyl stress and AGE formation across the gastrointestinal tract.

Introduction

Reactive α-dicarbonyl compounds (DCs) are formed during Maillard-induced carbohydrate degradation. Glyoxal (GO), methylglyoxal (MGO), diacetyl (DA), and 3-deoxyglucosone are the most extensively characterized markers, and they are key precursors of advanced glycation end products (AGEs).[1−3] DCs can be formed also in vivo where they can contribute to the development of chronic diseases, oxidative and inflammatory cascades and to the formation of endogenous AGEs.[4,5] In addition, DCs may influence normal physiological functions through carbonylation of lipids, proteins, or DNAs.[6] While for dietary AGEs a cause–effect relationship with health outcomes is still a hypothesis, DCs generated during glycolysis are involved in the pathogenesis of diabetic complications including obesity, nephropathy, and neuropathy,[7,8] and their excessive production was demonstrated in diabetes.[9]

Dicarbonyl trapping has been proposed as an effective strategy to control undesired outcomes in foods and to prevent further complications in vivo.[10] Several natural foods and their extracts inhibit dicarbonyl-induced reactions by their antioxidant activity and by reducing GO and MGO concentration. Effective examples in foods include tea catechins, ginger shogaols,[11] secoiridoid derivatives in olive leaf and oil,[12] and hydroxycinnamic acid derivatives,[13] while creatine can scavenge reactive DCs in meat products[14] and in vivo.[15] A major role in scavenging in vivo formation of DCs is played by glyoxalase, an ubiquitous enzyme connected with glycolysis.[16]

There is no consensus on the correlation between dietary intake of DCs and their in vivo concentrations. A correlation between endogenous MGO and a DC-rich diet has been recently proposed; however, it cannot be ruled out that the correlation is driven by many confounding factors such as lipid and protein concentrations in the diet.[17,18]

The hypothesis that reducing the dietary intake of DCs or scavenging them within the gastrointestinal tract could reduce the endogenous DC concentration is interesting[4] and can be of particular relevance in pathological conditions. Treatment with dietary flavonoids was effective in promoting a significant decrease in the level of DCs in the human body.[19]

Melanoidins, high-molecular weight brown polymers present in foods, are a bioactive source of reducing power combining reductones and condensed phenolic compounds with antioxidant, antimicrobial, and prebiotic capacities.[20,21] Melanoidins act as a functional dietary fiber, favoring the delivery of reducing capacity along the gastrointestinal tract.[22] As melanoidins are formed during food processing, their structure and composition change with the food; the principal constituents of coffee and cocoa melanoidins are polysaccharides; in bread, melanoidins are formed by gluten proteins and starch cross-linked by Maillard reaction products.[22] In coffee, cocoa, nuts, and fruits, besides carbohydrates and proteins, phenolic compounds can be incorporated into melanoidin structures. Fragments of chlorogenic acids (CGAs) have been found in coffee melanoidins,[23] and cocoa melanoidins contain polyphenols such as epicatechins.[24] Fogliano and Morales[25] estimated that the dietary intake of melanoidins from all of the possible sources could be close to 10.0 g per average consumer. Therefore, their role as DC scavengers can have a physiological relevance modulating the amount of dietary DCs inside the gut lumen and their bioavailability.

In this study, high molecular weight melanoidins from cocoa, coffee and bread were selected as they represent phenolic-rich (coffee and cocoa) and nonphenolic-rich (bread) melanoidins. We aimed at comparing the dicarbonyl trapping capacity of these different melanoidins sources in an in vitro model system in order to elucidate possible mechanisms behind the trapping process and to explore the possibility of using melanoidins as a bioactive source of reducing compounds.

Materials and Methods

Chemicals and Melanoidin Sources

GO aqueous solution (GO, 40%), MGO aqueous solution (MGO, 40%), DA, quinoxaline (QX), 2-methylquinoxaline (2-MQX), 2,3-dimethylquinoxaline (2,3-MQX), pyridoxamine (PM) dihydrochloride , diethylenetriaminepentacetic acid (DETAPAC), o-phenylenediamine (OPD), ethylenediaminetetracetic acid (EDTA), 3-caffeoylquinic acid (3-CQA), caffeic acid (CA), ferulic acid (FA), and Pronase E were purchased from Sigma-Aldrich (St. Louis, MO). Hydrochloric acid (37%), sodium hydroxide, disodium hydrogen phosphate dihydrate, sodium dihydrogen phosphate dihydrate, and acetic acid were obtained from Merck (Darmstadt, Germany). Dried cocoa beans (Forastero) were provided by CACEP (Villahermosa, Mexico). Dark roasted arabica coffee beans and bread crumbs were purchased from a local supermarket.

Preparation of High Molecular Weight Coffee Melanoidins (HMW-CM)

Extraction of HMW-CM followed the procedure described by Nunes and Coimbra[26] with some modifications. Coffee beans were ground to an average particle size of 0.4 mm. Lipids were removed by using dichloromethane (1:30, w/v, three times), and 100 g of coffee powder was extracted using 1200 mL of water at 80 °C for 20 min, and then filtered through a filter paper (Whatman 595, Billerica, MA) under vacuum. The filtrate was dialyzed (MW cutoff 14 kDa, D9402, Sigma-Aldrich) at 4 °C with 10 water renewals (1200 mL for each cycle). After dialysis, the retentate was freeze-dried to obtain HMW-CM.

Preparation of High Molecular Weight Cocoa Melanoidins (HMW-COM)

HMW-COM was obtained from toasted cocoa beans according to the methods described by Summa et al.[27] In brief, 200 g of deshelled cocoa beans was roasted using a convection oven (150 °C, 210 min, Memmert, Schwabach, Germany) and cryogenically ground to a fine powder by a cryogenic grinder (6875D Freezer/Mill, SPEX SamplePrep, UK). Cocoa powder was defatted with petroleum ether at 30 °C for 1.5 h and centrifuged at 3000g for 15 min at 4 °C. The supernatant was discarded, and the process was repeated three times. After air-drying overnight, the defatted cocoa powder (50 g) was extracted with 500 mL of water at 80 °C for 20 min. Then, the aqueous solution was centrifuged at 5000g for 10 min, and the supernatant was filtered through a filter paper (Whatman 595) to remove the insoluble materials. The filtrate (400 mL) was dialyzed using a dialysis membrane (MW cutoff 14 kDa) for 3 days with eight water renewals until conductivity reached a value lower than 2 μS/cm detected by a conductivity meter (WTW inoLab Cond 7110, Fisher Scientific, Sweden). After storage at −20 °C, the retentate was lyophilized to yield HMW-COM. All freeze-dried HMW-COM samples were kept at −20 °C.

Preparation of High Molecular Weight Bread Melanoidins (HMW-BM)

Before preparing HMW-BM as described by Borrelli et al.,[28] bread crumbs were heated at 200 °C for 15 min to generate enough bread melanoidins. After milling and sieving to a particle size of 0.15 mm, 50 g of bread powder was mixed with 600 mL of 0.2 M Tris-HCl buffer (pH 8.0) containing 0.7 U/mL of Pronase E and digested at 37 °C for 72 h. After centrifugation (4000g, 4 °C, 15 min), the supernatant was filtered through a Whatman 595 filter paper and dialyzed (MW cutoff 14 kDa) against 4 L of distilled water at 4 °C for 4 days with 10 water renewals. The collected retentate was freeze-dried and kept at −20 °C until used.

Evaluation of the Direct Dicarbonyl Trapping Capacity

Direct GO, MGO, and DA trapping capacities were determined using the method described by Glomb and Tschirnich[29] with some modifications. GO (0.37 mg/mL), MGO (0.46 mg/mL), DA (0.55 mg/mL), PM (1.08 mg/mL), CA (1.15 mg/mL), and 3-CQA (2.27 mg/mL) were separately dissolved in phosphate buffer (0.1 mol/L, pH 7.4) in order to obtain the same molarity (6.4 mmol/L). QX and 2,3-MQX (10 mmol/L) were dissolved in 20% aqueous methanol separately. GO, MGO, or DA (100 μL) was mixed with 750 μL of phosphate buffer and 100 μL of either phosphate buffer (blank), PM solution (positive control), or melanoidin solutions (0.1–25 mg/mL) and then incubated at 37 °C up to 168 h. Melanoidin solutions were prepared in the range of 0.01–2.5 mg/mL in model systems, while the final concentration of PM and DCs was 0.64 mmol/L. Additionally, in order to monitor the kinetic profile of DCs scavenged by HMW-CM and its predominant phenolic acids, 750 μL of phosphate buffer, 100 μL of one of the DCs, and 100 μL HMW-CM (20 mg/mL) or phenolic acids mentioned above were incubated at 37 °C for 2, 4, 24, 48, 120, and 168 h. All of the incubated samples were mixed with 200 μL of 0.2% OPD solution containing DETAPAC (9.6 mM) and 50 μL of 2,3-MQX (in reaction system with GO) or QX (in reaction system with MGO and DA) as the internal standard; all solutions were vortexed for 5 s. The mixture was kept at 37 °C in the dark for 2 h and filtered using a 0.22 μm polyvinylidene fluoride (PVDF) filter before high-performance liquid chromatography analysis. To evaluate the physiological relevance of dicarbonyl scavenging ability of HMW-CM and HMW-COM, the estimated dietary intake of coffee melanoidins (1.0 g/person per day)[25] and MGO (1.9 mg/person per day)[30] was reacted within an assumed digestive volume of the 1 L mimicking upper intestinal phase. In brief, 1.0 mg/mL of HMW-CM or HMW-COM and 0.026 mmol/L of GO, MGO, or DA were incubated together at 37 °C (pH 7.4) up to 2 h. Then, incubated mixtures were derivatized using OPD as described above and subsequently subjected to liquid chromatography–tandem mass spectrometry (LC–MS/MS) analysis.

Determination of QX Derivatives

Liquid Chromatography UV

Determination of three QX derivatives was performed on a Thermo Ultimate 3000 ultrahigh-pressure liquid chromatography system coupled with an RS 3000 diode array detector (DAD, Thermo Fisher Bremen, Germany). A Kinetex EVO C18 column (150 mm × 2.1 mm, 2.6 μm; Phenomenex, Aschaffenburg, Germany) equipped with a security guard of the same stationary phase was used. Separation was achieved through a gradient mixture of (A) 0.1% formic acid in water and (B) 0.1% formic acid in methanol at a flow rate of 0.4 mL/min, while the column was thermostated at 40 °C. The eluant program was as follows: 0–2 min, 2% B; 2–8 min, 2–25% B; 8–10 min, 25–50% B; 10–13 min, 50–95% B; 13–14 min, 95–2% B; and 14–17 min, 2% B. Chromatograms were recorded at 315 nm, and the retention times of QX, 2-MQX, and 2,3-MQX were 9.12, 11.13, and 12.03 min, respectively.

Liquid Chromatography–Tandem Mass Spectrometry

Determination of QXs formed during the simulated upper intestinal phase was conducted under the same chromatographic and UV conditions described above but with a TSQ triple quadrupole mass spectrometer as the detector (Thermo Fisher Scientific, Bremen, Germany). Positive electrospray ionization was used for detection, and the electrospray source parameters were set as follows: spray voltage 4.0 kV; capillary temperature 350 °C; dwell time 100 ms; sheath gas and aux gas were set to 10.0 and 5.0 AU (arbitrary unit). The QX derivatives of GO, MGO, and DA were identified by the selected reaction monitoring mode using the following transitions (in parenthesis collision energy, CE): GOqx: m/z 131 → 77 (CE: 25 V); MGOqx: m/z 145 → 77 (CE: 28 V) and DAqxm/z 159 → 77 (CE: 32 V).

Quantification

The amount of unreacted GO, MGO, and DA in different samples was calculated through the ratio of peak area of QX, 2-MQX, 2,3-MQX, and their corresponding internal standard, respectively. Calibration curves were in the concentration range of 2.0–100.0 ppm for UHPLC analysis and 0.02–5.0 ppm for UHPLC–MS/MS analysis, with linearity higher than 0.99 for all the investigated compounds in both conditions. Percentage decrease in each DC was calculated using the following equation (eq )

Release of Bound Phenolic Acids from HMW-CM

Adsorbed Phenolic Compounds

Noncovalently bound phenolic compounds were released according to the method reported by Delgado-Andrade and Morales[31] with few modifications. Briefly, 45 mg of HWM-CM was incubated in 5.095 mL of NaCl solution (2 M) overnight at 4 °C. Then, the extracts were centrifuged (4000g, 4 °C, 10 min), and the clear supernatant was filtered through a 0.22 μm PVDF filter to give saline-treated HWM-CM.

Acidic Hydrolysis

Bound phenolic compounds were extracted using acidic hydrolysis of HMW-CM as described by Oracz et al.[32] with some modifications. HMW-CM (3 mL, 15 mg/mL) in 50% aqueous methanol was mixed with 1 mL of HCl (10.2 M) and placed in a heating block at 75 °C for 150 min. After hydrolysis, the solution was neutralized with 1095 μL of 10 M NaOH resulting in a final volume of 5095 μL and centrifuged at 4000g and 4 °C for 10 min. The clear supernatant was filtered using a 0.22 μm PVDF filter to give acid-hydrolyzed HMW-CM. Another 45 mg of HMW-CM was dissolved in 5095 μL of water and centrifuged (4000g, 4 °C, 10 min), and then the supernatant was filtered through a 0.22 μm PVDF filter as nontreated HMW-CM.

Alkaline Hydrolysis

Covalently bound phenolic compounds were released by alkaline hydrolysis of HMW-CM using the method described by Coelho et al.[33] with some modifications. Briefly, 45 mg of HMW-CM was dissolved in 3 mL of 2 M NaOH solution containing 20 mM EDTA. After incubation at 30 °C for 90 min, the mixture was adjusted to pH 7.0 with 950 μL of 6 M HCl, and 1145 μL of water was added to achieve a 5095 μL of final volume. The solution was centrifuged (4000g, 4 °C, 10 min), and the resulting supernatant was filtered with a 0.22 μm PVDF filter to give alkali-hydrolyzed HMW-CM.

The obtained nontreated, saline-treated, acid- and alkali-hydrolyzed HMW-CM were directly subjected to analysis of predominant phenolic acids and evaluation of the dicarbonyl trapping capacity.

Analysis of Predominant Phenolic Acids

Predominant phenolic acids in saline-treated, acid- and alkali-hydrolyzed HMW-CM were analyzed by LC–DAD on a Kinetex EVO C18 column (150 mm × 2.1 mm, 2.6 μm, Phenomenex) equipped with a security guard of the same stationary phase. Eluant A was 0.1% formic acid aqueous solution, and eluant B was 0.1% formic acid in acetonitrile. The gradient was as follows: 0–5 min, 2% B; 5–15 min, 2–17% B; 15–30 min, 17% B; 30–50 min, 17–50% B; 50–70 min, 50–90% B; 70–71 min, 90–2% B; and 71–75, 2% B. The flow rate was 0.4 mL/min. and the column oven was set at 35 °C. Detection was performed at 325 nm, while the chromatographic stream was continuously monitored from 200 to 600 nm. The identification of CA, 3-CQA, and FA was conducted through comparison of their retention times and UV–vis spectra with that of pure CA, 3-CQA, and FA standards under the same chromatographic conditions. 4-caffeoylquinic acid and 5-caffeoylquinic acid were tentatively identified by comparison of their retention times and UV–vis spectra according to Fernandez-Gomez et al.[34] Total CQAs were quantified by the external standard technique using a 3-CQA calibration curve, while CA and FA were quantified by using their respective standards.

Analysis of the Adducts in the MGO–CA Model System by LC–MS/MS

CA (5 mM in 100 mM phosphate buffer, pH 7.4) was incubated alone or with MGO (5 mM in 100 mM phosphate buffer, pH 7.4) for 72 h at 37 °C to give the CA or CA–MGO model system, and then the incubated samples were subjected to UHPLC–MS/MS analysis. Chromatographic separation was performed using a Kinetex EVO C18 column (150 mm × 2.1 mm, 2.6 μm) at 30 °C. The mobile phase consisted of 0.1% formic acid in water (mobile phase A) and 0.1% formic acid in methanol (mobile phase B) at a flow rate of 0.4 mL/min using the following gradients: 0–5 min, 2% B; 5–15 min, 2–17% B; 15–25 min, 17–100% B, and 25–30 min held at 100% B for 5 min, and then the column was re-equilibrated with 2% B for 5 min. The ion source was operated in the negative electrospray ionization mode with the spray voltage at 3.0 kV. Sheath and auxiliary nitrogen gas were used at a flow rate of 50 and 25 AU, respectively. A preliminary trial identified molecular ions of the possible DC-hydroxycinnamic derivative adducts; the selected ion monitoring mode was used in three different channels: m/z 179 for CA, m/z 251 for mono-MGO–CA and m/z 323 for di-MGO–CA. Structural information on CA and the major MGO adducts of CA was obtained by tandem mass spectrometry through collision-induced dissociation with 30 eV collision energy. The mass range was measured from m/z 50 to 350, and data were acquired with Xcalibur version 4.0 (Thermo Fisher).

Statistical Analysis

All experiments were performed in triplicate unless otherwise stated. Significant differences (p < 0.05) in the dicarbonyl trapping capacity of samples were analyzed by Tukey’s HSD test using the SPSS statistics (v. 23.0, IBM, Armonk, NY). The error bar in all figures correspond to the standard deviation (SD).

Results and Discussion

Evaluation of the Direct Dicarbonyl Trapping Capacity of HMW-CM, HMW-COM, and HMW-BM and Their Physiological Relevance

HMW-CM, HWM-COM, and HMW-BM dicarbonyl trapping capacities are reported in Figure . All three DCs were effectively scavenged by both HMW-CM and HMW-COM when the concentration of melanoidins is higher than 0.5 mg/mL. HMW-CM showed 68.0, 76.9, and 64.8% trapping capacities for GO, MGO, and DA, respectively, at a concentration of 1 mg/mL. Similarly, HMW-COM was more effective in scavenging GO and MGO while less efficient in scavenging DA than HMW-CM. Considering nonphenolic-rich melanoidins, only MGO (38.1%) was quenched by the highest concentration of HMW-BM, revealing that polysaccharides and protein-based melanoidins have a specific chemical nature able to block MGO, while phenolic compounds in HMW-CM and HMW-COM contribute to the elimination of a wider spectrum of DCs. None of the three DCs was detected after 7 days incubation of three melanoidins (2.5 mg/mL), indicating that melanoidins were not able to release DCs in physiological conditions (data not shown).

Figure 1

GO, MGO, and DA trapping capacities of HMW-CM, HMW-COM, and HMW-BM with different concentrations (0.01–2.5 mg/mL) and PM (0.108 mg/mL) at 168 h. Results are expressed as mean ± SD for n = 3. Bars with the same letter are not significantly different according to Tukey’s HSD test at p > 0.05.

PM was used to compare melanoidins activity to a well-known carbonyl scavenger. The concentration of HMW-CM and HMW-COM was higher than that of PM (0.108 mg/mL) used in model systems; however, both were more effective than PM in quenching DCs considering differences in their molecular weight. PM exhibited a specific trapping capacity; it was an efficient MGO scavenger, an intermediate DA scavenger, and a weak GO scavenger. This result was in line with previous studies, where MGO was found to be the most efficient α-DCs in reacting with amino residues and in particular with PM because of its ability to form stable heterocyclic compounds.[35,36]

To investigate the physiological relevance of DC trapping capacity by coffee and cocoa melanoidins, they should be tested at a concentration compatible with the daily intake of melanoidins which is in the range 0.5–2.0 g for moderate and heavy consumers.[25] Data summarized in Figure showed that around 40% of GO and MGO were scavenged within 2 h by HMW-CM and HMW-COM, respectively, and for DA, HMW-CM showed a significantly higher efficacy than the other one, reaching about 60%, which is in line with the results presented in Figure . A concentration of coffee melanoidins fluctuating between 0.25 and 1 mg/mL in the colon was assayed, assuming that the colon accumulates coffee melanoidins over at least 24 h in a maximum volume of about 2 L.[37] These values showed that melanoidins were effective in trapping DCs not only in the intestinal phase within 2 h but also in the colon to exert further scavenging activity in combination with the microbial population.[38] This evidence suggested that a regular intake of HMW-CM from coffee brew could be effective in controlling carbonyl loading in human bodies.

Figure 2

Dicarbonyl trapping capacity of HMW-CM (1.0 mg/mL), HMW-COM (1.0 mg/mL), and PM (0.108 mg/mL) within 2 h under simulated physiological conditions. The concentration of DCs and melanoidins was calculated according to the estimated daily intake. Results are expressed as mean ± SD for n = 3. Bars with the same letter are not significantly different according to Tukey’s HSD test at p > 0.05.

Time-Course Evaluation for the HMW-CM Dicarbonyl Trapping Capacity

HMW-CM and its predominant phenolic acid, 3-CQA, and CA were subjected to a time-course investigation in order to gain insights into the reaction between DCs and phenolic compounds in melanoidins. As shown in Figure , the GO concentration was reduced by CA up to 11.8% within the first 4 h, while HMW-CM diminished MGO and DA down to 11.8 and 10.2%, respectively. HMW-CM was a better MGO scavenger (50% reduction within 18 h) than GO and DA (50% reduction in about 40 h) upon longer incubation time. HMW-CM at a concentration of 2 mg/mL was characterized by a similar DC trapping capacity toward 0.115 mg/mL of CA (0.64 mmol/L), while its activity was significantly higher than 0.227 mg/mL of 3-CQA (0.64 mmol/L). The amounts of DCs scavenged by 3-CQA increased continuously during incubation, reaching 82.2, 87.6, and 100% for GO, MGO, and DA, respectively, after 168 h, following the same trend of HMW-CM and CA. MGO reduction was in line with Mesías and co-workers,[39] although in other conditions higher values have been reported by Yoon,[40] probably because of the higher concentration of 3-CQA.

Figure 3

Time-course of GO (A), MGO (B), and DA (C) trapping capacity of PM (0.108 mg/mL), HWM-CM (2 mg/mL), CA (0.115 mg/mL), and 3-CQA (0.227 mg/mL). Results are expressed as mean ± SD for n = 3.

Moreira et al.[41] pointed out that CGA and CGA derivatives are the most abundant phenolic compounds in HMW-CM, and the amount of phenolic compounds incorporated in HMW-CM ranged from 76.5 to 224.0 mg/g (in equivalent weight of CA and CGA).[33] According to these values, bound phenolic compounds in melanoidins were at the same level with the concentration of CA and 3-CQA used in the model system. Altogether, these results supported the hypothesis that CGAs and CA extensively contributed to the overall dicarbonyl trapping capacity of HMW-CM.

Influence of Hydrolysis on the Dicarbonyl Trapping Capacity of HMW-CM

A deeper examination of the relationship between the bound phenolic compounds and dicarbonyl trapping capacity of HMW-CM was achieved through a partial cleavage of the HMW-CM structure. Two hydrolysis methods were used in order to understand the role of melanoidins in scavenging DCs. It was found that 3-CQA, 5-caffeoylquinnic acid, and 4-caffeoylquinnic acid were identified as the most abundantly released phenolic compounds in saline-treated HMW-CM as shown in Figure S1B. Table outlined that the content of absorbed CQAs (298.7 mg/100 g HMW-CM) was in the same order of magnitude as previous studies,[33] although the relative proportion of each CQA was characterized by slight differences probably because of isomerization occurring during roasting and dialysis.[34,42]

Table 1

Content (mg/100 g HWM-CM)a of Released Phenolic Acids from HMW-CM after Different Treatments

	saline treatment	acidic hydrolysis	alkaline hydrolysis
total CQAs	298.7 ± 6.3	nd	nd
CA	nd	65.5 ± 1.7	671.1 ± 21.5
FA	nd	34.7 ± 5.5	95.7 ± 3.6

All values are shown as means ± SD (n = 3). nd, not detected.

A significant amount of phenolic compounds was detected upon acidic and alkaline hydrolysis, especially CA (65.5 and 671.1 mg/100 g HMW-CM) and FA (34.7 and 95.7 mg/100 g HMW-CM) as shown in Table . The amount and nature of phenolic compounds bound in coffee melanoidins were previously investigated by releasing ester-linked, condensed, and glycosidically linked phenolic compounds through alkaline saponification, alkaline fusion, and acidic hydrolysis, respectively.[43,44] Our results are in line with previous studies,[33,45] where high amounts of CA and FA were detected in coffee brew or in high-molecular weight melanoidins after using alkaline hydrolysis, highlighting that most of the phenolic compounds in coffee melanoidins are covalently bound to the polysaccharide skeleton. In contrast, the content of CA and FA released after acidic hydrolysis was lower than that of the one reported by Moreira and co-workers,[43] probably because of the differences in the hydrolysis time.

In Figure , the total dicarbonyl scavenging activity of HMW-CM before and after saline and hydrolytic treatments was highlighted: the total amount of reactive carbonyls scavenged in the nontreated and saline-treated HMW-CM groups was similar, both significantly higher than that in alkali- and acid-hydrolyzed HMW-CM groups. The acid-hydrolyzed HMW-CM exhibited the lowest amount of free phenolic compounds and blocked 53.7, 63.2, and 46.3% of GO, MGO, and DA, respectively, resulting to be more effective than alkaline-saponified HMW-CM. According to our preliminary trials, ascorbic acid was not added as it could react with DCs, leading to an overestimation of the result (data not shown). The loss of total dicarbonyl trapping capacity in alkali- and acid-hydrolyzed HMW-CM, therefore, is probably due to the oxidation of phenolic acids during hydrolysis. In addition, DC scavenging contributed by free phenolic compounds was estimated according to their content and concentration–response relationship shown in Figure S2. We observed that the DC trapping capacity of coffee melanoidins was mainly associated to the bound phenolic acids, especially CGAs and CA, as the amount of free phenolic acids in nontreated HMW-CM was not sufficient to scavenge DCs, as presented in Figure . It was also found that most of the dicarbonyl trapping capacity was still related to the bound phenolic compounds after hydrolysis, while the proportion of DCs scavenged by bound phenolic compounds decreased with the release of free phenolic acids. Previous studies reported that phenolic compounds in coffee melanoidins are mainly in the condensed form, which can be released by alkaline fusion but not hydrolysis,[33] and acidic hydrolysis is not an adequate technique to release bound coffee phenolic compounds compared to alkaline saponification, keeping most of the melanoidins intact,[45] which is in agreement with our results. The interplay between bound phenolic compounds and complex macromolecular structures, as polysaccharides or polypeptides, could act as a “dicarbonyl sponge” following chemical mechanisms previously depicted for antioxidant activity in insoluble polymerized materials.[46] According to the concept of the “antioxidant dietary fiber”,[47] here, we demonstrated that melanoidin-bound polyphenols are able to quench DC compounds, thus contributing to the control of carbonyl stress.

Figure 4

GO, MGO, and DA trapping capacities of nontreated, saline-treated, acidic- and alkaline-hydrolyzed HMW-CM after 168 h incubation. Results are expressed as mean ± SD for n = 3. Different letters indicate significant differences according to Tukey’s HSD test at p > 0.05.

The ability of bound polyphenols to scavenge DCs paralleled previous reports on antioxidant activity of roasted coffee: the higher the degree of roasting, the more pronounced is the redox potential.[48] Besides pioneering studies on antioxidant activity of coffee melanoidins in vitro,[49,50] direct confirmation of the mechanisms can be inferred from the behavior of coffee melanoidins in vivo. Dittrich and co-workers[51] demonstrated that a melanoidin-rich diet promoted the oxidative stability of LDL up to 35%, suggesting that the bound undigested antioxidants can play an active role. Coelho et al.[33] reported that the association of the condensed phenolic structure with indigestible polysaccharides present in HMW-CM probably is one of the reasons for the observed relationship between coffee consumption and the antioxidant activity of feces.

Studying the Formation of MGO Adducts of CA under Simulated Physiological Conditions by LC–MS/MS.

We hypothesized that the CA moiety in the coffee melanoidin skeleton was mainly responsible for the dicarbonyl trapping capacity. MGO was incubated with CA to investigate the potential formation of carbonyl adducts. Figure A–D presented the total ion chromatograms of the model system with CA and CA–MGO, and the extracted ions of the new peaks formed during the incubation. After 3 days, CA decreased by nearly 48.7%, and two new peaks were annotated upon comparison with control without MGO. The peak belonging to CA appeared at 13.82 min with the molecular ion m/z 179 [M–H]−, and the two new peaks at 15.07 and 15.69 min had molecular ion m/z 251 [M – H]− (mono-MGO–CA adduct, mass shift +72 m/z) and m/z 323 [M – H]− (di-MGO–CA adduct, mass shift +144 m/z), respectively, suggesting that these two molecular ions were mono-MGO- and di-MGO-conjugated CA.

Figure 5

Total ion chromatogram of the CA (A) and CA–MGO (B) systems after incubation at 37 °C for 3 days, extracted ion chromatogram [M – H]− of mono-MGO–CA adduct [m/z, 251, (C)] and di-MGO–CA adduct [m/z, 323, (D)], and MS/MS spectra of CA (E), mono-MGO–CA adduct (F), and di-MGO–CA adduct (G).

Structural information of CA and its MGO adducts was obtained through MS/MS spectra. As shown in Figure E–G, the major fragment ions of CA were m/z 135 [M – 44 – H]− and m/z 134 [M – 45 – H]−, which are in line with the typical fragmentation pattern of CA.[52] Fragment ions with m/z 207 [M – 44 – H]− and m/z 135 [M – 44–72 – H]− from the mono-MGO–CA adduct could match with the decarboxylation of CA and subsequent loss of the MGO moiety. Major fragments at m/z 205 and m/z 190 could be obtained by cyclization followed by α-cleavage in the MGO moiety as highlighted in Figure S3 panel B. In MS/MS spectra of di-MGO–CA, fragment ions at m/z 277, m/z 251, and m/z 207 indicated the structure of mono-MGO–CA with the same fragmentation profile. This result suggested that MGO could be attached to unsubstituted carbons in the benzene ring of CA, following a similar reaction scheme as proposed for carbonyl trapping reactions of epicatechins and olive phenols.[12,53]

We hypothesized that also in physiological conditions, the reaction mechanism between CA and MGO can lead to the formation of two isomers as depicted in Figure . Several studies suggested that phenolic compounds undergo electrophilic aromatic substitution reactions with GO or MGO at physiological conditions, with the hydroxy group in the MGO moiety close to the aromatic ring.[53−55] However, Hidalgo and co-workers[56] observed that the carbonyl group was conjugated with the aromatic ring after heating GO and resorcinol together at 100 °C for 3 h, which implies that isomerization may also occur under physiological conditions to extend the conjugation. The unsubstituted carbon 2 and 6 should be the major active site for scavenging DCs. Specifically, the trapping reactions between phenols and DCs are hydroxyalkylation and aromatic substitution reactions, and the hydroxyl groups in the aromatic ring could increase the reactivity of unsubstituted carbon atoms to attack carbonyl carbon atoms. Conversely, the replacement of hydroxyl in C3 by methoxyl leads to the loss of MGO trapping capacity of FA and [6]-shogaol,[11] which means that the replaced methoxyl group decreases the activity of C2 and C6 because both hydroxyl and methoxyl are ortho–para directing groups.

Figure 6

Proposed mechanism of reaction for trapping of MGO by CA under simulated physiological conditions.

Concerning HMW-CM as sources of hydroxycinnamic acids, especially CGAs and CA, the presence of melanoidin-bound phenolic structures may also enhance DC trapping efficacy of each other as a result of additive effects. The ability of melanoidin-bound phenolics in scavenging DCs follows recently pointed out effects of the food matrix in modulating chemical reactivity:[57] structural organization at the molecular and macroscopic level is responsible of the reaction routes in a specific food environment.

In summary, the present study revealed that polyphenol-rich melanoidins, such as HMW-CM and HMW-COM, can scavenge DCs, thus mitigating the negative consequences of their reaction with other macromolecules in physiological conditions. This is the first report that presents the tentative detection of mono- and di-MGO adducts of CA, which indicates that CA scavenges DCs through trapping reactions. Partial hydrolysis and the release of melanoidin-bound phenolic compounds illustrate that phenolic compounds bound to macromolecules are still active to scavenge DCs in vitro and confirm the possibility of using antioxidant dietary fibers as functional ingredients to quench dicarbonyl species along the gastrointestinal tract.