Article pubs.acs.org/JAFC
New Insight into the Role of Sucrose in the Generation of α‑Diketones upon Coffee Roasting Luigi Poisson,† Noémie Auzanneau,† Frédéric Mestdagh,† Imre Blank,‡ and Tomas Davidek*,† †
Nestlé Product Technology Centre Orbe, Nestec Ltd., CH-1350 Orbe, Switzerland Nestlé Research Centre, P.O. Box 44, CH-1000 Lausanne 26, Switzerland
‡
ABSTRACT: The origin and formation pathways of the buttery-smelling α-diketones 2,3-butanedione and 2,3-pentanedione upon coffee roasting were studied by means of biomimetic in-bean experiments combined with labeling experiments. For this purpose natural sucrose in the coffee bean was replaced by fully or partially 13C-labeled sucrose or by a mixture of unlabeled and fully 13C-labeled sucrose (CAMOLA approach). The obtained data point out that sucrose contributes to both α-diketones; however, its importance and reaction pathways clearly differ. Whereas the major part of 2,3-pentanedione originates from sucrose (about 76%), its contribution to 2,3-butanedione is much lower (about 35%). Formation from intact sugar skeleton is the major pathway generating 2,3-pentanedione from sucrose, whereas 2,3-butanedione is mainly generated by recombination of sucrose fragments. The contribution of glucose and fructose moieties of sucrose to both α-diketones is comparable. Finally, kinetic experiments with fully labeled sucrose showed that the contribution of sucrose changes during roasting. KEYWORDS: coffee, flavor, precursor, sucrose, CAMOLA, roasting
■
mocha note.17 These mechanisms include generation of 2,3butanedione from intact C4 hexose backbone, recombination of C1/C3 (formaldehyde and 1-hydroxy-2-propanone) and C2/C2 (acetaldehyde and glycolaldehyde) sugar fragments, and glycine-mediated chain elongation of glyoxal and methylglyoxal.18−22 Similarly, several mechanisms were proposed to explain the formation of 2,3-pentanedione. These includes recombination of C1/C4 sugar fragments (e.g., formaldehyde and 2,3butanedione), recombination of C2/C3 sugar fragments (e.g., acetaldehyde and 1-hydroxy-2-propanone), or alanine-mediated chain elongation of methylglyoxal.15,18,23 In general, the number of precursors in model systems is strongly limited to reduce the complexity. Hence, such systems cannot reproduce the chemical and physical transformations of the coffee beans during roasting. To study the formation pathways of coffee aroma compounds under more real conditions, the so-called biomimetic in-bean experiments were developed, where the coffee bean itself is used as a pressurized reaction vessel.24−26 The results revealed among others an important role of the soluble saccharides in the formation of α-diketones, whereas free amino acids played only a minor role. In addition, different formation pathways leading to 2,3-butanedione and 2,3-pentanedione were highlighted by employing labeled precursors.26 The in-bean approach was also successfully applied to study the mechanism of coffee
INTRODUCTION The delightful aroma and taste of coffee are developed during coffee roasting at temperatures >200 °C, and it is generally accepted that the coffee’s intrinsic quality is predetermined in the green bean by its precursor composition. The main constituents in green coffee are carbohydrates, nitrogen-containing compounds (mainly proteins, trigonelline, and caffeine), lipids, organic acids, and water. 1 The carbohydrates represent about half of the dry basis of green coffee beans.2,3 The main part is the insoluble fraction that forms the structure of the cell walls consisting of cellulose, galactomannans, and arabinogalactans along with proteins and chlorogenic acids, all of them showing complex structures.4 Nevertheless, it is the water-soluble coffee fraction that is considered as the more important precursor pool. Particularly, the low molecular weight constituents, comprising free sugars, amino acids, trigonelline, and chlorogenic acids,4−6 rapidly degrade at the early stage of roasting and instantly participate in manifold reactions.7 Free sugars are almost exclusively represented by sucrose with about 8% of dry matter in Arabica and 3−6% in Robusta. Its fast hydrolysis at the beginning of the roasting process releases the reducing saccharides glucose and fructose, which thereupon are strongly involved in caramelization and Maillard-type reactions. Arabinose was also discussed as potential precursor, released from the arabinogalactans during roasting.7−10 Other free sugars such as galactose, mannose, or glucose are present in only trace amounts. A large number of model studies were conducted to understand the mechanism underlying the formation of different key coffee odorants such as thiols, α-diketones, furanones, and pyrazines under dry heating conditions.11−16 For example, several mechanisms were proposed to explain the formation of buttery-smelling 2,3-butanedione and 2,3pentanedione, which were recently shown to render the harsh roasty/sulfury note of 2-furfurylthiol to the pleasant coffee/ © 2016 American Chemical Society
Special Issue: 11th Wartburg Symposium on Flavor Chemistry and Biology Received: Revised: Accepted: Published: 2422
October 30, 2016 December 22, 2016 December 25, 2016 December 25, 2016 DOI: 10.1021/acs.jafc.6b04849 J. Agric. Food Chem. 2018, 66, 2422−2431
Article
Journal of Agricultural and Food Chemistry melanoidin formation6 and to investigate the presence and nature of thiol binding sites in raw coffee beans.24,27 Kinetic experiments applying time-resolved sampling during roasting28 as well as online mass spectrometric techniques such as proton transfer reaction mass spectrometry (PTR-MS) or photon ionization mass spectrometry were other approaches used to study coffee aroma formation upon highly dynamic coffee roasting process.29−32 Despite the significant effort devoted, the generation of aroma-active compounds upon coffee roasting is still not fully understood. The in-bean experiments were proven to be very useful in providing insights into formation pathways; however, the former study of Poisson et al.26 revealed also their limitations. Among others, the high cost of 13C-labeled precursors and relatively large batch size (125 g) enabled only partial replacement (17%) of natural sucrose by [UL-13C6fructose]sucrose. This consequently rendered the data interpretation rather complex. In addition, the biomimetic reconstitution of the water-extractable fraction is still a challenge, and the incorporation efficiency of different precursor classes is yet not sufficiently understood. Therefore, the present study aimed at substantiating the role of sucrose in the formation of α-diketones upon coffee roasting by applying an improved extraction and incorporation protocol combined with a down-sized roasting step. This allowed for an entire replacement of the natural sucrose by fully or partially 13 C-labeled analogues and also kinetic studies with labeled sucrose and CAMOLA study.
■
rotation speed (24 rpm agitator, 100 rpm mixer). The obtained green coffee extracts were combined and freeze-dried in a Lyobeta 35 freezedryer (Telstar, Terrassa, Spain) and stored at −40 °C until use. Incorporation of Biomimetically Recombined Extract (Reference Sample, BREB). For the preparation of the biomimetically recombined extract (BRE; based on the composition of water extract of green coffee beans, Table 1), the single components (see Table 2)
Table 1. Composition of the Green Coffee Extract in Percent (Expressed on a Dry Matter Basis) key component
amount (%/DM)
key component
amount (%/DM)
lipids ashes total phenols caffeine trigonelline
0.0 4.1 5.6 0.71 0.13
total organic acids free amino acids free sugars sucrose metals
2.7 0.56 8.1 7.9 0.013
Table 2. Amounts of Components Used for the Preparation of Biomimetically Recombined Extract (BRE)a,b component phenols chlorogenic acid trigonelline caf feine organic acids D-(−)-quinic acid L-(+)-lactic acid potassium acetate malic acid potassium citrate amino acids L-alanine L-threonine L-serine L-aspartic acid L-glutamic acid glycine L-histidine L-arginine L-proline L-tyrosine L-asparagine
MATERIALS AND METHODS
Raw Material. Green coffee beans (Coffea arabica, Costa Rica, La Giorgia, wet processed) were used. Chemicals. The following chemicals were commercially available: caffeine (99.5%), copper(II) sulfate (99%), dichloromethane (99.8%), D-(−)-arabinose (99%), D-(+)-galactose (99%), L-rhamnose (99%), D(+)-mannose (99%), D-(+)-sucrose (99.5%), D-(−)-quinic acid (98%), ethanol (99.5%), iron(III) chloride (97%), L-alanine (99.5%), Larginine (99.5%), L-asparagine (98.5%), L-aspartic acid (99.5%), Lcysteine (99.5%), L-glutamic acid (99.5%), L-glutamine (99.9%), glycine (98.5%), L-histidine (99.5%), L-isoleucine (99.5%), L-(+)-lactic acid (99%), L-leucine (99.5%), L-lysine (98%), L-methionine (99.5%), L-phenylalanine (98%), L-proline (99.5%), L-serine (98.5%), Lthreonine (98.5%), L-tyrosine (99%), L-tryptophane (99.5%), L-valine (99.5%), malic acid (99.5%), manganese(II) chloride (98%), potassium acetate (99%), potassium hydroxide (85%), trigonelline hydrochloride (98%), zinc(II) sulfate heptahydrate (99%), chlorogenic acid (95%), potassium citrate (99%) (Sigma-Aldrich Chemie GmbH, Buchs, Switzerland); [UL-13C6-fructose]sucrose (98%), [UL-13C6glucose]sucrose (98%), [UL-13C12]sucrose (98%), D-[UL-13C5]arabinose (98%), D-[UL-13C6]galactose, D-[UL-13C6]mannose (Omicron Biochemicals, Inc., South Bend, IN, USA). Water Extraction of Green Coffee Beans. Green coffee beans were extracted with hot water by applying the following conditions: Ten kilograms of green coffee beans was mixed with 20 L of water in a Scanima batch mixer (Tetra Pak Scanima, Denmark) and heated to 60 °C for 1 h. The obtained extract was drained, and the coffee beans were extracted another four times with 20 L of demineralized water at 60 °C for a total of 4 h to obtain the water-soluble substances (total 100 L of extract). The resulting extraction yield was found at 20.5% (23.2% on dry matter base). In a second step the exhausted green coffee beans were dried in the Scanima mixer by increasing the temperature of the double-jacket to 110 °C and applying a vacuum of 150 mbar. During the first 2 h of drying, the product temperature was raised from 60 to 80 °C, and then the final temperature was held constant for another 4 h (total 6 h of drying time). To prevent bean breakage, the agitator/mixer was regulated at the lowest possible
amount (mg) 324.7 49.7 41.2 34.6 0.1 3.2 15.9 104.8 1.9 1.6 1.3 4.5 8.3 0.2 0.7 3.0 1.0 0.9 4.0
component amino acids (continued) L-glutamine L-tryptophan L-valine L-isoleucine L-leucine L-phenylalanine L-lysine sugars D-(−)-arabinose L-rhamnose D-(+)-galactose D-(+)-sucrose D-(+)-mannose metals iron(III) chloride manganese(II) chloride copper(II) sulfate zinc(II) sulfate, heptahydrate
amount (mg) 0.6 1.3 0.6 0.4 0.6 1.0 0.7 0.4 0.1 7.9 500.6 1.2 0.3 0.2 0.2 0.03
a
For 5 g of exhausted beans (EB). bComposition of BRE was based on the composition of water extract of green coffee beans.
were dissolved in 2 g of demineralized water at 80 °C. The pH value of BRE was adjusted to 5.5 (corresponding to the pH of the natural extract) with a 16.5% w/w solution of KOH, and water-exhausted green coffee beans (EB, 5 g) were soaked with the BRE at 50 °C for 5 h and then overnight at room temperature. During soaking, the beans were gently stirred using a Rotavapor (Büchi, Switzerland). To improve the incorporation of BRE into EB, 1 mL of demineralized water was added into incorporated coffee beans and absorbed using a Rotavapor (gentle stirring for 1 h at 50 °C and then 5 h at room temperature). After water absorption, the treated beans were washed with the same mass of water as beans during 10 s. The washing losses were controlled by analyzing the washing waters with an ATAGO PAL3 pocket refractometer, which measures the total solid content on a Brix scale. The washed coffee beans were frozen to −80 °C and then freeze-dried for 24 h at 0.1 mbar and about −80 °C on an Alpha 2-4 LSC freeze-dryer (Christ, Germany) to reach a moisture content of 10 2423
DOI: 10.1021/acs.jafc.6b04849 J. Agric. Food Chem. 2018, 66, 2422−2431
Article
Journal of Agricultural and Food Chemistry ± 0.5% (experiment 2 in Table 3). Finally, the beans were roasted using the procedure described below.
Table 4. Roasting Profile for Experiments 1−7 time (s) temp (°C)
Table 3. Biomimetic In-Bean Experiments To Study the Formation of Key Aroma Compounds from Free Sugars upon Coffee Roasting expt
name
1 2
green beans BREB (ref)a
3
BREB + [13C12]-SUCb
4
BREB + [13C6-Fru]-SUC
5
BREB + [13C6-Glc]-SUC
6
BREB + other labeled sugars
7
BREB + CAMOLAc
8
BREB + [13C12]-SUC kinetics
0 148
42 168
95 183
176 196
300 203
Table 5. Roasting Profile for the Kinetic Study (Experiment 8)
description
time (s) temp (°C)
original untreated green coffee beans exhausted bean (EB) + biomimetically recombined extract (BRE) EB + BRE omitted in free sugars spiked with [UL-13C12]sucrose EB + BRE omitted in free sugars spiked with [UL-13C6-fructose]sucrose EB + BRE omitted in free sugars spiked with [UL-13C6-glucose]sucrose EB + BRE omitted in free sugars but sucrose spiked with D-[UL-13C5] arabinose, D-[UL-13C6]galactose, D[UL-13C6]mannose CAMOLA experiment: EB + BRE omitted in free sugars spiked sucrose and [UL-13C12]sucrose (50%/50%) kinetic study: EB + BRE omitted in free sugars spiked with [UL-13C12]sucrose
50 170
80 177
100 184
150 189
210 197
260 200
300 202
330 203
color) and counted to control their number against the initial number of provided beans. Free Sugar Analysis. Two grams of cryo-ground (Kryomill, Retsch, Germany) green beans or water-extracted beans (EB) or 0.5 g of green coffee extract powder was weighed in a 20 mL volumetric flask. After the addition of Milli-Q water (Millipore, Zug, Switzerland), the slurries were incubated for 1 h in a water bath at 100 °C, followed by centrifugation for 5 min at 10000g and filtration through a 0.2 μm filter (VWR International, Dietikon, Switzerland). Further sample cleanup was done by passing 3 mL of extract through a C18 cartridge (Sep-Pak C18, Waters, Montreux-Chailly, Switzerland), which was previously conditioned with 2 mL of methanol and 3 mL of Milli-Q water. Sugars were separated using high-performance anion-exchange chromatography (10 μL injection) on a PA-100 column (Thermo Fisher Scientific, Ecublens, Switzerland), using the ICS-5000 system (Thermo Fisher Scientific) with a constant flow of 1 mL/min and the following gradient: 100% eluent A (Milli-Q water) until 55 min; at 65 min, 75% eluent A and 25% eluent B (1 M NaOH); at 70 min, 50% eluent A, kept until 80 min; final equilibration at 100% eluent A until 95 min. A flow of 0.5 mL/min of 0.3 M NaOH was added postcolumn before the amperometric detector. Sugars were quantified using external calibration curves. Total Coffee Polyphenols. Total coffee polyphenols were measured using Folin−Ciocalteu’s phenol reagent by colorimetric detection according to the methodology described by Georgé et al.33 Free Amino Acids. Free amino acids were analyzed after aqueous extraction of green coffee samples by GC-MS using the Phenomenex EZfaast kit (Brechbuehler, Echallens, Switzerland). Sample preparation was performed as follows: About 3.3 g of cryo-ground green coffee beans or water-extracted beans or corresponding amount of green coffee extract powder (0.75 g) was exactly weighed into a 50 mL volumetric flask and filled with water (Milli-Q). The extraction was performed for 1 h during continuous stirring at ambient temperature. After settling of the solids, the supernatant was filtered using a 0.2 μm pore size syringe filter (SRI), and the amino acids were derivatized. The derivatization was carried out according to the manufacturer’s manual (EZ:faast for free amino acid analysis, Phenomenex). Therefore, a defined amount of the sample extract (100 μL; 50 μL for the abundant amino acids) was spiked with an internal standard solution (norvaline, c = 200 nmol/L). The derivatization was followed by liquid injection and GC-MS analysis. The standard solutions for the calibration curves (concentrations equivalent to 20−300 nmol/L) were spiked with 100 μL of internal standard solution and derivatized in the same way as the samples. The sample (1 μL) was provided to the GC column by liquid injection on split−splitless injector (in split mode; split of 60) at 250 °C. Separation was performed on a 10 m × 0.25 mm Phenomenex Zebron-AAA column, (Brechbuehler) using an Agilent 7890A gas chromatograph (Agilent, Basel, Switzerland). Helium was used as carrier gas with a constant flow of 3 mL/min. The following oven program was applied: initial temperature of 110 °C (0 min), raised to 320 °C at 32 °C/min, and final temperature held for 6.5 min. Mass spectrometry was performed on an inert MSD 5975C quadrupole mass spectrometer (Agilent). Electron impact ionization was applied, and the mass spectrometer was operated in the single ion monitoring mode. Chromatograms were processed using Agilent MassHunter software. Caffeine and Trigonelline. Caffeine and trigonelline contents were determined after aqueous extraction by HPLC-UV similar to the procedure of Casal et al.34
a BREB coffee beans obtained by re-incorporation of biomimetically recombined extract into water-extracted beans. bSucrose (SUC). c Carbohydrate module labeling (CAMOLA).
Labeling Experiments. A similar procedure was applied to prepare green coffee beans spiked with labeled precursors by replacing one or several free sugars in BRE with their labeled analogues. The performed experiments are summarized in Table 3. Labeled Sucrose. The BRE was omitted in all sugars and spiked with [UL-13C12]sucrose (500.6 mg; experiment 3), [UL-13C6fructose]sucrose (500.6 mg; experiment 4), or [UL-13C6-glucose]sucrose (500.6 mg; experiment 5) prior its incorporation into 5 g of EB. The spiking level corresponded to the naturally occurring total sucrose content. Other Labeled Sugars. The BRE was omitted in all sugars except sucrose and spiked with D-[UL-13C5]arabinose (0.4 mg), D-[UL-13C6]galactose (7.9 mg), and D-[UL-13C6]mannose (1.2 mg) prior its incorporation into 5 g of EB (experiment 6). Kinetic Study. For the kinetic study with [UL-13C12]sucrose the setup of experiment 3 was scaled up by a factor of 6 (3004 mg if [UL-13C12]sucrose per 30 g of EB; experiment 8). CAMOLA Experiments. To reach a ratio of 1:1 of unlabeled and labeled sucrose, the level of unlabeled sucrose in BRE was adjusted to compensate for the residual level of sucrose in EB (49.9 mg per 5 g of EB). Consequently, the BRE omitted in all sugars was spiked with [UL-13C12]sucrose (250.3 mg) and unlabeled sucrose (200.4 mg) prior to its incorporation into 5 g of EB (experiment 7). Laboratory Roasting Trials. A standard roasting profile assuring the same roasting conditions (without considering the final color) was defined and applied to all coffee samples. Consequently, the same thermal energy was provided to coffee beans, which allowed for an appropriate comparison among all samples. Temperature and air flow profile were optimized for the roasting of 15 g of green coffee. This profile led to a CTN (color test number, Neuhaus-Neotec, Germany) of 86 for untreated coffee beans (experiment 1 in Table 3). After freeze-drying, 5 g of reincorporated green beans was counted and then mixed with 10 g of nontreated coffee beans to increase the batch size to a critical amount needed for the roasting using a sample roaster. The individual batches (15 g) were roasted under the same conditions on an IKAWA (Ikawa, London, UK) sample roaster for 300 s (experiments 1−7) or for 330 s (experiment 8) until 203 °C. The roasting profiles are given in Tables 4 and 5. The color of the roasted coffee samples was not measured. After roasting, darker colored beans (reconstituted beans) were sorted out from the bulk beans (lighter 2424
DOI: 10.1021/acs.jafc.6b04849 J. Agric. Food Chem. 2018, 66, 2422−2431
Article
Journal of Agricultural and Food Chemistry Total Fat. Total fat content was determined by the Weibull− Berntrop method based on ISO method 8262, involving a treatment with boiling hydrochloric acid.35 Organic Acids. Organic acids were extracted from cryo-ground beans with water at 70 °C for 30 min. After filtration and C18 solid phase extraction (Sep Pak, Waters WAT 020515, Waters, MontreuxChailly, Switzerland), the acids were analyzed by high-performance anion-exchange chromatography conductometry using a Dionex IonPac AS11-HC column (Thermo Fisher Scientific, Reinach, Switzerland). Ash. Total ash content was determined after destruction of the organic matter at 550 °C in a muffle furnace, according to AOAC International method 920.93.36 Metals. Metal content was determined by ICP-OES according to AOAC International method 984.27.36 Aroma Analysis by SPME-GC-MS. Roasted and ground (R&G) coffee (0.5 g) was added into a 20 mL headspace vial, and the sample was equilibrated for 10 min at 40 °C. Aroma compounds were then extracted from the headspace by solid phase microextraction (SPME) at 40 °C during 10 min (2 cm fiber, 50/30 μm StableFlex, coated with PDMS/DVB/Carboxen; Supelco, Buchs, Switzerland) and thermally desorbed into the split−splitless injector (in split mode; split of 2) heated at 240 °C for 10 min. Separation was carried out on a 60 m × 0.25 mm × 0.25 μm polar DB-624 column (Phenomenex, Brechbühler, Switzerland) using an Agilent 7890B gas chromatograph (Agilent). Helium was used as carrier gas with a constant flow of 1.2 mL/min. The following oven program was applied: initial temperature of 40 °C held for 2 min, then raised to 240 °C at 5 °C/min, and final temperature held for 10 min. Mass spectrometry was performed on a 7200 accurate mass Q-TOF mass spectrometer (Agilent). Electron impact ionization was applied, and the mass spectrometer was operated in the full scan mode (m/z 30−250) at a spectra acquisition rate of 5 spectra/s. Chromatograms were processed using Agilent MassHunter software. All results were corrected for the 13C content of the natural isotope. The obtained percentage after correction
