Various varieties of apples (Boskoop, Granny, Pinova, Elstar, Cox, Jonagold and Jonagored) were used for the production of Liège syrup by Siroperie Meurens (Aubel, Belgium). Sampling was done at various stages:

Homogenate: after the washing and the grinding of the apples (with skin, seeds…)

Juice: after the first cooking (3 h at 90°-95°C) and filtration (filter press) of the homogenate

Concentrate: after concentration of the juice with a concentrator (120°-140°C for 40 min)

Syrup: after mixing concentrates of various fruits at 65°C and heating at 105°C during 20 min. Sugar can be added during mixing.

The samples were diluted in water and centrifuged at 10000 g for 10 min before the various measurements.

In in vitro model systems, the heating procedure was modified from Ajandouz et al.17. An equimolar (100 mM) mixture of one amino acid and glucose or fructose (1 mL), without pH adjustment was heated in a 10 mL screw-sealed tubes in boiling water (100°C) for 20, 60, 120 or 180 min or autoclaved at 120°C for 20 min. The tubes were then cooled down on ice and stored at 4°C. Various pH were also tested at 100°C for 180 min. The following buffers were used: 0.1 M sodium acetate adjusted to pH 3 or 5 with 0.1 M acetic acid, 0.1 M sodium phosphate adjusted to pH 7 or 9 with 0.1 M hydrochloric acid, 0.1 M sodium phosphate adjusted to pH 11 with 0.1 M sodium hydroxide.

Marmalades (lab preparations) were prepared by mixing and crushing fruits (10 g of apple or pear) with sucrose (6 g) and distilled water (6 mL). The mixture was heated in 50 ml disposable plastic tubes for 60 min in boiling water. In some experiments, the heating was extended to 120 min or the mixture was buffered to pH 5 or 9. In one experiment, sucrose was replaced by fructose (5 g instead of 6 g because the sweetness of fructose is higher than this of sucrose).

Total phenolics were determined according to the Folin-Ciocalteu method described by Mihalache Arion et al.18. In a 96-well microplate, 200 µL of water, sample (appropriately diluted), or standard were mixed with 100 µL of Folin-Ciocalteu reagent (10%). After 3 minutes, 80 µL sodium carbonate solution (7.5% w/v) were added. The plate was incubated at 30°C for 1 h in a microplate reader (Multiskan Ascent, ThermoLabsystems, Finland). After incubation, the absorbance at 750 nm was measured. Gallic acid (GA) was used as standard, and results are expressed in mg gallic acid equivalents (GAE) per mL. All assays were done in duplicate.

The antioxidant capacity was first determined by scavenging of the 2,2-diphenyl-1-picryhydrazyl (DPPH) radical as described by Sipel et al.19. The stock solution was prepared by stirring 7.5 mg DPPH in 100 mL methanol overnight. In the assay, 100 µL of extract, standard (0–100 µM Trolox), or blank (methanol) and 200 µL DPPH solution were mixed in a well of a 96-well microplate. The absorbance of samples, standards, and blanks at 520 nm was determined after 5 min of incubation in a Multiskan Ascent reader (ThermoLabsystems, Finland) at 30°C.

The antioxidant capacity was also determined by the ORAC assay as described by Kevers et al.20. Briefly, AAPH was used as a peroxyl radical generator, Trolox as a standard, and fluorescein as a fluorescent probe. 25 µl of diluted sample, blank, or Trolox calibration solution (0–100 µM) were mixed with 150 µl of 4 µM fluorescein and incubated for 15 min at 37°C before addition of 25 µl AAPH solution (173 mM). The fluorescence was measured every 2 min for 4 h on a Victor 3 plate reader (Perkin Elmer) at 37°C. Filters were used to select an excitation wavelength of 485 nm and an emission wavelength of 520 nm. All samples were analysed in duplicate at three different dilutions. The final ORAC values were calculated from the net area under the decay curves.

The results obtained with both assays are expressed in µmol Trolox equivalents (TE) per mL.

In a 96-well microplate, 50 µL of water, sample (appropriately diluted), or standard (HMF) were mixed with 125 µL of p-toluidine solution (10% in isopropanol). 25 µL of barbituric acid (0.5%) were added. The absorbance at 550 nm was measured with a microplate reader (Multiskan Ascent, ThermoLabsystems, Finland). Results are expressed in µg HMF per mL.

Even if this method is not specific (it can also detect the presence of aldehydes other than HMF), it allows evaluating the modifications of HMF in in vitro reactions where there is no interference with other compounds.

Melanoidin formation was evaluated by measurement of OD at 405 nm of 150 µL samples (triplicates) in a 96-well microplate according to Echavarria et al.22. Results are expressed in OD of the sample without dilution.

The assay was done with the Thermo scientific Pierce BCA protein assay kit according to the manufacturing instructions. Albumin was used as standard.

All results presented are means (±SEM) of three independent experiments except for table 1 for which five experiments were done.

The data were compared by ANOVA to evaluate the significant differences between samples using Tukey HSD’s post test P<0.05.

During the process, the fruits were first washed and grinded to obtain a homogenate. Then this homogenate was cooked at 90-95°C for 3 h and then filtered. The juice so obtained was concentrated at around 130°C for 40 min. and finally, the Liège syrup was obtained after cooling and further heating to 105°C. Thus this process includes at least 4 hours of heating at temperatures between 90°C and 140°C.

The antioxidant capacity (ORAC) and the total phenolic content were higher in juice than in homogenate (table 1). The process of cooking was probably responsible of this increase23 because cooking induces thermal inactivation of oxidative enzymes and /or the destruction of cell walls and subcellular compartments that causes the release of antioxidant compounds as phenolics24. The heating of the juice induced another increase of total phenolics but no modification (ORAC) or a decrease of the antioxidant capacity (DPPH). The decrease was confirmed in the syrup for these three parameters. The final values of antioxidant capacity (ORAC) and total phenolics were similar to the values of the homogenate. Finally, all the process of syrup preparation did not decrease the antioxidant capacity nor the phenolic content. Correlations were already observed between total phenolics and antioxidant capacity suggesting that phenolic compounds were partially responsible of the antioxidant capacity1820.

The proteins present in the homogenate drastically decreased during the concentrate preparation. Proteins are easily denatured at high temperature and partially hydrolysed in acidic conditions. The released amino acid can then be used in the Maillard reaction. Indeed, in parallel, HMF appeared in the concentrate while in syrup, the final product, the content of this compound was very low. HMF is a furanic compound which forms as an intermediate in the Maillard reaction13. The amount of HMF detectable in foods is directly related to the heat load applied during processing of carbohydrate-rich products14. In Liège syrup, the protein content was very low and the HMF had almost totally disappeared. We can suppose that the Maillard reaction has continued during syrup mixing and heating and that melanoidins were formed from the HMF. These compounds as other MRPs are known to have an antioxidant capacity411 and are probably responsible of the deep brown color of the Liège syrup and of its taste and smell.

Aiming at understanding how antioxidant capacity was modified and how the content in HMF and melanoidins varied during the process, in vitro model systems were studied. Each consisted in mixing one amino acid with either glucose or sucrose in equimolar concentration (100 mM). The different mixtures were heated at 100°C during 180 min.

Development of antioxidant activity due to MRPs can be influenced by the characteristics of the food matrix. Therefore we also analysed the antioxidant capacity (DPPH and ORAC), the HMF content and absorbance at 405 nm during apple or pear marmalade cooking. According to the previous results, the cooking conditions used were: 100°C during 60 and 120 min in non buffered condition, or during 60 min at pH 5 or pH 9, or during 60 min with fructose added instead of sucrose).

Cooking increased both the DPPH antioxidant capacity (figure 4A) and ORAC values (data not shown) of apple and pear marmalade’s. Doubling the heating time from 60 to 120 min in non buffered conditions led to an increase of HMF content (figure 4B). Absorbance at 405 nm (indicative of melanoidin content) increased also during the first hour (figure 1C), but was stable during the following hour for pear marmalade. The slight decrease observed with apples indicated that there was probably an aggregation of particles of melanoidin with time at high temperature.This behavior is quite different from what was observed in amino acids / sugar in vitro model systems. However melanoidin composition in real foods is probably more complex than in model systems due to the more diverse pool of reactants.

When the mixture was buffered to pH 5, the antioxidant capacity after 60 min of heating was similar to that observed in non buffered conditions (figure 4A). But at pH 9, an increase of the antioxidant capacity was observed for apples and pears. Such an increase was already observed in this study in model aminoacid – sugar mixtures. On another hand, shifting the pH from 5 to 9 led to an increase of the HMF content in apple but not in pear marmalades. The melanoidins (OD at 405 nm) increased with the increase of pH from 5 to 9 in apple marmalade’s, but was stable in pear marmalade’s. Although melanoidins are chemically diverse, many studies reported that they are negatively charged in both real foods and in in vitro model systems at neutral pH 9. Under these conditions, the type of amino acid present during the reaction determined the anionic properties of the melanoidins30. The antioxidant properties of melanoidins have been partly ascribed to the metal chelating capacity of these compounds. This can explain the difference observed between apple and pear marmalade’s.

The replacement of sucrose by fructose induced an increase of the antioxidant capacity in apple marmalade. Reducing sugars in the fruit puree, mainly glucose and fructose, participate directly in the non enzymatic browning reactions while some disaccharides, such as sucrose are less reactive because they must be first hydrolyzed during thermal treatment, leading to the formation of glucose and fructose31. This fact can explain the rapidity of Maillard reaction in the presence of fructose.

The results of this study may have some useful practical implications at both the food technology and nutritional levels. Melanoidins are produced during the processing and storage of foods. The antioxidant properties of melanoidins can inhibit the oxidation of unsaturated lipids and functional food ingredients, such as vitamins, polyphenols and flavonoids. Moreover, their antimicrobial activity can inhibit the growth of microorganisms32 and prevent the spoilage and deterioration of foods. Furthermore, at the end of the Maillard reaction, the volatile aromatic compounds contribute directly to the attractiveness of the products. Consequently, the Maillard reaction has both desirable and undesirable effects on products33. Hence, researchers should optimize the formation of these components.

Considering that melanoidins may preserve the quality and safety of foods10, some parameters of the processing could be modified:

the cooking times and temperature,

the pH of the preparation,

the relative proportion of fruits or the addition of amino acids and reducing sugars responsible of the formation of the MRPs (reducing sugars are more rapidly transformed in MRPs than sucrose),

the addition of plant extract containing polyphenolics. The composition in polyphenols themselves responsible of antioxidant capacity can be an important factor that affects MRP formation434.

Concerning the cooking time and temperature, we have also to take in account that the degradation of ascorbic acid in food is one of the major sources of furan compounds3536. In our in vitro model, we have evidenced that an equimolar mixture of ascorbic acid with amino - acids is associated with a dramatic increase of HMF (data not shown).

An increase of health properties of cooked fruits is associated with a higher antioxidant capacity, a decrease in HMF and an increase of melanoidins. Some phenolic compounds and plant extracts containing phenolic compounds could be used to prevent the formation of some MRPs before thermal process applications or during storage37.