*2.7. Determination of Chelating Activity on Fe2*<sup>+</sup>

Chelating activity of the HMW isolates from raw and roasted cocoa beans was evaluated using the method of Gu et al. [24]. Briefly, 1 mL of the suitably diluted lyophilized HMW cocoa materials in high-purity deionized water (2 mg/mL) were added to 1.85 mL of high-purity deionized water and 50 μL of 2.0 mM FeCl2. After mixing, the solution was allowed to stand at room temperature for 30 s, followed by the addition of 100 μL 5 mM ferrozine. The reaction mixture was then vortexed and left to stand at room temperature for 15 min. The absorbance of the solution was measured spectrophotometrically at 562 nm with UV-1800 UV-VIS Spectrophotometer (Shimadzu, Tokyo, Japan). Worth noting, that low absorbance of the resulting solution indicated a strong ferrous ion chelating ability. A reaction mixture containing 1 mL of high-purity deionized water instead of sample solution served as the blank. The results are expressed as the percentage chelating activity (%).

#### *2.8. Attenuated Total Reflection Fourier Transform Infrared Spectroscopy*

The infrared spectroscopy is based on the absorption of radiation due to vibrations bonds of molecules [25,26]. The FTIR spectra of lyophilized HMW cocoa melanoidin fractions were obtained using an infrared Fourier transform spectrometer, model IRTracer-100 (Shimadzu, Tokyo, Japan) equipped with an attenuated total reflection (GladiATR) accessory with diamond crystal (PIKE Technologies, Inc., Madison, Wisconsin, USA) at room temperature. The spectral range was 400–4000 cm−<sup>1</sup> with 40 scans and a resolution of 4 cm−1. Around 5 mg of lyophilized HMW cocoa fractions was deposited on the diamond platform prior to measurement. Background and sample spectra were acquired at 4 cm−<sup>1</sup> resolution with 40 scans from 400 to 4000 cm−<sup>1</sup>

#### *2.9. Statistical Analysis*

The results are presented as mean ± standard deviations of three replicates. Statistical tests were evaluated by using the Statistica 13.0 software (StatSoft, Inc., Tulsa, OK, USA). The obtained data were tested for normal distribution (Shapiro-Wilk test) and equal variances (Levene's test). As all values showed normal distribution and homogeneity of variance, considering *p* ≥ 0.05, the data were subjected to the analysis of variance (ANOVA). The effects of variety, roasting temperature, roasting air humidity, and their interaction on total phenolic content and antioxidant activity of HMW melanoidin fractions were tested by means of two-way ANOVA. The significant differences among the means were estimated through Tukey's HSD test. For all statistical analysis, *p* < 0.05 was considered as statistical significance. The error bars in all figures correspond to the standard deviations. The correlation coefficients between investigated parameters were assessed by means of the Pearson correlation test using Microsoft Office Excel 2016 (Microsoft Corporation, Redmond, WA, USA).

#### **3. Results and Discussion**

The present study is a continuation of our previous work [19,20] exploring the physicochemical properties and the profiles of free and bound phenolics of HMW melanoidin fractions of different *T. cacao* groups and origins. In this work, the total phenolic content and the antioxidant activity of the HMW melanoidin fractions (>12 kDa) from raw and roasted, at different temperatures and relative air humidities, cocoa beans of different groups were evaluated by using different in vitro spectrophotometric assays.

#### *3.1. Total Phenolics Content*

The total phenolics content (TPC) of HMW melanoidin fractions from the three cocoa types is shown in Figure 1. The data were expressed as milligrams of gallic acid equivalents (GAE) per gram dry weight (mg GAE/g dw). In this study, significant differences (*p* < 0.05) in TPC of the HMW fractions were found between the three cocoa types. The HMW fractions from both raw and roasted Criollo beans were characterized by the highest TPC, ranging from 121.99 to 149.90 mg GAE/g DW. Less phenolics were observed in HMW isolates from Forastero beans (117.46–139.12 mg GAE/g DW), while HMW materials from Trinitario beans contained the lowest amounts of these compounds (93.02–129.00 mg GAE/g DW). In order to assess the influence of cocoa variety, roasting temperature, roasting air humidity, and their interactions on TPC content of HMW melanoidin fractions, a two-way ANOVA was carried out. The results showed that the cocoa variety and roasting process parameters significantly affected the TPC of HMW melanoidins isolated from the studied cocoa beans (*p* < 0.01). Generally, more phenolics were contained in the almost all of HMW fractions isolated from roasted beans than from the raw ones. The greatest increase in TPC (by 25.3–38.1% of initial value) of the HMW fractions was observed when cocoa beans of the Trinitario type were roasted at 110 ◦C. Moreover, it was also noticed that thermal processing of Criollo and Forastero beans at 150 ◦C with RH of 0.3% led to the highest increase in the TPC (by 18.4–22.9% of initial value). As can be seen from the results, the changes in the TPC of the HMW fractions also depended on the RH and in almost all samples were slightly

less advanced when the air humidity was increased from 0.3% to 5.0%. Thus, it was confirmed that influence of the roasting parameters on TPC of the HMW fractions is complex, and huge differences are observed depending on applied cocoa bean variety and process conditions, like temperature and relative air humidity level. According to Pérez-Martínez et al. [27], the Folin–Ciocalteau assay measures the reducing capacity of a sample, and the results of TPC may be affected by the presence of other electron donors, such as non-phenolic substances and nitrogen-containing compounds. Thus, the increase in the level of compounds able to react with the Folin–Ciocalteu reagent could be linked to the formation of new substances, especially reductones, during the roasting of cocoa beans. These compounds, which are formed in the advanced and final stages of Maillard reaction, can act as reducing agents, due to the presence of hydroxyl and pyrrole groups [24,28].

**Figure 1.** Total phenolics content of melanoidin fractions isolated from raw and roasted, at different at different temperatures and relative air humidities, cocoa beans of different groups. Results are presented as means ± SD from triplicate assays. Bars with the same lowercase letter (a–g) within each variety do not differ significantly according to Tukey's HSD test at *p* < 0.05.

Summa et al. [16] reported that pre-roasting of cocoa beans (at 80–90 ◦C for 10 min) significantly reduced the concentration of Folin–Ciocalteu reactive substances in the 30–10 kDa fraction while accurate roasting (at 130–160 ◦C for 15–20 min) increased their content. As suggested by these authors, this phenomenon might be caused by formation of reducing substances in the 30–10 kDa fraction during roasting at high temperatures. They also found that the level of reducing compounds in the >30 kDa fraction increased significantly during cocoa beans pre-roasting but declined markedly after accurate roasting. In addition, the observed behavior can be a consequence of the incorporation of phenolic compounds into the structure of melanoidins during heat treatment [15,29]. The oxidized polyphenols may react with thiol groups on amino acids, peptides, and proteins or nucleophilic amines via Michael-type addition or Schiff base reaction to form thiol−quinone and amine–quinone adducts or benzoquinone imines [15,30–32]. Our previous studies have shown that roasting of cocoa beans of different types at temperatures ranging from 110 to 150 ◦C led to a marked increase in the bound phenolics content as compared to HMW fractions from unroasted beans [20]. Therefore, we can concluded that significant differences in the TPC observed between the HMW cocoa melanoidin fractions appeared to be brought about by the structural changes of their active components, involving oxidation of phenolic compounds, their condensation with proteins and polysaccharides, as well as

cross-linking and polymerization of low molecular weight MRPs during roasting, which decide their electron donating abilities [20,31,33].

#### *3.2. Free Radicals-Scavenging Capacity*

To make a comprehensive evaluation on the antioxidant effect of samples, it is necessary to employ different methods due to the fact that this assays exhibit different mechanisms of by which its antioxidant activity takes place. In vitro free radical scavenging activities of the HMW melanoidin fractions from raw and roasted at various conditions cocoa beans of different *T. cacao* groups and origins were assessed against DPPH and ABTS radicals (Table 2).

**Table 2.** The free radical scavenging capacity (2,2-diphenyl-1-(2,4,6-trinitrophenyl) hydrazyl (DPPH) and 2,2 -azino-bis (3-ethylbenzothiazoline-6-sulfonic acid) diammonium salt (ABTS)) and ferric reducing ability (FRAP) of high-molecular weight (HMW) fractions isolated from raw and roasted at different conditions cocoa beans of Criollo, Forastero, and Trinitario groups originating from various geographical regions.


T, temperature. RH, relative air humidities. Data are presented as mean ± SD of three replications. The means followed by the same lowercase letter (a–h) within each variety in the same column do not differ significantly according to Tukey's HSD test at *p* < 0.05. Significance: \* *p* < 0.05; \*\* *p* < 0.01; \*\*\* *p* < 0.001; ns: not significant. DPPH, the DPPH radical-scavenging capacity expressed in μmol Trolox equivalents per gram of HMW fraction dry weight (μmol TE/g DW); ABTS, the ABTS radical-scavenging activity expressed in in μmol Trolox equivalents per gram of HMW fraction dry weight (μmol TE/g DW). FRAP, the ferric reducing antioxidant power expressed in μmol Trolox equivalents per gram of HMW fraction dry weight (μmol TE/g DW).

The ABTS•<sup>+</sup> and DPPH assays are widely used methods for the assessment of the antioxidant activities of many vegetable or food matrices [34–37]. These methods are both based on quenching of stable colored radicals and show the free radical quenching activity of antioxidants even when present in complex biological matrices such as plant or food preparations (extracts or fractions). The values of antioxidant potential obtained in both assays were equal or higher than those reported by other authors for the HMW melanoidin fractions of cocoa beans, chocolate, or even coffee [16,23,38], meaning that all of studied HMW cocoa melanoidin samples exhibited good DPPH and ABTS radical scavenging activities.

However, there were significant differences between HMW melanoidin fractions from both the raw and roasted cocoa beans regarding their antioxidant abilities. Two-way ANOVA revealed significant effect of cocoa variety (*p* < 0.01), roasting temperature (*p* < 0.001), roasting air humidity (*p* < 0.01), and interaction of variety and roasting conditions (*p* < 0.001) for the DPPH radical-scavenging activity (Table 2). In addition, ABTS scavenging capacity were significantly affected by variety, roasting temperature, and interaction of variety and roasting conditions, but the levels of relative humidity did not have significant influence. The HMW materials from unroasted Criollo beans demonstrated the highest scavenging capacity against both DPPH and ABTS radicals (726.26 and 849.84 μmol TE/g dw, respectively). Compared with HMW melanoidins from Criollo beans, the HMW fraction from unroasted Forastero beans showed the significantly lower (*p* < 0.05) activity against DPPH (689.16 μmol TE/g dw) but the similar against ABTS radical cations (806.22 μmol TE/g dw). The lowest radical scavenging abilities against the ABTS and DPPH assays was exhibited by HMW isolates from unroasted Trinitario beans, which is consistent with the results of TPCs discussed above. These results showed that, depending on the cocoa type and roasting conditions, an increase or decrease in the antioxidant capacity of HMW melanoidin fractions was observed. Roasting of all types of cocoa beans at temperatures ranging from 110 to 150 ◦C caused a significant increase (*p* < 0.05) in the DPPH radical-scavenging activity of HMW fractions. However, the changes in the free radical-scavenging activity of HMW fractions were considerably smaller at the higher RH. Interestingly, we observed that HMW fractions from Criollo beans roasted at 150 ◦C with lower RH exhibited a higher antioxidant activity by DPPH scavenging assay (1475.61 μmol TE/g dw) compared to all other melanoidin samples. Roasting of Forastero and Trinitario beans at temperatures ranging from 110 to 150 ◦C caused a significant increase (*p* < 0.05) in the ABTS radical-scavenging activity of their HMW melanoidin fractions. It was also observed that the HMW fractions from Criollo beans, roasted at temperatures of 110, 120, and 150 ◦C, displayed the significantly increased ABTS scavenging capacity, while after roasting at 135 ◦C, exhibited a slightly lower antioxidant activity by this method. Nevertheless, among HMW melanoidin fractions isolated from roasted beans of the three cocoa varieties, the highest ABTS radical-scavenging capacity (1172.37 μmol TE/g dw) was exhibited by the fraction from Criollo beans roasted at 120 ◦C and RH of 0.3%. Based on the results presented in Table 2, it is possible to observe that, in almost all HMW fractions, antioxidant activity is higher in the DPPH method (except HMW fractions of Criollo beans roasted at 110 ◦C). This behavior is similar to the observed by other researchers [23] who have observed a stronger scavenging activity against DPPH• than ABTS•<sup>+</sup> radicals in the case of chocolate melanoidins. Our results corroborated also with those reported by Summa et al. [16], who showed that the >30 kDa fraction obtained from pre-roasted cocoa beans had higher DPPH and ABTS radical-scavenging activity compared to the >30 kDa fractions of raw and roasted beans. The differences between the ABTS and DPPH radical-scavenging activity could be explained by the wide variety of chemical mechanisms involved in the antioxidant activity of HMW cocoa melanoidin fractions, as it was mentioned earlier. The reaction mechanism with DPPH• radical involve transfer of a hydrogen atom, while the reactions with ABTS•<sup>+</sup> radicals involve electron transfer process [39]. The observed differences indicate the complexity of the mechanism of action of melanoidins formed in real food. The increase in the antioxidant capacity of the HMW fractions can be a consequence of accumulation of high molecular weight MRPs-like melanoidins during roasting. In addition, the observed behavior can be explained by the presence of residues of certain active compounds,

containing more than one active group (OH or NH2), such as phenolic compounds, quinones, and low molecular weight MRPs, in the HMW materials. These compounds might be attached to the structure of melanoidins via non-covalent bonds and influence their biological properties [4,5,18]. Although the content of free phenolic compounds in HMW fractions of roasted cocoa beans decreased [19], it is possible that it could be explained by the presence of quinones generated by oxidation of these compounds that spontaneously form covalent bonds to functional groups of melanoidins during roasting [4,40]. We suppose that the radical-scavenging activity of tested HMW fractions from cocoa beans depends on the structure and the number of the included active group (OH or NH2). This was a clear indication that the synergistic effect between different bioactive compounds present in the structure of melanoidins could determine their biological properties. Since, the HMW cocoa melanoidins have the ability to scavenge free radicals, thereby preventing lipid peroxidation chain reactions that cause damage a wide range of molecules found in living cells, they could serve as potential nutraceuticals and functional ingredients.

#### *3.3. Ferric Reducing Antioxidant Power*

The reducing capacity may serve as a significant indicator of potential antioxidant activity. Thus, the HMW melanoidin fractions of unroasted and roasted cocoa beans of tree different *T. cacao* beans was estimated for their ability to reduce TPTZ–Fe (III) complex to TPTZ–Fe (II). In order to identify the influence of cocoa variety, roasting temperature, roasting air humidity, and their interactions on ferric reducing antioxidant power (FRAP), a two-way ANOVA was performed. As demonstrated in Table 2, significant differences (*p* < 0.05) in the reducing properties of HMW fractions were found between the three cocoa types. The HMW fractions from Criollo beans exhibited the highest FRAP values, ranging from 622.42 to 1077.18 μmol TE/g dw, followed by isolates of Forastero beans (533.90–947.46 μmol TE/g dw), while HMW materials of Trinitario beans had the lowest reducing power (440.49–779.49 μmol TE/g dw). The results also showed that the roasting process parameters significantly (*p* < 0.05) affected the reducing properties of HMW materials isolated from the studied cocoa beans. Generally, all of HMW fractions isolated from unroasted beans showed a lesser reducing power compared to the isolates from the roasted samples. The greatest increase in the ferric reducing ability (by 71.1%–77.0% of initial value) of HMW fractions was observed when cocoa beans of all studied cocoa groups were roasted at 135 ◦C with RH of 0.3%. As can be seen from the results, the changes in the reducing power of HMW fractions depended on the RH and in almost all samples was slightly more pronounced when the air humidity was decreased from 5.0% to 0.3%. We can conclude that all studied HMW cocoa melanoidin fractions behaved as reductants and inactivators of oxidants due to the presence of electron-donors in their structure. The observed increase in the reducing capacity of HMW cocoa fractions could be linked to the formation of new substances able to donate electrons or to terminate radical chain reactions, during roasting of cocoa beans. As described above, the HMW cocoa melanoidin fractions contained different amounts of free and bound phenolic compounds and the presence of these polyphenols along with compounds formed in the advanced and final stages of Maillard reaction could contribute to the antioxidant properties observed by FRAP method. These residues due to the presence of hydroxyl and pyrrole groups can act as reducing agents via their redox potential of transferring electrons [24,33]. In a study performed by Summa et al. [16], it was observed that roasting of cocoa beans (at 130–160 ◦C for 15–20 min) significantly increased the concentration of reducing substances in the 30–10 kDa fraction. They also found that the level of reducing compounds in the >30 kDa fraction increased significantly during cocoa beans pre-roasting but declined markedly after roasting.

#### *3.4. The Weighted Average Antioxidant Capacity*

Our results suggest that the antioxidant capacity of HMW fractions from raw and roasted beans depend on the cocoa type and roasting conditions, wherein no significant correlation was observed between DPPH, ABTS, and FRAP values. This may be explained by the fact that different constituents in the HMW fractions would have different mechanisms of action regarding their antioxidant activities. According Tabart et al. [41], the weighted average of the results obtained by the different assays should be calculated to get an overall impression of the antioxidant potential of the samples. Therefore, the results of antioxidant capacity of HMW fraction obtained by the specific assay (DPPH, ABTS, and FRAP) was divided by the average activity of the all HMW samples by the same assay, and then the calculated values in each assay were summarized and divided by the number of assays used (three in our case). The weighted average antioxidant capacity of each HMW samples are shown in Figure 2. The results of the present study show that cocoa beans roasting temperatures of 110–150 ◦C cause significant rise in the weighted average antioxidant capacity (WAAC) of the HMW materials. The highest WAAC values were obtained at a roasting temperature of 150 ◦C and RH of 0.3% (in the case of Criollo and Forastero groups) and a roasting temperature of 120 ◦C and RH of 0.3% (in the case of Trinitario group). It was also observed that the HMW fractions of both raw and roasted Criollo beans exhibited the highest WAAC values. The HMW fractions derived from almost all beans of Forastero type has a slightly higher WAAC compared to the samples of Trinitario type (except for beans roasted at 110 ◦C and RH of 5.0%). This is reflected in the results of the two-way ANOVA (Figure 2) indicating that WAAC was significantly affected by the cocoa variety (*p* < 0.001), roasting temperature (*p* < 0.001), roasting air humidity (*p* < 0.01), and their interactions (*p* < 0.001).

**Figure 2.** The weighted average antioxidant capacity of HMW fractions isolated from raw and roasted, at different at different temperatures and relative air humidities, cocoa beans of different groups. Results are presented as means ± SD from triplicate assays. Bars with the same lowercase letter (a–f) within each variety do not differ significantly according to Tukey's HSD test at *p* < 0.05.
