The post-harvest processing of coffee beans leads to a wide range of reactions involving proteins. The formation of crosslinks between proteins and phenolic compounds present in high concentrations of coffee beans represents one of the most challenging and still not fully characterized reactions. The aim of this work was to assess the presence of products from such reactions in coffee samples, focusing on the adducts between cysteine and chlorogenic acids (CQAs). For this purpose, 19 green and 15 roasted coffee samples of the Coffea arabica, Coffea canephora, and Coffea liberica varieties were selected for this study and basically characterized. Then, targeted liquid chromatography mass spectrometry (LC-MS/MS) methods were developed to assess the formation of adducts between CQA and cysteine, glutathione, and N-acetylcysteine as the amino acid and peptide models, and quantified such adducts in coffee samples. The results of the characterization showed a heterogeneous distribution of the protein content (8.7–14.6%), caffeine (0.57–2.62 g/100 g), and antioxidant capacity (2–4.5 g ascorbic acid/100 g) in Arabica, Canephora, and Liberica samples. Glutamic acid, arginine, and proline were found to be the major amino acids, while 5-CQA (38–76%), 3-CQA (4–13%), and 4-CQA (4–13%) were the most abundant CQA derivatives of all coffee varieties. The model experiments for adduct formation demonstrated that cysteine binds to CQA via thiol groups and 5-CQA initially isomerizes to 3- and 4-CQA, depending on the conditions, allowing cysteine to bind to two different sites on 3-, 4- or 5-CQA molecules, thus, forming six different Cys-CQA adducts with m/z 476. The reaction was more favored at pH 9, and the adducts proved to be stable up to 90 °C for 10 min and up to 28 days at room temperature. The relative quantification of adducts showed peak area values ranging from 1100 to 3000 in green coffee bean samples, while no adducts were detected in roasted coffee beans. Overall, this work was the first attempt to demonstrate the presence of Cys-CQA adducts in coffee beans and paves the way for further investigations of such adduct formation at the protein level.

Iron deficiency is a major global health concern, primarily attributable to inadequate dietary intake and limited absorption of iron. Foliar fertilization of vegetables like bell pepper (Capsicum annuum) may increase their iron content. When especially rich in ascorbic acid, intestinal iron absorption might be improved even further. However, iron content in foods and/or bioaccessibility after in vitro digestion alone would be unreliable predictors of iron bioavailability. Consequently, it seems to be necessary to reconsider digestion models and bioavailability evaluation. The objective of this study was to establish a mucin-enhanced in vitro digestion model to assess the bioavailability of non-heme iron from food matrices in combination with the widely utilized Caco-2 model of intestinal iron absorption. Compared with Fe(III), incubation with 20 µM and 200 µM Fe(II) sulfate increased ferritin formation (normalized to total protein (TP)) in differentiated Caco-2 cells by 80% and 130%, respectively. Furthermore, no loss of cellular viability was observed across the tested Fe(II) and Fe(III) concentrations (20–2000 µM). Three in vitro digestion models (DIN, DIN-lite, and DIN+G), differing mainly in digestive enzyme content, were evaluated for iron bioaccessibility and bioavailability. Only DIN-lite and DIN+G were compatible with Caco-2 cells. Although DIN-lite yielded 25% higher bioaccessible iron than DIN+G, both models resulted in comparable ferritin formation in Caco-2 cells. The DIN+G/Caco-2 model was applied to bell pepper cultivars (‘Ferrari’, ‘Morbidelli’, and ‘Jack Miller’), treated with foliar Fe(II) sprays during cultivation, achieving up to 3.9-fold increased iron content. However, this increase did not translate into enhanced in vitro iron bioavailability in the bell pepper pericarp. Consistent with previous studies, these findings indicate that iron content and bioaccessibility alone are insufficient predictors of iron bioavailability in plant-based foods. At the same time, the mucin-enhanced DIN+G/Caco-2 model proved to be a suitable approach for investigating iron bioavailability in plants.