White lupin (Lupinus albus) seeds from the (intentionally left blank for review), were supplied by a local vendor in sealed plastic bags. Prior to cooking, the seeds were stored in a light-resistant container at room temperature (22–24 °C) in a dry environment. The dried seeds were sorted and cleaned three times to remove foreign material as well as immature, and damaged seeds. Only visually intact and uniform seeds were included to ensure sample standardization. Cooking procedures were performed in a domestic setting, and the prepared samples were subsequently stored at +4 °C. To remove naturally occurring alkaloids, which may hinder widespread consumption and impart a bitter taste, thereby simulating traditional domestic preparation methods, the seeds were soaked in tap water for 5 days, with the water replaced every 12 h. Sodium bicarbonate (NaHCO₃, 10 g/L) and sodium chloride (NaCl, 10 g/L) were added to the soaking water (14, 15). This initial soaking step, essential for making the lupin seeds palatable and safe for human consumption, is primarily an alkaloid removal process and is not expected to significantly alter the major protein structure. Subsequently different cooking techniques were applied. Cooked lupine samples were placed in airtight, transparent glass containers and transported to the laboratory in insulated boxes with ice packs (16, 17). The following cooking methods were applied to the L. albus samples (18).

In this study, specific processing conditions were selected to represent typical household practices and standard biotechnological protocols. The aim was to compare the fundamental effects of these distinct processing mechanisms on the allergenic profile of lupin, rather than to optimize individual processing parameters (e.g., time-temperature combinations). Boiling: Samples were cooked in boiling water at 100 °C for 60 min under controlled conditions, with time and temperature monitored using a thermometer. At the end of the process, cooking was halted by rinsing the samples with ice-cold water. Microwaving: Soaked lupin samples were placed in a glass container, covered with plastic wrap, and microwaved at 900 W for 30 min, with inspection at 10-min intervals. Fermenting: Soaked lupin samples were boiled in water at 100 °C for 30 min. Subsequently, 50 g of boiled lupins were added to 200 mL of hot water containing 5 g of rock salt. Lactic acid fermentation was then performed using Lactobacillus brevis in a whey-based culture medium at 43 °C for 24 h in the dark, protected from sunlight (19). Enzyming: The soaked lupin samples were boiled in water at 100 °C for 30 min, then subjected to treatment with 5 g of a commercial powdered bromelain preparation (800 GDU) (20, 21). Samples were periodically tested for softness by gentle manual pressure. Cooking was considered complete, and samples were placed in storage, once a uniformly soft texture was achieved (22). Figure 1 shows images of the cooked samples.

A laboratory-based system was established using organic and inorganic chemicals and enzymes to simulate the mouth, stomach, and small intestine phases of the human digestive process. The procedure was based on a modified version of the INFOGEST protocol, as described by Yaman and Mızrak (23, 24). Enzyme solutions were prepared based on the specific activities provided by the manufacturer to achieve the required enzymatic activities in the final digestion mixtures. Samples were numbered, and in parallel, a blank solution without any sample was processed through all experimental steps. After each measurement, the pH meter probe and the magnetic stir bar used for mixing were rinsed thoroughly with distilled water. During intervals between experiments, mixtures in the beakers were covered with Parafilm and stored at +4 °C. A schematic representation of the digestion steps is provided in Figure 2.

For the oral phase, 2 g of sample was placed into a 10 mL Falcon tube and 2 mL of simulated saliva solution containing α-amylase from porcine pancreas (Sigma A3176, Type VI-B, ≥ 500 units/mg protein) was added. The volume was adjusted to 10 mL with distilled water, and the mixture was homogenized for 20 s. Samples were then incubated for 5 min at 37 °C in a shaking water bath, in sealed tubes (pH 7.0 ± 0.2) (23). For the gastric phase, 4.8 mL of simulated gastric solution containing pepsin (Sigma P7012, derived from porcine gastric mucosa, activity ≥ 2,500 units/mg protein) was added to the mixture from the oral phase. The mixture was homogenized for 20 s and incubated for 2 h at 37 °C in a shaking water bath, in sealed tubes (pH 3.0 ± 0.2). For the intestinal phase, 4 mL of simulated intestinal fluid containing pancreatin (Sigma P7545, from porcine pancreas, 8 × USP specifications). Based on the manufacturer’s specification (equivalent to ≥ 200 USP protease units/mg), the pancreatin solution was standardized to achieve a trypsin activity of 100 U/mL in the final digestion mixture and 2 mL of bile solution containing bile salts (Sigma B8631, bovine) were added to the mixture from the gastric phase. The samples were incubated in a shaking water bath at 37 °C for 2 h (pH 7.50 ± 0.2) (25, 26).

To compare the physiological INFOGEST model against a non-physiological, simultaneous enzymatic degradation, a standard protein hydrolysis method was applied as described (27). It should be noted that this method was not intended to simulate human digestion but served as a technical control representing an aggressive, single-step enzymatic hydrolysis often used in proteomic sample preparation. The enzyme solutions were prepared in 100 mM Tris-HCl buffer at pH 8.0. Given that 100 g of lupin contains approximately 38 g of protein, each 2 g sample contained about 0.76 g of protein. Two grams of sample were weighed and transferred into 10 mL falcon tubes. Pepsin (enzyme-to-substrate ratio 1:200; 152 mg) (28); chymotrypsin (1:100; 76 mg); and trypsin (1:100; 76 mg) (29) enzymes were dissolved in 50 mL distilled water. Although pepsin is known to be inactive at alkaline pH, it was included in the mixture to strictly adhere to the reference protocol (27) and to represent a total protease challenge. The solution was mixed using a magnetic stirrer. One Falcon tube was designated as a “negative control” containing no sample. Five milliliters of the prepared enzyme solution were added to each tube, and the mixtures were incubated in a shaking water bath at 37 °C for 2 h (pH 7.50 ± 0.2). To denature the proteolytic enzymes in the digested solution, the pH was lowered to 3.0 using 0.1 M trichloroacetic acid, and the volume was adjusted to 50 mL with distilled water. The digested samples were centrifuged at 4,000 rpm for 5 min, and the supernatants were filtered through a 0.22 μm cellulose acetate (CA) filter. The filtrates were stored at −20 °C until analysis.

Peptide analyses were performed using a Waters Acquity UPLC I-Class Plus system coupled to a Xevo TQ-XS triple quadrupole mass spectrometer (Waters, Milford, MA, USA), equipped with a BEH C18 column (2.1 mm × 50 mm, 1.7 μm particle size). The mobile phase constitutes the liquid passing through the column in the LC section. The mobile phases consisted of: phase A, 0.1% formic acid in water; and phase B, 0.1% formic acid in acetonitrile. The LC gradient program is presented in Table 1.

To ensure stable and controlled elution of peptides on the column, the column was maintained at 40 °C and the sample tray temperature at 20 °C. The injection volume was 5 μL, and the mobile phase flow rate was 0.5 mL/min. Mass spectrometric detection was performed using electrospray ionization (ESI) in both multiple reaction monitoring (MRM) and full-scan modes. The capillary voltage was set to 0.5 kV, the cone voltage to 70 eV, and the collision energy ranged from 20 to 40 eV.

In silico analysis was used to predict the peptides generated after digestion, to verify the accuracy of LC-MS/MS results, and to compare these peptides with known allergenic protein sequences in established databases.

Analyses were performed by importing protein sequences obtained from publicly available databases into the software. Protein analysis was simulated using single, double, or triple enzyme combinations. To model gastric and small intestinal digestion of lupin proteins, two enzyme combinations were used: (i) pepsin (pH > 2; EC 3.4.23.1) + pancreatin, and (ii) pepsin (pH > 2; EC 3.4.23.1) + trypsin (EC 3.4.21.4) + chymotrypsin (EC 3.4.21.1). The ExPASy PeptideCutter tool, available through the Swiss Bioinformatics Resource Portal, was used to perform virtual enzymatic digestion of protein sequences and generate peptide fragments of varying lengths for further analysis (30). Peptide sequences were expressed using the standard single-letter amino acid code (IUPAC) following database comparisons.

Protein mass spectral data were processed using MassLynx software, with a noise reduction algorithm applied to remove peaks with a signal-to-noise ratio below 1,000 AU. The filtered spectra were then matched against protein databases.

Peptide sequences were determined from the molecular mass data of degradation products in the LC-MS spectra. The sample data were uploaded to a protein database search engine to identify the corresponding peptide sequences (31). The Swiss-Prot section of the UniProtKB database was used for peptide sequence identification (32). Allergenicity predictions were conducted using AllergenOnline.org, a database maintained by the Food Allergy Research and Resource Program at the University of Nebraska–Lincoln (33). The AllergenOnline database was searched using the “8-mer” option, enabling the detection of short-sequence peptides. Allergenicity potential was further evaluated using the AllerTOP v2.0 prediction tool, which applies multiple computational approaches, including machine learning algorithms (34, 35).

All experiments were conducted with at least three replicates, and the results were expressed as mean ± standard deviation (SD). The Chi-square test was used to analyze categorical data. For normally distributed continuous variables, one-way ANOVA followed by Tukey’s post-hoc test was applied to determine significant differences. Statistical significance was set at p < 0.05.

In our study, the chromatographic profiles the samples exhibited peaks of varying intensities over a wide time range. A pronounced signal plateau was observed particularly after the 24th minute. In contrast, the chromatographic pattern of the blank sample differed considerably. In LC-MS/MS analysis, due to the reverse phase chromatography method used, short-chain hydrophilic peptides interact weakly with the column and are eluting earlier and generating peaks at shorter retention times, whereas more hydrophobic and larger peptides are retained longer on the column (36, 37). The initial peaks observed in the microwave-treated sample (between 4–15 min) are presumed to correspond predominantly shorter-chain, low molecular weight digestion products of peptides. In contrast, the peaks observed after the 24th minute likely represent larger peptide sequences, possibly derived from enzymatically specific cleavage sites. The observation of specific signals in the blank sample during this time frame suggests that the peptides may have formed through autolysis over time in the in vitro incubation period (38).

In our study, peptides belonging to the basic globular proteins of lupin were identified. Conglutin beta1 (β-conglutin); peptides components to the conglutin delta protein (δ-conglutin) and gamma conglutin 2 (γ-conglutin) have been identified. varying cooking methods have been shown to cause significant shifts in the sequences of detectable allergenic peptides. One notable finding was that in vitro digested samples (n = 60) yielded a lower diversity of detectable peptide sequences compared to undigested samples (n = 67). In a comprehensive study conducted on Lupinus angustifolius, the capture process of peptides with the highest peak signal intensity and the lowest technical variability for each conglutin protein was filtered for liquid chromatography multiple reaction monitoring-mass spectrometry (LC–MRM-MS). This approach identified 22, 37, 13, and 10 peptides from α-, β-, δ-, and γ-conglutin proteins, respectively, across 46 lupin genotypes (39).

Proteomic analyses have shown that lupin protein isolates are dominated by β-conglutin fragments, with numerous proteins involved in germination and stress responses identified in Lupinus albus (40–42). As indicated by these studies, the differences among the peptides identified in the literature can vary depending on the method used in the analysis (such as LC-MS/MS), the ionization energy settings, the mobile phase gradient protocol, the proteolytic enzymes applied during the sample preparation process, and the data processing parameters used for peptide identification. Such technical variables influence not only in the identified peptide sequences but also in the mass spectra of diversity of identified sequences (43, 44). Furthermore, most of the peptide sequences predicted by in silico analyzes were confirmed by LC-MS/MS after digestion supporting the predictive capability of the computational models. The absence of statistical significance in the differences between groups may be attributable to the limited sample size.

Proteomic analyses on boiled beans revealed a decrease in the high molecular weight isoforms of the main storage globulin proteins, phaseolin (average change by −3.7-fold) and legumin (average change −2.5-fold) (45). In a study, low molecular weight peptides (< 3 kDa) were isolated from beans before and after grinding into flour and boiling. HPLC results showed greater peptide concentrations after the thermal treatment. Following the treatment, the bean peptides significantly decreased compared to the peptides obtained from native protein (46). In our study, the in vitro digestion process exerted different effects on the peptide profiles of the proteins found in lupin, cooked using the boiling method, depending on the protein family. One of the possible reasons for this observation is that the proteins in the in vitro gastrointestinal system were further degraded into smaller peptide sequences, rendering them undetectable by mass spectrometry (47). Indeed, in our study, the boiling process induced significant changes in the structure of lupin proteins, affecting their peptide profiles. Although each protein group responds differently, the general trend is that boiling breaks proteins into smaller fragments, enhancing enzyme accessibility during digestion.

Additionally, in another study, microwave cooking had the greatest impact on green peas and dry beans, while no thermal treatment significantly altered the peptide profile of chickpeas. Differences in the peptide profile depending on the type of legume and the applied thermal treatment were quantitatively demonstrated (48, 49). A combined approach has been proposed ultrasound and microwave techniques to obtain protein isolates from the seed of the L. angustifolius species grown in Ireland. SDS-PAGE analyses revealed that applying these processes preserved the primary structure of the proteins. However, high-power microwaves caused significant changes in the secondary structure of the proteins, transforming them from highly ordered forms (β-sheet and α-helix) to disordered forms (random coil and β-turn) (50).

The in vitro digestion model using the INFOGEST protocol enables the investigation of how proteins in processed foods are broken down in the gastrointestinal system, along with the resulting peptide profiles, and their biological activities in a laboratory setting (51). Consistent with the literature reporting increased protein digestibility and structural unfolding through microwave treatment of beans (52). Induces extensive denaturation of the protein structure, weakens intramolecular interactions, and thus unfold the protein chains, facilitating the access of digestive enzymes to the protein (53). In our study, conglutin delta showed an almost comparable peptide sequence diversity between treatments. In contrast, broader profile of unique peptides from gamma conglutin 2 were identified in the in vitro digested sample. In microwave-cooked samples, the in vitro digestion process exerted varying effects on the peptide profiles of the proteins found in lupin. It is likely that the partial denaturation of the protein or the formation of aggregates induced by thermal processing occurs differently depending on the lupin species (54).

The fermentation method, like other cooking methods, partially degrades proteins, but via a biological mechanism (19, 55, 56). During the fermentation of lupin seeds, proteases secreted by microorganisms (such as lactic acid bacteria) capable of degrading storage proteins, releasing smaller peptides and amino acids. It has also been concluded that the tannins, α-galactosides, phytic acid, and trypsin inhibitors of bacterial fermentation are also degraded (57, 58). Fermentation of Lupinus angustifolius flour has been suggested as a successful strategy to increase the acceptability of legume flavor (59, 60). Fermentation is known to degrade storage proteins and reduce allergenicity through microbial protease activity (61). Aligned with studies conducted on lupin, fermentation in alters protein content and quality in our study (62, 63). Tahmasian et al. (64) investigated the effect of fermentation induced by Rhizopus oligosporus on the proteome composition and allergenic protein content of Lupinus albus seeds. A significant reduction in the main Lupinus albus seeds allergen was reported, based on the peptide amount of the β-conglutin protein. Therefore, authors indicated that the fermentation process could be beneficial in reducing the potential allergenicity of lupin-based foods (64). The starter culture used in the current research was expected to consist of lactic acid bacteria derived from yogurt whey. Specific proteolytic cleavages carried out by microorganisms during fermentation may have contributed to the formation of these unique peptides. Additionally, the fermentation process may have induced covalent modifications in proteins (e.g., phosphorylation, deamidation), leading to differences in peptide masses, which could appear as new peaks in the LC-MS/MS profile. Consequently, these modifications may have led to the detection of different peptides. These findings support the view that the fermentation process is not only a traditional processing method but also a potential biotechnological tool for enhancing functional properties. A detailed investigation of the biological activities and health effects of peptides specifically identified in fermentation is warranted in future studies.

Enzymatic processes used as pre-treatments in foods represent a processing technology that can alter the protein structure and change their physicochemical properties. Such changes can be beneficial for formulating improved functional components and improving nutritional properties, adding value to legume proteins (65–67). Boschin et al. (68) showed that three types of lupin species exhibit similar activity scales, but after the application of pepsin + trypsin and chymotrypsin, L. luteus peptide mixtures have emerged as significantly the most active. In the study by Santos-Sánchez et al. (69), protein hydrolysate from Lupinus angustifolius was administered to mice, resulted in an increase in total plasma cholesterol, cardiovascular risk indices, and plasma antioxidant capacity in mice. Hydrolysates consisting of alcalase, papain, and pepsin were identified as the most effective for the degradation of α- and β-conglutinins, The enzymatic degradation of lupin has the potential to modify polypeptides that react with IgE and improve the techno-functional properties of the isolates, (70). In our study, the Gamma conglutin 2 protein was observed to have the most stable protein profile among all four different cooking methods examined. This protein largely maintained its structural integrity during the applied processes and produced thirteen peptides that were consistently identified under all conditions. This finding indicates that gamma conglutin 2 is structurally resistant to thermal treatments and other pre-treatments. Indeed, it has been reported in the literature that lupin seed proteins are highly resistant to thermal processes such as boiling or roasting (54). For legumin-like and conglutin beta 1 proteins, more peptides were detected in in vitro digestion, In conglutin delta, sequence counts were almost equal between treatments. In gamma conglutin 2, more peptides were detected in the in vitro digested sample. The in vitro digestion process, in addition to the enzyme treatment, exerted variable effects on the peptide profiles of the proteins found in lupin.

“YPSSTKDQ” and “TRYEEIQR” sequences matched in all four cooking methods. Their stability in all four cooking methods indicates that it is less affected by cooking processes and is highly conserved (71). The persistence of the specific peptide sequences YPSSTKDQ and TRYEEIQR across all four cooking methods warrants a molecular explanation regarding their resistance to proteolysis. The stability of YPSSTKDQ can be attributed to the presence of proline (P) at the second position. Proline is an imino acid with a rigid cyclic structure that imposes conformational constraints on the peptide backbone. It is well-documented that peptide bonds involving proline residues are often resistant to cleavage by broad-spectrum gastrointestinal proteases due to steric hindrance (72). Similarly, the resistance of the TRYEEIQR sequence is likely driven by its specific charge distribution. This sequence contains a cluster of acidic residues (Glutamic acid [E]) adjacent to bulky hydrophobic residues [tyrosine (Y), isoleucine (I)]. The high negative charge density created by the glutamic acid cluster can generate electrostatic repulsion or local structural folding that interferes with the binding efficiency of trypsin (73). Consequently, these molecular features proline-induced rigidity and charge-mediated shielding likely protect these epitopes from complete hydrolysis, preserving their structural integrity and potential IgE-binding capacity despite thermal and enzymatic processing.

In summary, comparative studies from the last decade have consistently shown that the use of these processes can transform lupin proteins. Each method has different effects but integrating them (e.g., optimizing the sequence) can often provide the most comprehensive improvements in lupin protein profiles and functionality (74, 75).

One study, the in vitro digestibility of certain legumes was assessed to determined that among all cooking methods, pressure cooking is the most effective (76). In another study, microwave cooking yields the highest levels of bioactive peptides after digestion in green peas and dried beans, while pressure cooking and traditional boiling were found to be less effective in terms of peptide diversity (48). In another study, autoclaving markedly increased the degree of sequential pepsin/pancreatin hydrolysis, whereas boiling increased it only in certain legumes (77).

This study used a comparison of two models to evaluate the in vivo persistence potential of the identified epitopes. The INFOGEST protocol was used to simulate the complex, stepwise physiological conditions of human digestion, while a simplified protease assay was applied in parallel as a high-sensitivity biochemical control to test structural resistance. Consequently, peptide sequences demonstrating stability in both models may provide the strongest inference of resistance to proteolysis. As shown in Figure 4, although proteins from the same family were identified in both groups, notable differences in peptide diversity were observed, highlighting the selective nature of the physiological model. This divergence reconciles the perceived contradiction resolved by the various results shown in Figure 4 (an increase in some proteins and a decrease in others). Using a selective method like LC-MS/MS, this study demonstrated that the in vitro digestion significantly impacts the peptide profiles of proteins with effects varying by protein family.

In a study, the degradation of protein isolates obtained from various legumes and the peptide and free amino acid profiles reflecting the proteomic results under the INFOGEST protocol were analyzed. It was found that legumes, particularly soybeans, show resistance during gastric digestion, leading to the release of substantial amounts of peptides and amino acids (78). In lupin species, the closest study to this has investigated how different protein extraction methods of Lupinus angustifolius affect the proteome profile of lupin seeds. it demonstrates that the choice of method significantly affects the proteomic outcomes. This is crucial for understanding how digestion protocols can influence proteome comparisons (39, 79). To our knowledge, this is the first research conducted on Lupinus albus, which can compare the proteome profile obtained from the INFOGEST protocol with the standard protease method. In vitro digestion applied and unprocessed samples show that both peptide sequences belonging to the same protein (Conglutin beta 1) and those belonging to a different protein (legumin-like) matched exactly the allergen database. Although the same cooking methods were applied, treatment with different enzymes resulted in the formation of distinct peptide sequences in the lupin samples, which in turn produced different allergen matches. Additionally, the greater number of matches to known epitopes observed in samples subjected to an in vitro gastrointestinal system model suggests that this approach yields results more representative of allergenic peptides that may arise in a real biological environment.

Key strengths of this study include the use of high-resolution LC-MS/MS to determine the sequences of peptides formed after different cooking methods and in vitro digestion, as well as potential protein matches. This method is currently the most preferred technique in terms of sequence accuracy, precision, and comprehensiveness of data acquisition. As a cooking method, lactic acid bacteria fermentation is considered a relatively natural process that is safer than chemical/physical modification methods and can enrich the functional properties of legume proteins through biological degradation and modification. The use of lactic acid fermentation method in our study adds relevance to the work. Additionally, in silico analyses and bioinformatics databases were extensively utilized, providing detailed data and strengthening the analyses.

However, this study has certain limitations that should be acknowledged. First, the data presented relies on a label-free peptidomic approach, which provides semi-quantitative information based on peptide diversity and sequence counts rather than absolute quantification recommending future targeted studies for absolute quantification. Second the fact that the peptides providing the best match were identified using the LC-MS/MS method does not necessarily indicate the absence of other important protein-bound peptides; it only means that these peptides were not captured. Several factors could explain why these peptides were not captured. The peptide may have been degraded to the amino acid level, or it may have broken at a different point in the sequence. In addition, the ionization energies may not have been sufficient to detect these peptides. Furthermore, databases like Structural Database of Allergenic Proteins (SDAP) are less optimized for short peptide sequences, and some identified peptides did not meet the specific structural criteria of AllerTOP v2.0. Therefore, a definitive judgment cannot be made about the potential allergenicity of unassessed peptides. Nevertheless, peptides showing complete matches with Lupinus allergens in undigested samples (Supplementary Table 7) provide a valuable baseline for future risk assessment.