Effects of deamidation by protein glutaminase on the flavor binding properties of pea protein isolate

Abstract

This study aimed to investigate the effects of protein deamidation by protein glutaminase (PG) on the flavor binding properties of pea protein isolate (PPI), using vanillin as a model. The binding behaviors of native PPI and deamidated PPI were assessed at different temperatures (5–25 °C). The results showed that the number of binding sites (n) increases with decreasing temperature. In addition, the binding constant (K) and overall binding (nK) are considerably lower after deamidation. Thermodynamic analysis revealed negative ∆G° values for both proteins, confirming spontaneous binding. Additionally, positive ∆H° and ∆S° values suggested that the interactions are entropy-driven and primarily hydrophobic in nature, which was confirmed using molecular docking studies, with stronger bonding to vanillin observed for PPI than for deamidated PPI. Sensory evaluation revealed that deamidation promoted flavor release. Thus, PG deamidation enhances flavor delivery performance, positioning deamidated PPI as promising protein-based component for improving flavor interactions in food applications.

Highlights

• •Temperature-dependent protein rearrangement alters flavor-binding site exposure.
• •Protein glutaminase deamidation reduces vanillin binding to pea protein.
• •Vanillin-PPI interactions are entropy-driven dominated by hydrophobic interaction.

1. Introduction

The consumption of high-protein foods has been increasing as individuals aim to improve dietary protein intake and reduce fat consumption. As a result, proteins have been widely incorporated into foods as supplements or functional ingredients. However, high-protein foods can present dietary limitations for certain consumer groups (Toews & Wang, 2013). Therefore, the development of alternative proteins that could replace animal-derived proteins or allergenic plant proteins, such as soy protein, is paramount for expanding the scope of available high-protein foods (Verhoeckx et al., 2015). Peas (Pisum sativum L.) have attracted notable attention as a safe, plant-based protein source. Comparative studies have shown that pea protein contains higher levels of total protein, essential amino acids, and non-essential amino acids than soybean protein (Lam et al., 2016). However, these advantages are limited by protein-flavor interactions in pea protein systems, which can reduce aroma release and alter the overall flavor profile of the foods.

Flavor remains a critical factor in the consumer acceptance of high-protein foods. Proteins can interact with flavor compounds, potentially leading to flavor loss or distortion through reversible and irreversible mechanisms and thereby diminishing the sensory quality of food products (Anantharamkrishnan & Reineccius, 2020; Guo et al., 2019; Wongprasert, Mathatheeranan, Siripitakpong, et al., 2024; Suppavorasatit & Cadwallader, 2010). These interactions are governed by multiple factors, including the functional groups of the flavor compound and the physicochemical properties of proteins (Zhang et al., 2021). Wongprasert, Mathatheeranan, Siripitakpong, et al. (2024) reported that flavor compounds with different functional groups, such as esters, aldehydes, ketones, and alcohols, have different binding patterns with pea protein isolate (PPI), leading to an unbalanced flavor profile in the foods. Additionally, Wang and Arntfield (2015) reported that aldehydes and ketones with identical carbon chain lengths exhibit different binding behaviors toward canola and pea proteins. Both proteins show a stronger affinity for aldehydes than for ketones. The chain length of both aldehydes and ketones also affects binding behavior because of steric hindrance.

Among the strategies developed to address this issue, modification of protein structures has shown considerable potential. Some plant proteins, such as soy, coconut, and pea proteins, have high glutamine content, which were reported as a potentially key residue for flavor interactions (Temthawee et al., 2020; Wang & Arntfield, 2015). Therefore, protein modification by targeting glutamine in the protein chain can directly change protein conformation, thus altering the interactions between the protein and flavor compounds. Among protein modification methods, enzymatic approaches are preferred because of their specificity, safety, and ability to operate under mild conditions (Suppavorasatit et al., 2011; Wang & Arntfield, 2016; Yong et al., 2004). Protein glutaminase (PG; EC 3.5.1.44) is an enzyme that catalyzes the deamidation of glutamine residues, converting them into glutamic acid while releasing ammonia (Yamaguchi et al., 2001). This deamidation can improve the functional properties of proteins, potentially affecting their interactions with flavor compounds (Suppavorasatit & Cadwallader, 2010). Previous studies have shown the effectiveness of PG deamidation in reducing the binding affinity between proteins and flavor compounds. For example, Temthawee et al. (2020) reported a decrease in overall affinity between coconut protein and vanillin after PG deamidation, which is consistent with findings for soybean protein (Suppavorasatit & Cadwallader, 2012). As mentioned above, pea protein contains higher levels of total protein, essential amino acids, and non-essential amino acids than soybean protein (Lam et al., 2016). Notably, its dominant amino acids are glutamine/glutamic acid and asparagine/aspartic acid (Millar et al., 2019; Pownall et al., 2010). Given that glutamine is abundant in pea protein, it is an ideal substrate for PG deamidation.

Recent studies have shown that hydrophobic interactions, hydrogen bonds, and van der Waals forces are key contributors to the binding of flavor compounds to proteins (Bi et al., 2022). Carbonyl containing compounds such as vanillin are known to form such interactions, which can be affected by protein structure. Although these mechanisms have been investigated in computational and experimental studies, the information on the effects of PG deamidation on flavor binding in pea protein is very limited. Wongprasert, Mathatheeranan, Chen, et al. (2024) investigated the interactions between pea protein and vanillin and found that hydrophobic interactions are the primary binding mode. Their molecular docking results further supported this finding, indicating that vanillin predominantly interacts with pea protein through hydrophobic interactions. Based on these observations, PG deamidation could be employed to alter the bonds between pea protein and flavor compounds through the modulation of the structural features that contribute to hydrophobic interactions.

The present study aimed to investigate the impact of protein deamidation using PG on the flavor binding properties of PPI using vanillin as a model. To the best of our knowledge, this is the first study on PG deamidation of the PPI affecting flavor interactions and releases. The elucidation of these interactions could support the development of plant-based protein ingredients with improved flavor release and enhanced consumer acceptability.

2. Materials and methods

2.1. Materials

PPI (NUTRALYS® S85F) was obtained from Roquette Co., Ltd. (Paris, France). The composition of protein powder was reported as 82.37% protein, 8.92% lipid, 5.09% ash, and 3.62% carbohydrate (dry basis). PG “AMANO” 500 (purity >95%) was provided by Amano Enzyme (Nagoya, Japan). Vanillin-d₃, used as an internal standard, was synthesized following the methods described by Schneider and Rolando (1992). All other reagents and chemicals (analytical grade) were purchased from Bio-Rad Laboratories (Hercules, CA, USA), Sigma-Aldrich (St. Louis, MO, USA), and Thermo Scientific (Pittsburgh, PA, USA).

2.2. PPI preparation

PPI was prepared by defatting pea protein using Soxhlet extraction with hexane following a method reported by Wongprasert, Mathatheeranan, Chen, et al. (2024). The protein content of PPI was determined to be 92.35% based on nitrogen analysis via the Kjeldahl method and an N-to-protein conversion factor of 6.25 (Thaiphanit et al., 2016). The defatted PPI sample was vacuum-sealed and stored at −18 °C for further experiments.

2.3. Preparation of deamidated PPI (DPPI)

The deamidation of PPI using PG was conducted following the method employed by Suppavorasatit et al. (2011) and Kunarayakul et al. (2018), with some modifications. The reaction mixture consisted of 30 mg/mL PPI and PG in 0.01 M citrate-phosphate-borate buffer at pH 7.0, with an enzyme-to-substrate ratio of 36 U/g protein. The mixture was incubated at 50 °C for 1 h in triplicate. Enzymatic activity was terminated by holding the mixture at 80 °C for 10 min. All samples were subsequently dialyzed, freeze-dried, vacuum-sealed, and stored at −18 °C for further analysis. The degree of deamidation (DD) and the degree of hydrolysis (DH) were determined following previously reported methods with some modifications (Kunarayakul et al., 2018; Yong et al., 2006). The DD was determined by quantifying the amount of ammonia released during deamidation using an ammonia assay kit (ab83360, Abcam, Cambridge, UK). The DD was expressed as the percentage of ammonia released from the DPPI sample relative to that released from a fully hydrolyzed PPI sample. The DH was calculated as a ratio (percentage) of soluble protein present in the supernatant after precipitation with 0.2 N trichloroacetic acid to that in a fully hydrolyzed sample. The concentration of soluble protein in the supernatant was measured using the detergent-compatible protein assay kit (Bio-Rad Laboratories, Hercules, CA, USA).

2.4. Preparation of the flavor compound solution

A vanillin solution (1050 μg/mL) in phosphate buffer (pH 7.0) was prepared using odorless distilled water. A stock solution of vanillin-d₃ (920 μg/mL) was prepared in methanol. All solutions were kept in amber glass vials and stored at −40 °C prior to analysis.

2.5. Free (unbound) vanillin isolation and quantification

The isolation and quantification of free vanillin were performed according to the method described by Temthawee et al. (2020) with some modifications. Specifically, 5 mL of reaction mixture containing protein (3% w/v) and vanillin (Sigma-Aldrich, St. Louis, MO, USA) was added to an Amicon® Ultra-15 centrifugal filter tube with a 3 K molecular weight cutoff (Merck Millipore Ltd., Cork, Ireland). The mixture was then centrifuged at 5000 rpm for 30 min in a refrigerated centrifuge (Hybrid high-speed 6200, Kubota, Tokyo, Japan), with the temperature maintained to match the incubation temperature (5 °C, 15 °C, or 25 °C). The permeate (0.5 mL) was then transferred to a 2 mL glass vial, spiked with 10 μL of an internal standard solution (920 μg/mL vanillin-d₃), and thoroughly mixed. The mixed solution was then extracted with 0.3 mL dichloromethane (RCI Labscan, Ireland). The extracted fraction was subsequently analyzed using gas chromatography–mass spectrometry.

To determine the content of free vanillin, a 7890B GC/7000D MS Triple Quadrupole System (Agilent Technologies, Inc., Palo Alto, CA, USA) was used. Each 2 μL sample was injected in hot splitless mode at 250 °C. A DB-Wax column (30 m × 250 μm i.d. × 0.25 μm film thickness; Agilent J&W, Agilent Technologies, Palo Alto, CA, USA) was used for compound separation. The oven temperature was increased from 40 to 220 °C at a rate of 10 °C/min, with a final holding time of 5 min. Helium was employed as a carrier gas with a constant flow rate of 2.0 mL/min. Mass spectrometry was performed under the following conditions: transfer line temperature of 250 °C, ionization voltage of 70 eV, a mass range (scan mode) of 35–400 amu, and a scan rate of 4.2 cycles/s.

Vanillin content was calculated using the method described by Temthawee et al. (2020). The mass spectrometry response factor (f_i) was used for comparison against the internal standard (vanillin-d₃). The f_i of vanillin relative to vanillin-d₃ was calculated based on the gas chromatography–mass spectrometry peak areas of ions 151 (vanillin) and 154 (vanillin-d₃), and was determined to be 1.01 (data not shown). The mass of vanillin was calculated as shown in Eq. (1).

$$\text{Mass of vanillin}=\text{Mass of vanillin}-d_3\times {\mathrm{f}}_{\mathrm{i}}\times \frac{\text{Peak area of vanillin}\left(\mathrm{ion}151\right)}{\text{Peak area of vanillin}-d3\left(\mathrm{ion}154\right)}$$ (1)

2.6. Determination of equilibration time

The equilibration time for protein–vanillin interactions was determined using the equilibrium dialysis method described by Suppavorasatit and Cadwallader (2012) and Temthawee et al. (2020) with some modifications. All glassware was silanized before use, according to the procedure of Tsutsumi et al. (2003) to enhance surface hydrophobicity. Protein solutions (3% w/v PPI and DPPI) were prepared in 0.05 M phosphate buffer (pH 7.0) and allowed to hydrate overnight at 4 °C. Each hydrated protein sample was placed in a 20 mL vial, followed by the addition of vanillin to achieve a final concentration of 50 μg/mL. Each vial was covered with a Teflon-lined cap and placed in a low-form jacketed beaker (VR Glasstrade, Thailand) on a magnetic stirrer connected to a water bath to maintain a constant temperature (5 °C, 15 °C, or 25 °C). The mixtures were continuously stirred throughout the incubation period. Samples were collected over a 72-h period, and the concentration of free vanillin was determined using the method described above. The equilibration time for each sample at each temperature was determined as the time at which the plots of free vanillin against incubation time reached a plateau.

2.7. Determination of binding properties

The flavor binding properties of the proteins were assessed using the methods described by Temthawee et al. (2020). Suspensions of 3% (w/v) PPI and DPPI in 0.05 M phosphate buffer at pH 7.0 were prepared and refrigerated overnight at 4 °C. Ten mL of each sample was placed in 20 mL silanized vials. Vanillin was added to achieve final concentrations of 10, 20, 40, 60, 80, and 100 μg/mL, and the vials were sealed with Teflon-lined caps. The samples were incubated under continuous stirring at three constant temperatures (5 °C, 15 °C, or 25 °C) for the thermodynamic study. After equilibration, the amount of unbound vanillin in each sample was quantified. Binding parameters, including the number of binding sites (n) and the binding constants or affinity (K), were calculated from Klotz plots of $\frac{1}{\mathrm{\nu }}$ versus $\frac{1}{\left[\mathrm{L}\right]}$, using the Klotz Eq. (2), where ν is the number of moles of vanillin bound per mole of total protein, and [L] is the concentration of free vanillin. The slope of the plot equals $\frac{1}{\mathrm{Kn}}$ and the y-intercept equals $\frac{1}{\mathrm{n}}$.

$$\frac{1}{\mathrm{\nu }}=\frac{1}{\mathrm{n}}+\frac{1}{\mathrm{Kn}\left[\mathrm{L}\right]}$$ (2)

2.8. Determination of thermodynamic parameters

Thermodynamic parameters, including the Gibbs' free energy of binding ($∆$G°), enthalpy of binding ($∆$H°), and entropy of binding ($∆$S°), at each temperature were evaluated. The binding constant (K) was determined from the Klotz equation as described above.

The Gibbs' free energy of binding ($\mathsf{∆}$G°) was calculated from Eq. (3).

$$∆{\mathrm{G}}^{°}=-\text{RTlnK}$$ (3)

where R is the gas constant (8.314 J/mol·K), T is the absolute temperature (Kelvin, K), and K is the binding constant.

The enthalpy of binding ($\mathsf{∆}$H°) was calculated using the van't Hoff Eq. (4).

$$∆{\mathrm{H}}^{°}=\frac{\left(-\mathrm{R}\right)\times \left(\ln \left({\mathrm{K}}_1-{\mathrm{K}}_2\right)\right)}{{\mathrm{T}}_1-{\mathrm{T}}_2}$$ (4)

where K₁ and K₂ are the binding constants at 5 °C and 25 °C, respectively, T₁ and T₂ are the absolute temperatures in degrees Kelvin at 5 °C and 25 °C, respectively, and R is the gas constant.

The enthropy of binding ($\mathsf{∆}$S°) was calculated using Eq. (5).

$$∆{\mathrm{S}}^{°}=\frac{∆{\mathrm{H}}^{°}-∆{\mathrm{G}}^{°}}{\mathrm{T}}$$ (5)

2.9. Molecular docking

Molecular docking studies were performed to predict the interactions between pea proteins and vanillin, following the method described by Bi et al. (2022) with some modifications. The crystal structure of native pea protein was obtained from the Protein Data Bank (PDB ID: 3KSC) based on the study by Dai et al. (2020). The structure of DPPI (PDB ID: 3KSC) was modified using the Phyre² server (http://www.sbg.bio.ic.ac.uk/phyre2) (Kelley et al., 2015). Prior to modification, all water molecules and ligand atoms were removed. The crystal structure of vanillin was retrieved from the PubChem database (CID: 1183) (https://pubchem.ncbi.nlm.nih.gov). The target binding site was defined within the hydrophobic region at the terminal end of the protein monomer, as reported by Wongprasert, Mathatheeranan, Siripitakpong, et al. (2024).

Molecular docking simulations were performed using AutoDock Vina (https://vina.scripps.edu). The structures of both the protein targets and the ligand were converted to the required protein data bank in PDBQT format using AutoDockTools (Morris et al., 2009). The grid box size was set to 10 × 10 × 10 Å³, with the center coordinates positioned at X = 5.02, Y = 256.76, and Z = 56.82 to cover the hydrophobic binding pocket. The docking results were visualized using UCSF Chimera (RBVI, UCSF, CA, USA) and BIOVIA Discovery Studio 2023 (Dassault Systèmes BIOVIA Ltd., France).

2.10. Sensory evaluation

The procedure for sensory evaluation employed in the present study was authorized by the Research Ethics Review Committee for Research Involving Human Research Participants, Health Sciences Group, Chulalongkorn University, Bangkok, Thailand (COA No. 081/67). The odor detection thresholds for vanillin in PPI and DPPI samples were determined using the sensory evaluation method based on ASTM E697-04 (ASTM, 2004) and Temthawee et al. (2020) with some changes. A protein solution containing 3% (w/v) of both PPI and DPPI was prepared in 0.05 M phosphate buffer at pH 7.0. Each solution (14 mL) was placed in individual Teflon sniff bottles (Nalgene™, USA), and each bottle was spiked with 1 mL of vanillin to achieve six concentrations for each protein. The mixtures were kept for 48 h at 4 °C to allow the interactions to reach equilibrium. Protein samples were served in six sets, with each set consisting of three bottles: two without vanillin and one with vanillin. Before testing, all protein samples were placed at room temperature for 1 h. All sniff bottles were covered with aluminum foil, and each bottle was labeled with a 3-digit random code to avoid bias. Using the 3-alternative forced choice (3-AFC) method, all sets were presented to panelists in order of ascending vanillin concentration. The panelists, 10 males and 18 females aged between 19 and 45 years old, were instructed to sniff one set of samples at a time and choose the bottle with the strongest vanillin odor, or guess if they could not differentiate among the three samples. The group's best estimate threshold (BET) was calculated from the mean of individual BETs using the methods described by Temthawee et al. (2020).

2.11. Statistical analysis

Data were analyzed using analysis of variance. Duncan's new multiple range test, and independent-sample t-test were used to compare differences among treatments at a 95% confidence interval using the IBM® SPSS® Statistics software version 29.0 (IBM, Armonk, NY, USA). The DH and DH were determined in triplicate.

3. Results and discussion

3.1. Equilibration time for the binding of vanillin to PPI and DPPI

PG was used to selectively induce deamidation in PPI by modifying glutamine residues within the protein sidechain. This enzymatic reaction resulted in the formation of glutamic acid and the release of ammonia without affecting free glutamine (Liu et al., 2022; Yamaguchi et al., 2001). The treatment was conducted under optimized conditions to achieve a high DD while minimizing peptide bond cleavage (to maintain a low DH). The resulting DD and DH of DPPI were 63.4% and 4.6%, respectively. Despite the relatively high DD, the DH remained low, indicating that PG catalyzed deamidation without causing substantial hydrolysis. These findings confirm the enzyme specificity and are consistent with previous reports on soy, wheat gluten, and coconut proteins (Kunarayakul et al., 2018; Suppavorasatit et al., 2011; Temthawee et al., 2020; Yong et al., 2004).

The equilibration time for vanillin binding to PPI and DPPI at different temperatures according to the temperature coefficient theory (Q₁₀) (5, 15, and 25 °C) is presented in Fig. 1 and Table 1. At 5 °C, binding equilibrium was reached after 48 h of incubation (Fig. 1a), indicating that lower temperatures require longer equilibration times for stable interactions compared with higher temperatures, such as 25 °C (Fig. 1c). This observation aligns with previous reports, suggesting that lower temperatures reduced the rate of biochemical reactions between the protein and ligand (flavor compound) (Anantharamkrishnan & Reineccius, 2020; Lamikanra et al., 2000). Similarly, at 15 °C (Fig. 1b), the binding of PPI and DPPI with vanillin reached equilibrium after 48 h of incubation (same as at 5 °C), indicating that equilibration at this moderate temperature still requires a relatively extended period for the binding process to reach maximum efficiency. Notably, at 25 °C (Fig. 1c), equilibrium was achieved after only 36 h, indicating that equilibration proceeded faster at a higher temperature (Table 1).

Fig. 1.

Equilibrium time of vanillin to PPI ● and DPPI at 5 °C (a), 15 °C (b), and 25 °C (c). ν = number of moles of vanillin bound per mole of total protein.

Table 1.

Equlibration time for the binding of vanillin to vanillin to PPI and DPPI.

Protein	T (οC)	Equilibration time (h)
PPI	5	48
	15	48
	25	36
DPPI	5	48
	15	48
	25	36

These findings are consistent with the general principle that elevated temperatures accelerate biochemical reactions (Anantharamkrishnan & Reineccius, 2020). These findings collectively indicate that temperature considerably affects the rate of binding between pea proteins and vanillin. The longer equilibration times observed at lower temperatures suggest that food processing or flavoring applications involving pea protein may require extended processing times or incubation periods to achieve optimal flavor binding. Conversely, at higher temperatures, binding occurs faster, which can be advantageous for applications requiring shorter processing times. This observation is consistent with studies on the equilibration of protein–vanillin interactions for soy protein and coconut protein, which also reported more rapid binding at higher temperatures (Suppavorasatit et al., 2013; Temthawee et al., 2020).

3.2. Binding affinity of the vanillin to PPI

The double-reciprocal plots (Klotz plots) showing the binding interactions between vanillin and PPI or DPPI at 5 °C, 15 °C, and 25 °C are presented in Fig. 2(a–c). All plots exhibit linearity within the temperature range, suggesting that vanillin binds independently (non-cooperatively) with both PPI and DPPI (Li et al., 2000). These findings are consistent with a study on the binding affinity of vanillin to coconut and deamidated coconut proteins, which also reported linear plots for vanillin binding to both proteins (Temthawee et al., 2020). The Klotz plots for vanillin binding to soy protein isolate (SPI) and deamidated SPI are also linear (Suppavorasatit & Cadwallader, 2012). In addition, Bi et al. (2022) studied the binding of PPI to (E)-2-octenal, hexanal, and (Z)-3-penten-1-ol, with the results suggesting that these flavor compounds independently bind to the PPI. Similarly, Wongprasert, Mathatheeranan, Chen, et al. (2024) reported the non-cooperative binding of ethyl butanoate, ethyl hexanoate, ethyl isopentanoate, and methyl anthranilate to PPI within the same temperature range. In a separate study, non-cooperative binding was observed for vanillin, γ-decalactone, furaneol, and (Z)-3-hexen-1-ol at 5 °C, 15 °C, and 25 °C, with linear Klotz plots indicating a proportional, non-cooperative binding with PPI across these temperatures (Wongprasert, Mathatheeranan, Siripitakpong, et al., 2024). These findings collectively suggest that PPI can independently bind various flavor compounds. The linear equations from the Klotz plots for vanillin binding to PPI and DPPI are shown in Table 2, with coefficients of determination (R²) exceeding 0.985, which corresponds to more than 98.5% of the total variation being explained in the plots.

Fig. 2.

Klotz plots for binding of vanillin to PPI ● and DPPI at 5 °C (a), 15 °C (b), and 25 °C (c). ν = number of moles of vanillin bound per mole of total protein.

Table 2.

Linear equations from Klotz plots for the binding of vanillin to PPI and DPPI.

Protein	T (οC)	Replication 1		Replication 2
		Equation	R2	Equation	R2
PPI	5	y = 0.0625× + 0.8120	0.9918	y = 0.0610× + 0.8187	0.9923
	15	y = 0.0629× + 0.8505	0.9901	y = 0.0645× + 0.8992	0.9953
	25	y = 0.0698× + 0.9097	0.9924	y = 0.0680× + 0.9555	0.9980
DPPI	5	y = 0.0844× + 0.7739	0.9982	y = 0.0860× + 0.7446	0.9893
	15	y = 0.0876× + 0.8001	0.9957	y = 0.0865× + 0.8505	0.9970
	25	y = 0.0880× + 0.9361	0.9954	y = 0.0927× + 1.0283	0.9859

The binding parameters, including the number of binding sites (n), binding constant (K), and overall binding value (nK) at different temperatures, were calculated from Klotz plots using Eq. (2.2). All binding parameters are shown in Table 3. The n values for the binding of vanillin to PPI are 1.23, 1.17, and 1.07 at 5 °C, 15 °C, and 25 °C, respectively. At the same time, the n values for vanillin bound to DPPI at 5 °C, 15 °C, and 25 °C are 1.32, 1.21, and 1.02, respectively. These results indicate that n values increase with decreasing temperature. However, no significant differences (p > 0.05) in n values are observed among proteins at the same temperature. For the same protein, the n value at 25 °C is significantly lower than those observed at 5 °C and 15 °C. The higher n values for vanillin binding to both PPI and DPPI at lower temperatures were attributed to enhanced protein unfolding. At lower temperatures, structural rearrangements in the protein are promoted because of weakened hydrophobic interactions. This leads to the exposure of additional binding sites on the protein surface, thereby increasing the number of available binding sites for vanillin. Consequently, more binding sites for vanillin are accessible at 5 °C than at 25 °C. This aligns with the results of previous studies, which indicated that lower temperatures can enhance protein flexibility and reveal more hydrophobic regions, thereby increasing the number of available binding sites (Suppavorasatit & Cadwallader, 2012). However, elevated temperatures can also lead to protein unfolding, but with different consequences. Whereas moderate thermal unfolding may promote binding in some cases, excessive heat can induce protein aggregation, reducing the flavor binding capacity by masking or altering binding sites. Kühn et al. (2008) demonstrated that heat treatment of whey protein isolate reduced flavor binding, which was occurred by the aggregation of unfolded β-lg molecules. Therefore, it could limit the accessibility or alter the conformation of binding sites of the protein. These results indicated that temperature-induced unfolding did not necessarily enhance flavor binding; instead, binding behavior depended on the balance between increased exposure of binding sites and aggregation or structural disruption that could hinder interactions with flavor compounds.

Table 3.

Binding parameters for the binding of vanillin to PPI and DPPI. Data represent mean ± SD of duplicates.^a-b, ^A-B

Parameter	T (οC)	PPI	DPPI
n	5	1.23b ± 0.01	1.32b ± 0.04
	15	1.17b ± 0.01	1.21b ± 0.05
	25	1.07a ± 0.04	1.02a ± 0.07
K (×104) (L/mol)	5	1320.67aA ± 30.36	891.38aB ± 36.15
	15	1334.11aA ± 25.50	948.30aB ± 49.41
	25	1354.22aB ± 72.02	1086.51bA ± 32.19
nK (×104) (L/mol)	5	1619.67bA ± 27.82	1173.81aB ± 15.59
	15	1560.61bA ± 41.32	1148.81aB ± 10.26
	25	1451.63aA ± 26.82	1107.56aB ± 40.74

^a-bWithin columns, values with the same lower-case letters are not significantly different at p > 0.05.

^A-BWithin rows, values with the same upper-case letters are not significantly different at p > 0.05.

The K values of 1320.67 × 10⁴, 1334.11 × 10⁴, and 1354.22 × 10⁴ L/mol (Table 3) are observed for vanillin bound to PPI at 5 °C, 15 °C, and 25 °C, respectively. The K values for DPPI at 5 °C, 15 °C, and 25 °C are 891.38 × 10⁴, 948.30 × 10⁴, and 1086.51 × 10⁴ L/mol, respectively. No significant differences are observed between the K values of each protein at different temperatures, except for DPPI, which shows a significantly higher K value at 25 °C than at 5 °C and 15 °C. However, the K values for DPPI are significantly lower than those for PPI. This reduction in binding strength was attributed to the increase in DD, as PG deamidation is known to modify the secondary structure of proteins, which can affect their interaction with flavor compounds (Kunarayakul et al., 2018). These results align with previous studies, where a decrease in K values after PG deamidation was reported for soy and coconut proteins (Suppavorasatit & Cadwallader, 2012; Temthawee et al., 2020). In SPI and deamidated SPI, vanillin and maltol showed similar binding behavior, likely because of their similar carbonyl and hydroxyl groups. The reduced affinities for deamidated SPI were attributed to the loss of reactive amide side chains, particularly glutamine residues, and structural changes caused by PG treatment, which limited the ability of the protein to bind flavor compounds (Suppavorasatit & Cadwallader, 2012).

Additionally, PG deamidation may alter the secondary structure of proteins, potentially contributing to the observed changes in n and K values (Temthawee et al., 2020). The increase in K values with temperature is in agreement with the findings of He et al. (2021), who investigated the effect of incubation temperature on the binding capacity of aldehyde compounds to myosin. Furthermore, a similar trend was reported for SPI, where K values increased with temperature (Suppavorasatit & Cadwallader, 2012). This temperature-dependent increase in K values was attributed to the loosening of the protein structure at elevated temperatures, facilitating access to specific amino acid residues and promoting the binding of flavor compounds (He et al., 2021). Higher K values indicate greater stability of the vanillin-protein complex (Dinu et al., 2022).

However, only n or K values are insufficient for describing the overall binding affinity of the flavor compound to proteins because n and K values can change owing to the rearrangements occurring at the outer surface of the binding site (Temthawee et al., 2020). A study by Zhou and Cadwallader (2006) indicated that the nK value, derived from the Klotz equation, more accurately reflects the overall binding affinity. The nK values of vanillin bound to PPI at 5 °C, 15 °C, and 25 °C are 1619.67 × 10⁴, 1560.61 × 10⁴, and 1451.63 × 10⁴ L/mol, respectively. At the same time, the nK values of vanillin bound to DPPI at 5 °C, 15 °C, and 25 °C are 1173.81 × 10⁴, 1148.81 × 10⁴, and 1107.56 × 10⁴ L/mol, respectively. The nK values for the binding of vanillin to DPPI are significantly lower than those for PPI. However, for the same protein, the nK values do not significantly differ at the studied temperatures, except for PPI, which shows a significantly lower nK value at 25 °C than at 5 °C and 15 °C. The decrease in nK values with increasing temperatures was attributed to the rearrangement of protein structures at lower temperatures, which has been shown to expose hydrophobic binding sites and facilitate stronger binding with vanillin (Damodaran, 2008). As previously discussed, protein–flavor interactions may involve both reversible and irreversible mechanisms. Among the irreversible interactions, Schiff-base formation is the generation of a covalent bond between the carbonyl group of the flavor compound and the amine group of amino acid side chains (Kühn et al., 2006; Suppavorasatit & Cadwallader, 2010; Temthawee et al., 2020). The observed decrease in nK values for DPPI may be partially explained by the reduced availability of amine groups after PG deamidation, during which glutamine residues are converted into glutamic acid (Suppavorasatit & Cadwallader, 2012). This structural modification likely limited the formation of Schiff bases, thereby inhibiting irreversible binding between DPPI and vanillin and contributing to the overall decrease in binding capacity.

3.3. Thermodynamics of the binding of vanillin to PPI and DPPI

The thermodynamic parameters of the binding of vanillin to PPI and DPPI, including the Gibbs' free energy of binding ($∆$G°), enthalpy of binding ($∆$H°), and entropy of binding ($∆$S°) are presented in Table 4. These parameters are essential for evaluating the stability and spontaneity of binding and are crucial factors in the development of protein-containing food products with a desired flavor (Wongprasert, Mathatheeranan, Siripitakpong, et al., 2024). The $∆$G° values obtained herein for both PPI and DPPI are negative at all studied temperatures. These negative $∆$G° values indicate that the binding of vanillin to both PPI and DPPI is spontaneous (Temthawee et al., 2020). Additionally, $∆$G° values significantly decrease with increasing temperature. At the same time, $∆$G° values are significantly higher after PG deamidation. The $∆$G° values observed for vanillin bound to PPI are −9.05, −9.06, and −9.72 kcal/mol at 5 °C, 15 °C, and 25 °C, respectively. At the same time, $∆$G° values observed for vanillin bound to DPPI are −8.83, −8.87, and −9.59 kcal/mol at 5 °C, 15 °C, and 25 °C, respectively. These results are consistent with the results of a study on the binding of vanillin to SPI, which reported a decrease in $∆$G° values with increasing temperature (Suppavorasatit & Cadwallader, 2012).

Table 4.

Thermodynamic parameters for the binding of vanillin to vanillin to PPI and DPPI. Data represent mean ± SD of duplicates.^a-c, ^A-B

Parameter	T (°C)	PPI	DPPI
$∆$G° (kcal/mol)	5	−9.05cB ± 0.01	−8.83cA ± 0.02
	15	−9.06bB ± 0.01	−8.87bA ± 0.03
	25	−9.72aB ± 0.03	−9.59aA ± 0.02
$∆$H° (kcal/mol)	5–25 ns	0.20 ± 0.25	1.63 ± 0.58
$∆$S° (cal/mol·K)	5 ns NS	33.29 ± 0.94	37.64 ± 2.00
	15 ns NS	33.28 ± 0.83	37.56 ± 2.11
	25 ns NS	33.88 ± 1.77	38.32 ± 2.96

ns, NS = not significance.

^a-cWithin columns, values with the same lower case letters are not significantly different at p > 0.05.

^A-BWithin rows, values with the same upper-case letters are not significantly different at p > 0.05.

The $∆$H° values of vanillin binding to PPI and DPPI are positive and do not significantly differ between the two proteins. Specifically, the $∆$H° values for vanillin bound to PPI and DPPI are 0.20 and 1.63 kcal/mol, respectively. The $∆$S° values for the binding of vanillin to PPI and DPPI are also positive and not significantly different. For PPI, the $∆$S° values are 33.29, 33.28, and 33.88 cal/mol·K at 5 °C, 15 °C, and 25 °C, respectively. For DPPI, the $∆$S° values are 37.64, 37.56, and 38.32 cal/mol·K at 5 °C, 15 °C, and 25 °C, respectively.

He et al. (2021) noted that the main types of interactions between small molecules and proteins can be determined based on the binding enthalpy ($∆$H°) and entropy ($∆$S°) values. Similarly, Ross and Subramanian (1981) identified the type of molecular interactions based on the changes in thermodynamic parameters. Both $∆$H° and $∆$S° being greater than zero typically indicate the predominance of hydrophobic interactions. Conversely, when both $∆$H° and $∆$S° values are less than zero, the interactions likely occur through hydrogen bonds and/or van der Waals forces. However, when $∆$H° is less than zero and $∆$S° is greater than zero, electrostatic interactions are prevalent. The present results suggest that the main interactions between vanillin and PPI, as well as DPPI, could be hydrophobic with an entropy-driven process. This inference is drawn from the fact that both $∆$H° and $∆$S° are greater than zero for both proteins. This result is consistent with previous studies, which reported that the interactions between proteins (SPI and PPI) and vanillin are driven by entropy, with hydrophobic interactions being dominant (Suppavorasatit & Cadwallader, 2012; Wongprasert, Mathatheeranan, Siripitakpong, et al., 2024).

3.4. Molecular docking of vanillin to PPI and DPPI

The potential binding pocket of vanillin on the protein monomer is located at the terminal region, as previously described by Wongprasert, Mathatheeranan, Siripitakpong, et al. (2024). It has been reported that the breakdown of proteins into monomeric forms increases the hydrophobic surface area, thereby exposing additional binding sites. This is consistent with the selected pocket in the present study, which is situated within a hydrophobic patch at the protein terminus, as also noted by Helmick et al. (2021).

The docked conformations and predicted interactions of vanillin with PPI and DPPI are shown in Fig. 3. The molecular docking results indicate that the main interactions involved in vanillin binding are hydrophobic interactions, van der Waals forces, and hydrogen bonding. The affinities predicted by AutoDock Vina (Table 5) are −3.9689 kcal/mol for vanillin bound to PPI and −3.7660 kcal/mol for vanillin bound to DPPI, indicating that vanillin exhibits stronger binding to PPI than to DPPI. Vanillin adopted slightly different conformations within the binding pocket of each protein, potentially altering the set of interacting amino acid residues. For PPI, hydrophobic interactions are observed between vanillin and residues Pro373, Val395, Ala420, and Ala422, whereas hydrogen bonds are observed with residues Gln393 and Ala422, and van der Waals forces with Met370, Val394, Val421, and Leu446. A similar interaction pattern is observed for DPPI. However, some interactions are lost or altered because of the substitution of glutamine with glutamic acid. Specifically, in DPPI, hydrophobic interactions involve residues Pro373, Val395, Ala420, and Ala422, hydrogen bonds formed with Ala422, and van der Waals forces are observed with Glu393, Val394, Val421, and Leu446.

Fig. 3.

Docked conformations and predicted interactions of vanillin within the binding pockets of PPI and DPPI. The superimposed structures illustrate the spatial orientation of vanillin in each protein. Interaction types are shown as follows: hydrophobic interactions (●), van der Waals forces (●), and hydrogen bonds (●). Both 2D and 3D representations of contact residues are included.

Table 5.

Predicted affinity values from molecular docking of vanillin with PPI and DPPI.

Protein	Affinity (kcal/mol)
PPI	−3.9689
DPPI	−3.7660

The higher affinity for DPPI suggests its weaker interaction with vanillin, which may have occurred owing to conformational changes induced by the transformation of glutamine to glutamic acid residues. This alteration likely affected the spatial arrangement of binding residues, thereby reducing interaction strength. These findings are consistent with those of Suppavorasatit and Cadwallader (2012), who demonstrated that deamidation can reduce the ability of proteins to bind with flavor compounds. Temthawee et al. (2020) also reported that partial deamidation of coconut protein using PG under optimal conditions reduced the overall binding affinity.

The computational results presented here also agree with experimental findings, which suggest that hydrophobic interactions are the dominant form of interactions in the binding of vanillin to PPI. Furthermore, the observed interaction types are consistent with those published by Wongprasert, Mathatheeranan, Siripitakpong, et al. (2024), who reported that vanillin–pea protein binding is governed by hydrophobic interactions, van der Waals forces, and hydrogen bonding.

3.5. Sensory evaluation

Binding interactions between proteins and flavor compounds are typically evaluated using instrumental techniques. However, instrumental methods cannot be used to determine the effects of these interactions on consumer perception. To address this limitation, sensory evaluation was conducted to assess the perceptual impact of vanillin binding to PPI and DPPI in an aqueous model system. The BET for each individual was calculated following the ASTM E679-04 method. This method defines BET as the average concentration where a panelist gives a consistent correct response after an initial incorrect response. Individual BETs were averaged across multiple trials to determine a personal BET, whereas the group BET was calculated as the geometric mean of all individual BETs.

The group BET values for vanillin in PPI and DPPI obtained herein are presented in Table 6. The group BET for vanillin in PPI was determined to be 31.61 μg/mL, whereas the value for vanillin in DPPI is 6.71 μg/mL. These results suggest that PG deamidation led to an approximately five-fold reduction in the group BET compared to that for PPI. Furthermore, the group BET values observed here are comparable to those reported in a previous study involving vanillin in coconut protein (33 μg/mL) and deamidated coconut protein (14 μg/mL) (Temthawee et al., 2020). In addition, the decrease in group BET values after PG deamidation is consistent with findings of a study on soy milk, which reported a reduction in group BET values after PG deamidation (Suppavorasatit et al., 2013). At the same time, the comparison of group BET values with the nK values of PPI and DPPI (1451.63 × 10⁴ and 1107.56 × 10⁴ L/mol at 25 °C, respectively) supports this conclusion. Lower nK values could imply that flavor compounds are more readily released from the protein matrix, allowing the detection of vanillin at lower concentrations in DPPI than in PPI.

The observed decreases in nK values and group BET values were attributed to deamidation catalyzed by PG, which reduces the number of available amide groups by converting glutamine residues in PPI into glutamic acid. Consequently, the reduced presence of amine groups may limit Schiff-base formation with the carbonyl groups of vanillin, thereby reducing the extent of irreversible binding (Suppavorasatit & Cadwallader, 2012).

4. Conclusions

This study demonstrated that the binding of vanillin to PPI and DPPI is temperature-dependent, with shorter equilibration times observed at higher temperatures. Binding affinity (nK) decreased with increasing temperature, and PG deamidation reduced the binding strength owing to structural changes. Despite this, PG deamidation improved flavor release, as indicated by lower odor detection thresholds and higher vanillin availability at lower concentrations. Thermodynamic analysis confirmed that vanillin spontaneously binds to both proteins, with the process being primarily driven by hydrophobic interactions. These findings were supported by molecular docking results, which revealed consistent interaction patterns. Overall, PG deamidation was shown to enhance the flavor delivery properties of pea protein, making DPPI a promising ingredient for flavor-sensitive food applications.

CRediT authorship contribution statement

Panatthida Siripitakpong: Writing – original draft, Methodology, Formal analysis. Thanakorn Wongprasert: Writing – review & editing, Formal analysis. Thanyada Rungrotmongkol: Writing – review & editing, Validation, Resources. Inthawoot Suppavorasatit: Writing – review & editing, Supervision, Resources, Methodology, Funding acquisition, Conceptualization.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability

Data will be made available on request.