Bacteria or yeast are involved in microbial fermentation of the protein-containing food to produce biopeptides. Briefly, these microorganisms secret proteolytic enzymes during their growth so that food proteins are hydrolysed into biopeptides. The production of biopeptide via microbial fermentation is straightforward. Firstly, the selected microorganism is grown into its exponential phase in the growing media at the optimum conditions. Subsequently, these microorganisms are harvested, washed, suspended in sterile media and added into the target food material as a starter to induce the fermentation and the production of biopeptides (Daliri et al., 2016; Aguilar-Toalá et al., 2017). During the fermentation process, the type of microorganism, nature of the food matrix, and fermentation conditions significantly affect the outcomes (i.e., the types and quantities of peptides released).

To date, although dairy products are still the main source of proteins for generating biopeptides using the fermentation method (Wang et al., 2015; Najafian and Babji, 2018; Worsztynowicz et al., 2020; Yu et al., 2020), the preparation and characterisation of plant-based biopeptides are drawing significant research interests (see Table 1). For instance, Xiao et al. (2018) fermented red bean [Phaseolus angularis (Willd.) W. F. Wight.] using Cordyceps militaris (L.) Fr to release small peptides. The authors observed significant inhibitory effects of the fermented bean on angiotensin I converting enzyme (ACE), with an IC₅₀ value of 0.63 mg/mL. As reported by Wu et al. (2018), whole-grain oats were fermented by Lactobacillus plantarum B1-6, Rhizopus oryzae, or their combination for 72 h, resulting in an increased degree of hydrolysis when both microorganisms were applied. In addition, the fermented oats had a higher ACE inhibitory activity, with an IC₅₀ value of 0.42 mg/mL. Enhancement of bioactivity during fermentation has been associated with degradation of the rigid cell wall of plants by microbial enzymes, resulting in enhanced protein digestibility and biopeptide release (Di Stefano et al., 2019).

Compared with the fermentation method, the hydrolysis of food proteins using enzymes to produce biopeptides has the advantages of time effectiveness, ease in scaling up, and better predictability. The general flow diagram of biopeptide preparation using this approach is shown in Figure 1. Essentially, the first step involves extraction of food proteins from the food to avoid interference of other non-protein components (Görgüç et al., 2020). For instance, enzyme activity can be significantly inhibited by phenolic compounds (Cirkovic Velickovic and Stanic-Vucinic, 2018). Subsequently, these extracted proteins are hydrolysed using the selected enzyme(s) at controlled temperature and pH for a certain period (He et al., 2016; Ou and Peng, 2016). Similar to the release of biopeptide using the fermentation method, in this enzymatic hydrolysis process, the selected enzyme(s), nature of food proteins, and environmental conditions during the hydrolysis significantly affect the outcomes. In a study by Zhang et al. (2012), rice bran protein was hydrolysed using peptidases, Alcalase, Neutrase, papaya latex papain or porcine pancreas trypsin at 25–50°C and pH 7.0–8.5 depending on the nature of the enzyme. The result showed the Alcalase hydrolysis process produced hydrolysates with the highest degree of hydrolysis. The subsequent in vitro test result confirmed a high micellar cholesterol inhibition ability of the released biopeptides. To date, although the association between specific proteolytic enzymes and the release of specific peptides in food proteins has not been completely understood, it is well-accepted that the biopeptides with low molecular weight are more bioactive than ones with high molecular weight (Ruiz-Ruiz et al., 2013; García-Tejedor et al., 2014). The third step is usually the termination of protein hydrolysis by inactivating the enzyme(s), followed by fractionation to separate hydrolysates or peptides from the reaction mixture (containing residual unhydrolysed proteins and buffer components). This separation process is usually performed using desalting and membrane-filtration technologies. Finally, the peptides can be recovered using freeze-drying technique, followed by characterisation of their physicochemical and biological properties.

In their study, Ferri et al. (2017) digested rice byproducts with five commercial proteolytic enzymes. The resulting digesta was then fractionated into four molecular weights using cross-flow membrane filtration techniques. This allowed bioactivity characterisation such as anti-tyrosinase, anti-inflammatory, cytotoxicity, irritation capacity, antioxidant, and anti-hypertensive activities to be performed. Similarly, in their studies, Nimalaratne et al. (2015) and Zhang et al. (2016) used various proteases to hydrolyse chicken egg white protein and Pseudosciaena crocea muscle, respectively. The hydrolysates were separated using desalting and ultrafiltration methods before the permeate was freeze-dried and characterised. It is worth noting that in both studies, reverse-phase high-performance liquid chromatography (RP-HPLC) was used to purify the peptide fraction further, but this was more for analytical purposes.

To date, many clinical trials have reported the correlation between the oxidative stress caused by reactive oxygen species (ROS) and the development of various chronic diseases including rheumatoid arthritis, diabetes, inflammation and even cancer (Sayin et al., 2014; Ibrahim et al., 2018). Therefore, the intake of a diet rich in antioxidants is well-recommended by clinicians, dietitians and health organisations. Meanwhile, food ingredients with antioxidant activity are also desired by the food industry, because these foods can delay and/or prevent oxidative deterioration of macromolecules in food product matrices, such as proteins and lipids, to improve the quality and shelf life of the food (Nwachukwu and Aluko, 2019).

Depending on the composition, structure, and hydrophobicity, some biopeptides show antioxidant properties to varying extents. For example, among 20 amino acids, Xu et al. (2017) observed stronger antioxidant activity in tryptophan, methionine, histidine, lysine, cysteine, arginine and tyrosine, than in others. Generally, biopeptides with antioxidant activities usually have short chains made from about 4 to 16 amino acids and low molecular weight in the range of 0.4–2 kDa.

The exact mechanism of antioxidant activities of these biopeptides is a continuing research endeavour. These peptides have been reported and used in different settings, such as inhibiting lipid peroxidation, scavenging free radicals, and chelating transition metal ions via various mechanisms (Wu et al., 2003; Rajapakse et al., 2005; Moure et al., 2006). For instance, in their studies, Ajibola et al. (2011) and Esfandi et al. (2019a,b) suggested that the scavenging of free radicals happens via hydrogen atom and single electron transfer by the tyrosine- and cysteine-containing biopeptides, respectively. In other studies, the histidine-containing biopeptides acted as antioxidants via the donation of hydrogen atom, entrapment of lipid peroxyl radical, and the chelating of metal ions by the imidazole group (Khan et al., 2014; Walters et al., 2018). Sarmadi and Ismail (2010) reported the antioxidant activity of cysteine due to sulfhydryl (-SH) groups' reacting with free radicals to form the disulfide bond (-SS-).

Some plant-based biopeptides with antioxidant properties are shown in Table 2. Other examples are given below. For instance, Megías et al. (2008) hydrolysed sunflower protein using pepsin and pancreatin and the hydrolysate exhibited significant copper-chelating properties due to the presence of histidine and arginine in the peptide sequence. Similarly, in a study by Girgih et al. (2014), hemp protein isolate was hydrolysed using pepsin, followed by pancreatin to generate biopeptides. Subsequently, the sequences of 23 peptides were identified in this protein hydrolysate and the subsequent in vitro and in vivo tests indicated the superior antioxidant properties of WVYY and PSLPA biopeptides. This indicates that gastrointestinal proteases can potentially generate antioxidant biopeptides during digestion of dietary proteins. Furthermore, Supawong et al. (2018) used Protease G6 to hydrolyse rice bran protein and the resultant biopeptides exhibited significant antioxidant activity. Subsequently, these peptides were incorporated into the fried fish cake as an example application. The authors reported that 2% (w/w) rice bran hydrolysate addition was able to reduce lipid oxidation by 79.8%, which was as equally effective as adding 0.02% (w/w) butylated hydroxyanisole/butylated hydroxytoluene (BHA/BHT). This demonstrates the potential of using biopeptides to preserve the quality of food products.

The incorporation of food ingredients with antimicrobial activity can help prolong the shelf life of the food product, as well as retain food quality during storage (Lucera et al., 2012). For this purpose, the mechanism and application of various natural antimicrobial compounds, such as essential oils from plants (e.g., basil, clove, and rosemary), enzymes from animals (e.g., lysozyme and lactoferrin), short-chain organic acids (e.g., acetic and citric acid), and biopolymers (e.g., chitosan) have been investigated and explored.

Some peptides exhibit antimicrobial activity due to their unique structural features such as having unusually long chains (with 20–46 amino acid subunits), the presence of basic groups such as lysine or arginine, and an amphipathic nature (Toldrá et al., 2018; Ahmed and Hammami, 2019). Generally, many antimicrobial peptides contain cationic amino acids and they usually have a high content of some particular hydrophobic residues such as leucine, isoleucine, valine, phenylalanine and tryptophan (Fjell et al., 2012). These biopeptides can form channels and/or pores on the surface of microbial membranes, leading to membrane disruption and even cell division (Yadavalli et al., 2016; Toldrá et al., 2018). The mechanism of antimicrobial activity of biopeptides was reported by Fjell et al. (2012); Barreto-Santamaría et al. (2020), and Li et al. (2021). Firstly, these biopeptides interact with oppositely charged groups on microbes' membrane surfaces. Subsequently, the biopeptides attach onto this surface (aided by electrostatic/hydrophobic interactions). Finally, the lipids of the microbe membrane are displaced, leading to the disruption of cell membranes.

The antimicrobial activity of various plant-based biopeptides have been reported, including peptide hydrolysate of chia flour (Salvia hispanica L.) (Segura-Campos et al., 2013), black pepper peptides with GTCVLVL, SSVVGRL, and ALGTLLL residuals (Umadevi et al., 2018), and soybean peptides containing PGTAVFL and ILAFLEATLVDLVVVLWTA residuals (Dhayakaran et al., 2016). In their study, Pu and Tang (2017) purified an antilisterial peptide (Alpep7) from the bromelain hydrolysate of rice bran proteins and identified LVDHFPL residual as being responsible for this antimicrobial activity. Furthermore, a liposome system was successfully developed to deliver this biopeptide to the listerial biofilm to confirm its listericidal activity. Similarly, Xiao and Zhang (2012) digested Jatropha curcas meal using multiple enzymes (including pepsin, trypsin, Protamex, Neutrase, Flavourzyme, papain, Alcalase, or acid protease) into hydrolysates with different degrees of hydrolysis and the authors evaluated the antibacterial activity of the hydrolysates. Based on the peptide sequence analysis, the CAILTHLR peptide was found to be responsible for the inhibitory effect against various microbes, including Escherichia coli ATCC 25922, Shigella dysenteriae ATCC 51302, Pseudomonas aeruginosa ATCC 27553, Staphylococcus aureus ATCC 25923, Bacillus subtilis ATCC 23631, and Streptococcus pneumoniae ATCC 49619. The minimum inhibitory concentrations were in the range of 29–68 μg/mL.

To date, many synthetic anticancer drugs show a wide range of side effects, such as nephrotoxic, neurotoxic, cardiotoxic and gonadotoxic effects (Oun et al., 2013; Ahar et al., 2014; Kamisli et al., 2015; Gutierrez et al., 2016; Van Acker et al., 2016). Therefore, biopeptides with anticancer activity are promising alternatives to prevent and/or decelerate cancer (Daliri et al., 2017). Some examples of plant-based biopeptides with anticancer activity are shown in Table 3. In a review by Chalamaiah et al. (2018), different mechanisms of anticancer activity of these peptides were reported. Some of the mechanisms included inducing cancer cell membrane damage, adhering to cell and inhibiting topoisomerases, modulating immune response, and/or inhibiting intracellular signalling of cancer cells.

Xue et al. (2015) reported that chickpea protein-derived peptide RQSHFANAQP dramatically increased p53 protein level, thereby inhibiting the proliferation of the MCF-7 and MDA-MB-231 breast cancer cells at the EC₅₀ values of 2.38 and 1.50 μmol/mL, respectively. In another study, based on a 2-day fermentation of rapeseed meal using Bacillus subtilis and Actinomucor elegans, the released biopeptides significantly inhibited the proliferation of human HepG2 liver cancer and MCF-7 breast cancer cells (Xie et al., 2015).

Among the plant-based biopeptides with anticancer benefits, those obtained from soybean proteins have been attracting particular research interests (Badger et al., 2005; Hwang et al., 2011). In a study by Rayaprolu et al. (2013), high oleic acid soybean protein from N98-4445A and S03-543CR lines was hydrolysed using Alcalase and the released biopeptides inhibited the proliferation of colon, liver and lung cancer cells. Moreover, anticancer activity has been observed in several other biopeptides, such as lunasin (a peptide found in soy), RLQLQGVN, GLTSL, LSGNL, GEGSGA, MPACGSS, and MTEEY peptides (Vital et al., 2014; Fernández-Tomé et al., 2017). Lunasin is a peptide with 43 amino acid residues and a molecular weight of 5 kDa (Udenigwe and Aluko, 2012; Rizzello et al., 2016). Within this peptide, the RGD residue and polyaspartic acid chain with nine aspartic acid residues have been reported as being responsible for its anticarcinogenic activities (Dia and de Mejia, 2011). In an in vitro study by Lumen (2005), lunasin showed the ability to enter mammalian cells within a few minutes when orally ingested, followed by localisation in the nucleus to inhibit histone acetylation and prevent skin cancers. Similarly, in vitro and in vivo studies have confirmed the inhibitory effect of lunasin on murine Lewis lung carcinoma (LLC) and B16-F0 melanoma cells, but do not affect the growth rate of normal cell lines.

Cholesterol levels of around 50 mg/dL in the serum are required by the human body to maintain normal functions (Steinberg and Witztum, 2009). However, excessive amounts of cholesterol in the blood may result in the formation of plaques in the arteries, resulting in cardiovascular diseases (Daliri et al., 2017). Although many synthetic drugs have been developed to lower blood cholesterol, side effects including liver injury or failure, myopathy and diabetes have been reported from their consumption (Carter et al., 2013; Mancini et al., 2016). Therefore, there is an increasing demand in sourcing natural and food-based alternatives of these drugs.

Hypocholesterolemic activity has been observed in some plant-based biopeptides, including ones from rice bran (Zhang et al., 2012), cowpea (Marques et al., 2015; Hernandez and de Mejia, 2017), cumin seed (Siow et al., 2016), and soy (Duranti et al., 2004; Lammi et al., 2015). Recently, Coelho et al. (2018) hydrolysed chia seed protein into biopeptides using Alcalase and Flavourzyme. After purification via ultrafiltration, the separated biopeptides with low molecular weight (<3 kDa) showed a significant inhibitory effect on cholesterol synthesis in vitro, by reducing the enzymatic reaction velocity of 3-hydroxy-3-methylglutaryl coenzyme A (HMG-CoA) reductase by up to 80.7%. Similarly, Lammi et al. (2016) reported that lupin biopeptides produced by enzymatic hydrolysis significantly inhibited the ability of HepG2 cells to secrete mature proprotein convertase (PC) subtilisin/kexintype 9 (PCSK9) in patients with moderate hypercholesterolaemia.

Some soy protein hydrolysate or biopeptides also exhibit hypocholesterolemic activities. For example, Lammi et al. (2015) treated HepG2 cells with three peptides separated from hydrolysed soy glycinin. The peptides were IAVPGEVA, IAVPTGVA and LPYP. The result showed that these three biopeptides significantly affected the catalytic activities of HMG-CoA reductase. Moreover, the cholesterol metabolism was modulated by the three peptides based on the activation of low-density lipoprotein receptor-sterol regulatory element-binding protein 2 (LDLR-SREBP2) pathway, finally promoting cellular uptake of low-density lipoprotein by the cultured hepatocytes. In an in vitro study by Pak et al. (2005), the hypocholesterolaemic activity of soy 11S globulin peptide was observed based on the binding of Bile acid. This biopeptide has the amino acid sequence of IAPGEVA, and a molecular weight of 755.2 Da. After bile acid was bound with peptide, it could not be reabsorbed through the enterohepatic circulation, which further stimulates the transformation of cholesterol into bile acids in the liver.

The overall prevalence of hypertension (i.e., high blood pressure) in the global population is estimated to be 1.13 billion people in 2015 and it is expected to be approximately 1.56 billion by 2025 (Kearney et al., 2005; Mills et al., 2016; World Health Organisation, 2021). The correlations between hypertension and many serious non-communicable diseases have been well-established (Lee and Hur, 2017). Therefore, the demand for functional foods or food ingredients with antihypertensive activity is increasing (Bhat et al., 2017). Unfortunately, many synthetic antihypertensive drugs have multiple side effects, including dizziness, dysgeusia, headache, angioedema, and cough (Daliri et al., 2017).

In the renin-angiotensin-aldosterone system of the human body, angiotensin I-converting enzyme (ACE) catalyses the conversion of angiotensin I to angiotensin II, finally increasing blood pressure. Therefore, compounds that influence this system have been drawing much research attention. Interestingly, some peptides have shown ability to lower blood pressure by inhibiting the activity of ACE. Piovesana et al. (2018) reported that ACE-inhibiting biopeptides usually have 2–12 amino acid subunits in the chain. Particularly, they often contain acidic (Aspartic and Glutamic acid), positively charged (in particular, an alkyl group at the C-terminus) and hydrophobic amino acid subunits. This is also confirmed in the study by Wu et al. (2006), where the correlation of ACE-inhibiting activity and structure of 168 di- and 140 tripeptides were investigated.

In a study by Marambe et al. (2008), flaxseed proteins hydrolysed using Flavourzyme (0.67 mg/mL) and the hydrolysate showed strong ACE inhibitory activity at a low IC₅₀ values of 0.07 mg/mL. Similarly, Wang et al. (2017) hydrolysed protein derived from rice bran using trypsin to obtain YSL peptide with a molecular weight of 395 Da, which exhibited strong ACE inhibitory activity at an IC₅₀ value of 76 μM. Li and Aluko (2010) hydrolysed pea protein using Alcalase and the separated IR, LF and EF peptides showed significant inhibition of the activities of ACE and renin. These peptides had IC₅₀ values below 25 mM. ACE inhibitory effects have been observed in biopeptides from soy (Wu and Ding, 2002), bamboo shoots (Liu et al., 2013) and cocoa beans (Sarmadi et al., 2011). Interestingly, Roy et al. (2010) suggested that the pea-based ACE inhibitory peptides were more resistant to digestion than the ones derived from milk.

The immune system of the human body is comprised of many biological structures and a healthy immune system that can identify and kill invading microorganisms (Yang et al., 2018). However, this vital system can be negatively affected by a wide range of factors including stress, unhealthy diet and lifestyle, and overwhelming presence of pathogens and/or antigens (Segerstrom and Miller, 2004). Although some drugs have been developed to modulate the human immune responses, they also have disadvantages like toxicity and high cost (Gertsch et al., 2011). This makes the modulation of the immune system via the intake of food-based compounds promising.

Immunomodulatory activities have been observed in some plant-based biopeptides, including the ones from soybean, wheat, yellow pea seed and rice (Morris et al., 2007; Egusa and Otani, 2009; Ndiaye et al., 2012; Hartati et al., 2017; Wu et al., 2017). Some examples of these biopeptides are shown in Table 4. Although the exact mechanism(s) of how biopeptide affects the immune system has not been fully understood, Chalamaiah et al. (2018) suggested that some peptides exhibit this bioactivity via activating macrophages, stimulating phagocytosis, increasing the amount of leukocytes, improving immune modulators (e.g., cytokines, nitric oxide, and immunoglobulins), stimulating natural killer cells, and enhancing the stimulation of splenocytes, CD4+, CD8+, CD11b +, and CD56 + cells.

Udenigwe et al. (2009) used pepsin, ficin and papain to hydrolyse flaxseed proteins, followed by releasing and separating biopeptides with a molecular weight below 1 kDa. These peptides showed significant inhibitory effects on the production of nitric oxide induced by lipopolysaccharide in RAW 264.7 macrophages. No cytotoxicity was observed. On the other hand, Kong et al. (2008) hydrolysed soy protein using various enzymes to produce low-molecular weight peptides (<1,000 Da), which improved lymphocyte proliferation and phagocytosis of peritoneal macrophages in mice. Interestingly, the authors observed the positive correlations between the immunomodulating activity and content of positively charged peptides.

Opioids such as morphine have been used as drugs due to their pain modulation functions (Bagley and Ingram, 2020). However, their long-term use leads to the development of tolerance and opioid use disorder. Therefore, biopeptides with opioid-like activity can be good alternatives to these drugs. Toldrá et al. (2018) described that YGGF and YP residuals as the common motifs present in most opioid peptides. Based on the study in the fundamental signalling mechanisms of opioid receptors, it is generally accepted that the three types of opioid receptors (μ, δ, κ) are activated by endogenous peptides derived from three different precursors, namely proopiomelanocortin, proenkephalin, and prodynorphin (Liu and Udenigwe, 2019). Unlike the animal-based opioid peptides which bind to μ receptors, the plant-based ones usually interact with δ receptors, except for soymorphins (Yoshikawa et al., 2003).

To date, researchers have identified opioid peptides from various plant proteins, including wheat (gluten, gliadin, and glutenin), barley (hordein), maize (zein), oats (avenin), rye (secalin), soybean (soya α-protein and cytochrome b), and spinach (rubiscolin) (Kaur et al., 2020). Leo Pruimboom and de Punder (2015) overviewed the degradation of gluten in the gastrointestinal tract and suggested that gluten exorphins, a morphine-like substance can be released. Following this, in Garg's et al. (2018) study, high opioid activity was observed in the gluten peptides containing YPG, YYPG, and YIPP motifs. Similarly, Yang et al. (2001) demonstrated that YPLDL and YPLDLF peptides identified in pepsin-digested spinach D-ribulose-1,5-bisphosphate carboxylase/oxygenase (RuBisCo) showed opioid activity in mouse vas deferens (MVD) assay.

Generally, most food proteins are not bitter. However, after proteolysis using some enzymes, the resultant hydrolysates or peptides may have bitterness. This bitter taste is due to the release of hydrophobic amino residues, which are usually buried within the native protein structure (Acquah et al., 2018) but get exposed after hydrolysis of the protein. For example, phenylalanine, tyrosine, tryptophan, and leucine, which are usually buried within the molecule, have bitter tastes in nature. During the hydrolysis of the proteins, these hydrophobic amino acid residues are gradually exposed and their bitterness can be detected by human bitter taste receptors (Maehashi and Huang, 2009). Generally, the bitterness of biopeptides intensifies with increase in degree of hydrolysis, since more hydrophobic amino acids are released. However, bitterness decreases when the proteins are intensively hydrolysed into peptides with low molecular weight or into free amino acids (Adler-Nissen, 1986; Fu et al., 2018). When native food protein is hydrolysed into peptides, it has been reported that the peptide bitterness intensity usually increases with the increase of hydrolysis degree and decrease of molecular weight, until the hydrolysis gives peptides with seven or fewer residues (Kim and Li-Chan, 2006; Maehashi et al., 2008; Kohl et al., 2013). Cho et al. (2004) investigated the correlation between the bitterness of soy protein hydrolysate and peptide molecular weights and showed that the hydrolysate with the strongest bitterness had a molecular weight of 4 kDa, while the least bitter hydrolysate was the one with a molecular weight of below 1 kDa. Contradictory results have also been reported. Humiski and Aluko (2007) found that the bitterness in pea hydrolysates was not related in any way to molecular weight.

In order to quantify the bitterness of biopeptides using their structures, Ney (1979) developed the Q-rule to illustrate and predict the bitterness of a peptide. Briefly, the average hydrophobicity, Q, is calculated by summing the amino acid side chain hydrophobicity of a peptide and dividing this value by the number of amino acid residues in the peptide. A mathematical representation of the Q value is as follows:

where Q is the average hydrophobicity of a peptide (cal/mol), Δf is the free energy of transfer of the side chains in the amino acid residues (hydrophobicity, cal/mol) and n is the number of amino acid residues (dimensionless).

According to this method, peptides with Q value over 1,400 cal/mol and molecular weights below 6 kDa are likely to be bitter, while those with Q-value below 1,300 cal/mol and molecular weights below 10 kDa should not be bitter. This principle was successful to a certain extent in interpreting the bitterness property in some peptides or hydrolysate from casein and soy (Murray et al., 2018; Iwaniak et al., 2020). However, there is an increasing number of studies confirming the correlation of amino acid position and their sequence and the bitterness property of peptides. For example, Kim et al. (2008) used computer simulation to study the structures of NALPE peptide and its 6 analogues. The results showed that the intensity of peptide bitterness could also be affected significantly by spatial orientation of hydrophobic regions in the structure, as well as proximity between polar groups and hydrophobic regions within the same plane space.

Mathematical models can be used to predict the bitterness of peptides. For instance, quantitative structure-activity relationship (QSAR) models were developed for this purpose, based on physicochemical properties of peptides, such as electronic charge, hydrophobicity and steric properties (Iwaniak et al., 2015; Agyei et al., 2016). Furthermore, Yin et al. (2010) introduced a descriptor, E, to the QSAR model, based on the multidimensional scaling of 237 physicochemical properties of the natural amino acid side chains. This improved the model's success in predicting the bitterness of 48 dipeptides (R² = 0.97). Kim and Li-Chan (2006) used a database of 224 di- to tetradecapeptides and five amino acids to investigate the correlation between peptide structure and bitterness. The results showed that bulky hydrophobic amino acids at the C-terminus and bulky basic amino acids at the N-terminus of peptides significantly impacts peptide bitterness. Later on, Soltani et al. (2013) developed several models, including multiple linear regression, support vector machine, and artificial neural network models, to quantify the correlation between peptide structures and bitterness.

Because biopeptides are promising food ingredients, the importance of detecting and quantifying the perception of bitterness in food biopeptides cannot be underestimated. So far, methods for the detection and quantification of bitterness in biopeptides involve sensory evaluation, use of databases, electronic tongue, and/or calcium imaging technique.

Since the exposed hydrophobic amino acid residues during food proteins hydrolysis result in bitterness in peptides, the separation of those bitter peptides from the hydrolysate seems like a reasonable approach.

Based on the π-π stacking interactions with aromatic side chains of peptides, activated carbon or macroporous resin can be used for debittering peptides (Clark et al., 2012). The separation efficiency depends on the affinity of the peptides to the adsorptive material. For example, 98.4 and 64.5% of phenylalanine was removed from whey protein hydrolysate using macroporous resin and activated carbon columns, respectively (Bu et al., 2020). Also, some bitter peptides can be removed by extraction using selected alcohols. Sinthusamran et al. (2020) studied the efficacy of different alcohols on decreasing the bitterness of salmon frame protein hydrolysates. The authors reported that 2-butanol worked more efficiently than iso-propanol in lowering the hydrolysates hydrophobicity and their bitterness intensity. Later on, 2-butanol and β-cyclodextrin were further used in combination to successfully debitter salmon frame protein hydrolysates (Singh et al., 2020).

However, it is worth noting the disadvantage of debittering peptides via the separation of hydrophobic amino acid residues. This process has been associated with the loss of essential hydrophobic amino acids and low processing efficiency. Moreover, the reported studies only focused on the development of debittering process and the bioactivity of debittered peptides still needs to be investigated.

The activity of a particular protease significantly affects the degree of hydrolysis of a protein, which is associated with the bitterness of peptides. For example, compared with the pea protein hydrolysed by papain, trypsin, bromelain and chymotrypsin, the hydrolysate prepared using Alcalase and Esperase had higher degree of hydrolysis and bitterness intensities (Arteaga et al., 2020). Similarly, soy protein hydrolysed by Alcalase was more bitter than the ones treated by Neutrase, papain, Corolase, and Flavourzyme (Meinlschmidt et al., 2016a). Generally, based on the mechanism of action and catalytic sites, proteases can be categorised into two groups: exopeptidases (E.C.3.4.11–3.4.19) and endopeptidases (E.C.3.4.21–3.4.99). Exopeptidases cleave free amino acids and/or low molecular weight peptides from the end of a polypeptide chain while endopeptidases act within a polypeptide chain (Stressler et al., 2015). Exopeptidases exhibit lower activity against intact protein molecules than endopeptidase and some exopeptidases show specificity toward hydrophobic amino acid residue at the N-terminal (Stressler et al., 2019). Therefore, exopeptidases can be used to cleave the bitter hydrophobic amino acid residue exposed during the protein hydrolysis by endopeptidases. For this purpose, aminopeptidases have been commonly used (Raksakulthai and Haard, 2003).

Microorganisms are a promising source of aminopeptidases and, to date, over 100 aminopeptidases have been identified from the genus Lactobacillus. Generally, aminopeptidase N (PepN, EC 3.4.11.2) and the proline-specific X-prolyl dipeptidyl aminopeptidase (PepX, EC 3.4.14.11) are two aminopeptidases that have been used for debittering purposes (Gonzales and Robert-Baudouy, 1996). Briefly, PepN hydrolyses almost all the amino acids from the amino (N-) terminus of a polypeptide chain, unless proline is present at the second position (which stops the hydrolysis) (Stressler et al., 2013). PepX is a proline-specific dipeptidyl peptidase that releases X-Proline dipeptides from the amino (N-) terminus of a polypeptide chain, as long as there are no proline or hydroxyproline at the second position (Stressler et al., 2014). Recently, Ewert et al. (2018) observed the superiority of PepN in debittering sodium caseinate hydrolysates, compared with PepX or aminopeptidase A (PepA). This might be due to the high specificity of PepX and PepA. PepX releases X-Proline dipeptides from the N-terminal of peptides, as a proline-specific aminopeptidase, while PepA is specific for the hydrophilic amino acids, such as Gtamic acid, Aspartic acid, and Serine. These amino acids are not associated with bitterness. During the enzymatic treatment, PepN was capable of cleaving bitter amino acids such as Leucine, Phenylalanine, Isoleucine, or Valine from N-terminus of most peptides. The debittering potential of PepN has been reported previously in a combination with PepX (Barry et al., 2000). Moreover, since aminopeptidases are released by microbes during microbial growth (Stressler et al., 2014), some particular strains have been used directly as debittering starters to produce protein hydrolysates. For example, the addition of Lactobacillus perolens, Rhizopus oryzae, and Actinomucor elegans significantly reduced the bitterness intensity of soy protein hydrolysates (Meinlschmidt et al., 2016b).

Alternatively, cross-linking of the peptides using transglutaminase (TG) has been used to decrease their bitterness. TG catalyses intra- and intermolecular cross-linking between glutamine and lysine residues in the peptide chain (Báez et al., 2011). This increases the molecular mass of peptides so that hydrophobic groups, which cause bitterness, can be buried again in the cross-linked polypeptide chain. As reported by Song et al. (2013), TG increased the content of 1,000–5,000 Da peptides in soybean protein hydrolysates via enzymatic cross-linking. The authors observed a significant decrease in the amount of bitter amino acids and improvement in sensory profile of the crosslinked peptide. However, it should be noted that the use of transglutaminase may compromise the functionality, biological activity and/or solubility of protein hydrolysates. Therefore, this approach may not be appropriate for certain food applications (Meng et al., 2020).

Encapsulation provides an opportunity to partially or completely mask the bitterness in peptides in addition to improving their other properties such as hygroscopicity and bioavailability. To date, several techniques such as spray drying, freeze drying, spray chilling, coacervation and double emulsion have been applied to encapsulate protein hydrolysates or biopeptides. The molecular and physicochemical properties of biopeptides, such as molecular weight, conformation, electrical characteristic, polarity, and stability, are the first factors to consider when choosing an appropriate carrier, as these properties will impact loading, retention, stability, and release from the encapsulated systems (McClements, 2018). Generally, for food applications, the selected carrier should meet some criteria such as edibility, biodegradability, non-toxicity, and inexpensiveness (Mohan et al., 2015). As a result, proteins, polysaccharides and lipids have been used as protective carriers for the encapsulation of protein hydrolysates, as shown in Table 6.