Application Potential of Food Protein Modification

Proteins are essential in foods, not only for their nutritional value, but also as modulator of structure and perception of a food product. The functional behavior of a protein is inherently susceptible to physico-chemical conditions as pH, ionic strength, temperature, or pressure, making them also an unpredictable, and at the same time, opportune component in food production. Proteins are generally also industrially costly, and with increasing world population and welfare the pressure on protein-availability for food purposes gives rise to some concerns. In view of a more sustainable use of protein-sources a number of routes have been followed in the past decades that provided big steps forward in protein availability: (i) more efficient production or protein refinery methods, (ii) use of alternative protein sources, and (iii) optimized usage of protein functionality. Especially in wheat production correlations between genetic expression and functional product behavior allowed breeders to optimize cultivars for geographic location (e.g. Payne et al., 1984). Alternatively, one has the ability to express specific proteins in non-original sources, for example human milk proteins in plants, such as rice (e.g. Lonnerdal, 2002). Directed alterations in the genome of food-producing organisms can lead to changes in the primary sequences of relevant proteins and thereby introduce potentially new functionality. If sufficient quantities of the novel protein are synthesized and become admixed with the basal levels of protein in the food, the functional properties of the food system (textureformation) may become improved. Alternatively, the modified protein can be isolated for use as food ingredient. More recently, a number of proteins from less-conventional origin have been identified as human food ingredients that one has started to exploit, e.g. algae, leafs, insects, and various seeds. Successful utilization of these new proteinaceous materials has thus far been rather limited, requiring breakthroughs in extractability, their digestibility, nutritive value, and overall functional and organoleptic properties. More downstream in the process is the modulation of protein functional behavior at an ingredient level. This can be physical-chemically, enzymatically, or via chemical engineering.


Functional role of proteins as food ingredient
This paragraph will briefly summarize the functional properties of proteins as food ingredients. The molecular mechanisms of their roles in food products will be highlighted. In the next paragraph these molecular mechanisms will be discussed as target for chemical engineering. On each of these topics extensive reviews have been published which will be pointed out.

Surface properties: Emulsions and foams
Surfaces or interfaces in food products are abundantly present in terms of emulsions (oilwater interface) or foams (air-water interface). Examples of foams and emulsions in food products include ice cream, dressings or margarine. In foams and emulsions respectively air cells or oil droplets are dispersed or in an aqueous phase; the latter case also water in oil emulsions exist. As a result of their amphiphilic character, proteins, being composed of polar and non-polar amino acids, can contribute to the formation and stability of such dispersions by adsorbing to the interface and developing stabilizing films by coating the oil or air droplets and hence act as emulsifiers. The ability of proteins to induce film formation depends on a number of parameters of molecular nature which have been studied into detail in the past and is collectively governed by a net energy gain from absorbing at an interface. Milk proteins have been identified as good foaming agents as a result of their aggregation state, molecular stability and flexibility, electrostatics, and (surface) exposed hydrophobicity (Hunter et al., 1991;Luey et al., 1991;Shirahama et al., 1990;Suttiprasit et al., 1992;Waniska & Kinsella, 1985). Chemical modifications have been employed in the past to improve surface activity of less performing proteins. These studies and consequences of the used modifications will be discussed in more detail in the following paragraph. A wide range of methods has been employed in literature to study the chemical and molecular properties of proteins adsorbed at interfaces which lead to a detailed understanding of the principle forces of importance to surface activity. These methods include ellipsometry and infrared reflection absorption spectroscopy (IRRAS) which provide information on denaturation and concentration of adsorbed proteins Grigoriev et al., 2007;Martin et al., 2003;McClellan et al., 2003). Stabilization of an air-water or oilwater interface is governed by a multiple step process. First, absorption at the interface requires proteins to diffuse to the interface and their retention at the interface is governed by the kinetic barrier of absorbance which, in turn, is influenced by factors such as exposed hydrophobicity , and net charge (Wierenga et al., 2005). Effective absorption onto the interface is followed by rearrangement of the protein molecules to form a thermodynamically stable but dynamic monolayer of molecules coating the droplets (de Jongh & Graham & Phillips, 1979a, 1979b. Some publications report that the rearrangement process coincides with structural extension of the protein molecules (reviewed in MacRitchie, 1978) while other results suggest that secondary structure does not undergo variation subsequent to adsorption onto the interface Graham & Phillips, 1979c). Despite the finding that local concentrations of 150-300 mg/ml can be reached at the interface (Meinders et al., 2001), proteins are still able to laterally diffuse as was shown by thiolated ovalbumin , illustrating that chemical modification has been used to underpin the molecular mechanisms of the surface activity of proteins.

Emulsions
Emulsions consist of two immiscible liquids, oil and water, in which the droplets are termed dispersed phase and the liquid surrounding the droplets is called continuous phase. Depending on the concentrations of each liquid and the environmental conditions, oil-inwater emulsions or water-in-oil emulsions can be formed. These consist of oil droplets in a continuous water phase and water droplets in a continuous oil phase, respectively. Examples of food related emulsions are milk, vinaigrette, and mayonnaise. Emulsions are often unstable by nature and will phase separate or coalesce over extended time, or storage. To stabilize emulsions, so-called emulsifiers can be used which form a small layer on the surface of the dispersed phase, thereby physically separating the continuous phase from the dispersed phase. Such treatment will allow the incorporation of emulsified solutions in food products which can be stored over an extended time frame. Amphiphilic proteins, containing both hydrophilic and hydrophobic regions, are effective emulsifiers. These proteins adsorb onto the interface between the oil and water phase, and stabilize the oil and water phase by selective interaction with both surfaces, thereby preventing the individual droplets from coalescing (reviewed in Rodríguez Patino et al., 2008). Other types of emulsifiers used in food products include lipids, phospholipids, surfactants or polysaccharides (Dickinson, 1992;McClements, 2005). Apart from a texture perspective, the absorption of proteins at the oil-water interface is of interest to the delivery of nutrients (reviewed in Malaki Nik et al., 2010). Upon digestion, changes occur at the oil-water interface of emulsified food products as a function of emulsifier stability, affecting the digestibility and subsequently the availability of encapsulated nutrients. The amphiphilic character of proteins can be modulated for example by the covalent attachment of lipid chains (see paragraph 3.2). Attachment of lipid chains of various lengths to a protein renders it more hydrophobic resulting in an improved ability to stabilize emulsions and/or foams (Aewsiri et al., 2011a(Aewsiri et al., , 2011b. Apart from exposed hydrophobicity, other factors have also been identified to determine the affinity of a protein for an oil-water interface. Aggregation or molecular weight of proteins for example adversely affects the emulsifying activity of proteins (Baldursdottir et al., 2010;Corzo-Martínez, 2011). That the aggregation effect is more complex than originally postulated was shown by single molecule total internal reflectance fluorescence microscopy of bovine serum albumin showing that the orientation at which the aggregates absorb to the interface largely determine the rate of absorbance to the interface (Walder & Schwartz, 2010). The identification of factors contributing to the surface activity of proteins allows the improvement of emulsifying activity by means of targeted chemical modification.

www.intechopen.com
Advances in Chemical Engineering 138

Foams
Foams consist of gas bubbles dispersed into a liquid. The stability of the air bubbles in a foam is determined by the foaming agent which forms a layer of adsorbed molecules separating the air bubbles from the continuous liquid phase, similar to the emulsifying activity described in paragraph 2.1.1 (reviewed in Halling, 1981;Wilde, 2000). Adsorption of a protein to the air-water interface induces partial dehydration of the molecule promoting protein-protein interactions. This effect is further amplified by the finding that local protein concentrations at the air-water interface can reach up to 150 to 300 mg/ml (Meinders et al., 2001). The rate of absorption to the air-water interface has been reported to depend largely on the hydrophobic nature of the protein under investigation (Kudryashova et al., 2003;Wierenga et al., 2003). Increasing the exposed hydrophobicity of proteins by means of conjugation with lipid chains was shown to increase the adsorption rate to the air-water interface . Net charge is a second parameter of interest determining adsorption kinetics, in which higher net charge slows down the adsorption process due to the electric repulsive forces involved Kudryashova et al., 2005;Le Floch-Fouéré et al., 2011). Highly aggregated heat-treated ovalbumin was further shown to induce a ten-fold decrease the diffusion rate of proteins to the interface compared to the native protein (Kudryashova et al., 2005). However, the tendency of aggregated ovalbumin to remain adhered to the interface upon first interaction is significantly larger than for nonaggregated protein, which was found to rapidly desorb from the surface after absorption (Kudryashova et al., 2005). Collectively, it has been shown that surface activity of proteins is not determined by a single molecular characteristic but rather depends on a combination of factors. Hence, several types of chemical modification can be employed to improve the airwater interface activity of proteins.

Aggregation and gelation
Protein aggregation is a major topic in the field of food science, the regulation of which is believed to markedly affect the texture of food products (Zhou et al., 2008;reviewed in Doi, 1993). Aggregated protein can act as a nucleation prerequisite to induce gelation, albeit at high protein concentration (Alting et al., 2003;Barbut & Foegeding, 1993;Ju & Kilara, 1998). Processing conditions and storage can induce protein aggregation, even at ambient temperatures (Promeyrat et al., 2010;Santé-Lhoutellier et al., 2008). The resulting protein aggregates can vary widely in size and morphology as a result of the environmental conditions under which they were formed, among other factors. For example, upon inspection using electron microscopy, amylin (Patil et al., 2011), hen egg white lysozyme (Arnaudov & de Vries, 2005), and -lactoglobulin (Arnaudov et al., 2003;Veerman et al. 2002) were found to form negatively stained and long fibrillar aggregates at pH values far away from the isoelectric point of the respective proteins. Near the isoelectric pH (Arnaudov & de Vries, 2005), or at high salt concentration (Arnaudov & de Vries, 2006;Veerman et al. 2002), spherical or amorphous aggregates are formed. That aggregate morphologies cannot always be categorized as either fibrillar or amorphous in a clear-cut manner was shown by various groups observing substantial heterogeneity in aggregate morphology, also called polymorphism, within the same preparation (Bauer et al., 1995;Jiménez et al., 2002Jiménez et al., , 1999. As a consequence, gels formed under these conditions of -lactoglobulin are particulate and the particle size depends on heating temperature and heating rate (Bromley et al., 2006).
These observations lead to the understanding that net charge plays a major role in determining aggregate and gel morphology (Krebs et al., 2009;Langton & Hermansson, 1992). Cryo-EM investigation of aggregates formed from ovalbumin which had been succinylated to various degrees provided further evidence that net charge dominantly determines aggregate morphology (Weijers et al., 2008). The propensity of proteins to aggregate, or rate of aggregation, has been shown to vary as a function of protein conformational stability (Chiti et al., 2000;Hurle et al., 1994;Kelly, 1998;Quintas et al., 1997;Ramirez-Alvarado et al., 2000;Siepen & Westhead, 2002), rate of unfolding (Broersen et al., 2007a), net charge (Calamai et al., 2003;DuBay et al., 2004), and secondary structure propensity (Fernandez-Escamilla et al., 2004). Exposed hydrophobicity (Calamai et al., 2003) and the possibility to form disulfide bonds naturally affect the aggregation and gelation propensity of proteins as these two forces are primarily driving the assembly process. This has been exemplified in a study which showed that -lactoglobulin A modified with Nethylmaleimide largely resisted aggregation induced by heating . Extensive knowledge of the molecular factors driving the aggregation process of proteins has lead to the development of a number of algorithms able to predict protein aggregation with high fidelity DuBay et al., 2004;Fernandez-Escamilla et al., 2004;Maurer-Stroh et al., 2010).
The energetics and kinetics of protein aggregation have been subject of many publications to date and has been reviewed on numerous occasions (e.g. Luheshi & Dobson, 2009;Straub & Thirumalai, 2011). It has been recognized that, regardless of primary sequence or physicochemical properties, all proteins have an inherent tendency to form aggregates in vitro under certain conditions (reviewed in Chiti & Dobson, 2006;Dobson, 1999). Protein aggregation is a multiple step complex process which can be viewed as a cascade of steps of assembly which may vary in molecular detail as a function of the protein studied or the environmental conditions. Nevertheless, the aggregate growth mechanisms of many different proteins share essential characteristics which have been elucidated both by experimental and computational methods (Teplow et al., 2006). The onset of aggregation often requires the (partial) unfolding or conformational rearrangement of proteins (Calamai et al., 2003;Dobson, 1999;Kelly, 1996;Rochet & Lansbury, 2000). Using human lysozyme (Canet et al., 1999) andHypF-N (Marcon et al., 2005) it has been shown that a population of less than 1% of partially folded protein can be sufficient to trigger the onset of the aggregation process. The resulting exposure of hydrophobic regions which normally reside in the core of a folded protein drives the self-assembly process to form small oligomers. These oligomers, or nuclei, are metastable and their transient and short-lived nature dictates that they can dissociate into monomeric protein, which has been shown for various proteins. Many efforts in the field of protein aggregation suggest that the nucleus has to reach a critical size which then allows further assembly by monomer addition to ultimately form mature fibrils or aggregated networks (Jarrett & Lansbury, 1993;Lomakin et al., 1996Lomakin et al., , 1997Sorci et al., 2011). An alternative scenario is the formation of intermediate protofibrils along the pathway which subsequently assemble into fibrils (Harper et al., 1997;Walsh et al., 1997). Mature fibrils have classically been viewed as the stable end-stage of the aggregation process which are not susceptible for dissociating conditions. Recently, using calorimetric methods (Morel et al., 2010), molecular dynamics simulations (Zidar & Merzel, 2011), and mechanical deformation studies (Paparcone & Buehler, 2011;Paparcone et al., 2010;Xu et al., 2010) it has been observed that fibrils can be dissociated albeit at high temperature.
Detailed knowledge of the molecular parameters determining aggregation propensity, rate and morphology resulted in the ability to tune protein aggregation through chemical engineering. The attachment of sugar chains to proteins has been shown to inhibit selfassociation (Marquardt & Helenius, 1992;reviewed in Helenius et al., 1997;Land & Braakman, 2001;Song et al., 2001). This effect was largely attributed to covalently linked sugar moieties affecting kinetic partitioning between folding and aggregation from an (partially) unfolded state. For example, glycosylation was found to increase the folding rate of the protein rapidly shielding exposed hydrophobic regions which could potentially act as a driving force for aggregation at ambient temperature (Broersen et al., 2007b;Shental-Bechor & Levy, 2008;Wang et al., 1996). Interestingly, high temperatures induced more rapid aggregation of glycosylated proteins compared to their non-glycosylated counterparts (Broersen et al., 2007b;Chobert et al., 2006). However, another study involving glycated bovine serum albumin concluded that glycation of the protein inhibited its aggregation upon incubation at moderate temperatures (Rondeau et al., 2010(Rondeau et al., , 2007. Glycosylation of proteins has also been shown to affect gel properties: attachment of a ketohexose to ovalbumin by means of the Maillard reaction resulted in the formation of gels with enhanced breaking strength (Sun et al., 2004). The effects of various types of chemical protein engineering on the physico-chemical functionality of proteins will be discussed in more detail in paragraph 3 of this review.

Protein structural integrity
Proteins in food products can lose their native structure as a result of processing conditions including storage, heat treatment, acidification, dehydration, mechanical processing or shear, and microbial hydrolysis. For example, long-term storage of milk powder has been found to induce lactosylation of the proteins present in the preparation which, in turn, results in affected powder solubility and emulsifying and foaming properties (reviewed in Thomas et al., 2004). Unfolded or hydrolyzed protein molecules can exert very different functionality to food products compared to folded proteins, a classical example being the boiling of an egg which converts the liquid-like transparent egg white into an opaque semisolid structure with very different textural properties. This paragraph will shortly discuss the principles of protein folding and structure and the forces that are implied. It was first recognized by Anfinsen (1973) that the primary sequence of a protein dictates the specific folded, or native, conformation a protein will assume to allow functional activity. Following urea-induced denaturation of ribonuclease A, the protein was allowed to refold by removal of urea. The protein was found to regain its native structure and functionality after this treatment suggesting that proteins can adopt their native conformation spontaneously (Anfinsen, 1973). This finding was awarded with the Nobel Prize in Chemistry in 1972 and opened up an avenue of experimental and theoretical work in the field of protein folding and unfolding. The structural insights into the folding and unfolding processes of many proteins have since then been explored using a vast range of biophysical instrumentation, both at the ensemble (reviewed in Sanchez-Ruiz, 2011) as well as at single molecule (reviewed in Borgia et al., 2008;Ferreon & Deniz, 2011) level. Protein conformational stability can be defined as the ability of the natively folded structure of a protein to resist unfolding. Two types of stability can be distinguished: the difference in energy content between the folded state and unfolded state of a protein is termed thermodynamic stability. Boltzmann's distribution law defines the distribution of folded www.intechopen.com Application Potential of Food Protein Modification 141 and unfolded protein molecules at a certain time. The rate at which the unfolded protein collapses to a folded state is reflected by the folding rate and represents the kinetic stability of a protein. Energy landscape theories and the folding funnel hypothesis have both been used as models to understand the energetic barriers of a protein separating the folded from the unfolded state (reviewed in Onuchic et al., 1997;Plotkin & Onuchic, 2002;Wolynes, 2005). Both models start from a similar principle in which there is an energy difference between the folded state of a protein and its unfolded state and that the folded state is defined as the favored entropic state of a protein. Local energy minima in the folding process can result in the accumulation of transient intermediate structures, which are neither folded, nor unfolded, to a larger or lesser extent, depending on the environmental conditions of folding or the primary sequence of the protein (reviewed in Baldwin, 2008;Englander et al., 2007). These intermediate structures, which often lack biological activity, are sometimes sufficiently stable to allow substantial accumulation (reviewed in Englander et al., 2007). Stable folding intermediates are related to a high propensity of aggregation as the hydrophobic core is not sufficiently shielded while opposing assembly forces are absent (reviewed in Ferreira et al., 2006). Kinetic protein stability is defined by a variety of molecular parameters including the proximity of native contacts in the primary sequence of a protein (Cieplak et al., 2004;Plaxco et al., 1998Plaxco et al., , 2000, internal friction or the energetic of intrachain interactions, energy barriers to backbone rotations and long-range residue interactions (Pabit et al., 2004;Qiu & Hagen, 2004), rate of diffusional motion of an unfolded peptide chain through the solvent (Pabit et al., 2004), and the presence or absence of intermediate state(s) (Baumketner, 2003;Onuchic et al., 1997). Forces that contribute to thermodynamic stability are the strength of intramolecular hydrogen bonds and solvent-protein interactions, both enthalpic in nature. The entropic contribution is mainly defined by the hydrophobic effect of folding through an increase in disorder of water molecules upon folding. Many of the forces retaining a protein structure intact can be disrupted, removed or introduced by chemical modification. For example, reaction of proteins with sugars by means of the Maillard reaction can lead to distinct changes in protein stability. The stability of proteins in food products can be affected by using proteins which have been modified by means of glycosylation or charge modification. Succinylation with the aim to increase net charge of soy protein hydrolysate lead to improved digestibility of the protein as investigated by a multienzyme method involving trypsin, chymotrypsin, and peptidase (Achouri & Zhang, 2001). This finding suggests that the protein had undergone structural rearrangement as a result of the succinylation process. Succinylation also lead to destabilization of Faba bean legumin (Schwenke et al., 1998). This group used differential scanning calorimetry (DSC) to study protein stability of legumin and found a decreased specific enthalpy for unfolding upon succinylation. Succinylation also resulted in an increased surface hydrophobicity of the protein suggesting at least partial unfolding of the molecule. Kosters and colleagues (2003) compared the effects of many different types of chemical modification on protein stability, including lipophilization using capric acid, glycosylation and succinylation of ovalbumin. Ovalbumin stability was probed by DSC, tryptophan fluorescence and circular dichroism (CD). Lipophilization resulted in a decreased denaturation temperature of ovalbumin reflected in an enthalpy decrease and a lower stability upon guanidine titration. Glycosylation was found to stabilize structural integrity of ovalbumin. A similar finding using CD and DSC to study temperature-induced unfolding has been reported upon glucosylation (Broersen et al., 2004;van Teeffelen et al., 2005) and fructosylation (Broersen et al., 2004) of lactoglobulin or glucosylation of codfish parvalbumin (de Jongh et al., 2011). Interestingly, assays performed at ambient temperature but involving denaturant-induced unfolding reported a decrease in protein stability upon conjugation of glucose to -lactoglobulin. This phenomenon, which appeared unique for a glycosylation reaction, has been studied in further detail by van Teeffelen and colleagues (2005). The observations could be explained in terms of a decreased change in heat capacity upon unfolding as a result of glucosylation indicating that the hydration pattern of proteins upon glycosylation is significantly affected.

Shelf life
Proteins can affect shelf life and stability of food products by enhancing antioxidant activity, affecting gas exchange, antimicrobial activity or by stabilization of emulsion or foam-based food products (del Rosario Moreira et al., 2011;Emmambux et al., 2004;Mendis et al., 2005). Protein films can be used as as packaging biomaterials as a result of their ability to form networks with rheologically advantageous characteristics (Arvanitoyannis, 1999;Audic & Chaufer, 2005, Longares et al., 2005. However, mixtures of for example proteins and polysaccharides have been found to exert superior functional properties compared to proteins or polysaccharides in isolation (reviewed by Pogaku et al., 2007). For example, the application of an edible coating of storage proteins obtained from sorghum, called kafirins, has been shown to extend the shelf life of freshly harvested pears . The shelf life of meat was shown to be extended upon application of a collagen and gelatin coating and led to reduced decoloration, antioxidant activity, and reduction of microbial spoilage (Havard & Harmony, 1869). Such coatings extend quality and shelf-life by acting as a slow-release gas barrier (Baldwin, 1994;Park, 1999). Nanobiocomposites of maize prolamin protein zein have also been employed as a gas barrier by coating tomatoes (Park et al., 1994), and apples (Bai et al., 2003). Even though collagen and gelatin coatings were reported to both effectively retain water in meat products (Antoniewski et al., 2007;Farouk et al., 1990), extensive moisture loss of kafirin coated pears compared to the uncoated product left them unacceptable toward consumers . The authors  therefore suggested to prepare wax or triglycerides/kafirin mixtures instead of pure kafirin coatings to prevent moisture loss as kafirin films themselves do not function effectively as water barriers (Emmambux et al., 2004;Gillgren & Stading, 2008). Because lipids form a very suitable moisture barrier as a result of their hydrophobic character, lipophilized proteins possibly form more effective coatings for fruit. However, to date, no work has been published to demonstrate the effect of lipid-incorporation into proteins to prepare stable coatings for fruit. Some proteins and peptides are known to have antimicrobial activity (Nizet, 2006;reviewed by Wimley, 2010). Some of these are also applied as food preservatives such as nisin, which is a potent antibacterial 34 amino acid peptide containing a number of uncommon amino acids. Nisin has been employed as an approved food preservative in cheese , fish, meat, and beverages (reviewed in Lubelski et al., 2008). Another known antimicrobial peptide is -poly-L-lysine which exhibits antimicrobial activity against bacteria and fungi and is used as a food preservative (reviewed in Hamano, 2011). This asset has also been explored within a food based environment, for example by applying mixed formulations of chitosan, a linear polysaccharide, and casein polymers to a number of food products including carrot, cheese, and salami (del Rosario Moreira et al., 2011). Pure caseinate films applied to squash slices showed limited antimicrobial activity (Ponce et al., 2008). While chitosan alone exerts significant anti-microbial activity, the inclusion of casein polymers into the formulation improved bactericidal properties even more (del Rosario Moreira et al., 2011) as a result of the ionic interaction between the two biopolymers (Pereda et al., 2008(Pereda et al., , 2009). Cao-Hoang and colleagues (2010) produced a nisin-containing sodium caseinate film to investigate the antimicrobial activity of both surface-and in-depth Listeria innocua inoculated soft cheese. The presence of the film reduced surface contamination with L. Innocua significantly, while antimicrobial activity within the cheese matrix depended on the distance from the film-coated surface. Antimicrobial films prepared from a mixture of oregano oil and whey protein isolate showed inhibition of growth of lactic acid bacteria, reduction of pseudomonads, total flora and growth rates when applied to fresh beef (Zinoviadou et al., 2009). Even though in many cases complex formulae have been employed, containing both a protein component as well as a carbohydrate or oil component, no publications are known that show the effects of covalently linked lipid or sugar to protein films as a potential edible film.

Sensory: Color, flavor, odor, texture
Sensory aspects of food products include sensation of flavor, odor, color, and texture. These factors play a large role in consumer acceptation of food products and the effects of various types of protein chemical engineering and their applications will be discussed in this paragraph. Many types of modification target the amino groups of lysine residues, including succinylation, lipidation and glycosylation through the Maillard reaction. Textural properties, including emulsifying, foam and gelling capacities, have been discussed in detail in paragraphs 2.1 and 2.2.

Flavor
Even though most proteins are tasteless, ingestion of a small number of proteins is perceived as sweet. These include thaumatin (Ohta et al., 2008;van der Wel & Loeve, 1972), monellin (Morris & Cagan, 1972), brazein (Ming & Hellekant, 1994), and lysozyme (Masuda et al., 2001). The sweetness of lysozyme results from the abundance of lysine residues which was shown by alanine substitution in lysozyme (Masuda et al., 2005a). It is therefore perceivable that modification of lysine residues by conjugation of a chemical group has consequences for the sweetness of the protein. Extensive acetylation and phosphopyridoxylation of lysine residues of lysozyme decreased the perceived sweetness of the protein further demonstrating that lysine residues play a major role in sensory aspects of this protein (Masuda et al., 2005, Kaneko & Kitabatake, 2001. No other reports on flavor modulating aspects as a result of chemical protein engineering have been reported.

Odor
One publication which studied the effect of acylation by acetic and succinic anhydride of flaxseed protein isolates reported no off-odors upon modification, although no results were presented (Wanasundara & Shahidi, 1997).

Color
Three types of modifications have been reported to affect the color of the protein preparation. First, succinylation was reported to convert the color of soy isolate from tan to chalk-white upon visual inspection (Franzen & Kinsella, 1976a). Upon measurement of surface reflectance using a colorimeter, increasing degrees of succinylation of flax seed protein lead to brighter protein preparations (Wanasundara & Shahidi, 1997). A similar observation has been reported for succinylation of fish muscle (Groninger, 1973), alfa-alfa leaf protein (Franzen & Kinsella, 1976b), and soy bean protein preparations in a U.S. patent for coffee whitener (Melnychym & Stapley, 1973). Acylation by treatment of protein with acetic anhydride also lead to brighter flaxseed protein isolate, although the effect of succinylation of the same protein was stronger (Wanasundara & Shahidi, 1997). Franzen and Kinsella (1976a) showed no effect on color upon acetylation of soy isolate. A third type of modification with a strong effect on color is glycosylation through the Maillard reaction. This reaction is a complex cascade of reactions initiated by the interaction of a reducing sugar with an amino group. Colored products are formed only at later stages of the reaction pathway and include aldols and melanoidins which are high molecular weight compounds (reviewed in Zamora & Hidalgo, 2006). Paragraph 3.1 further extends on the formation of browning products and Amadori compounds related to the Maillard reaction.

Texture
Texture is a complex consumer perception of mouthfeel, tastants and afterfeel. A number of most relevant texture attributes, like 'spreadability' (essential for attributes like 'creamy'), 'crumbliness' or 'separating/wateriness' have been shown to be directly related to the energy household in protein-based products when energy is exerted onto the system. This applied energy may arise from oral processing, exposure to heat, gravity or applied pressure during for example industrial processing. The energy balance in protein-based food structures has been proposed by van Vliet and Walstra (1995): W applied = W stored + W dissipated + W fracture. This model implies that all energy applied to the gel can be used either for fracture, can dissipate or can be stored in the network (and regained after release of exerted forces). At a microstructural level this translates into fracture nucleation points and fracture propagation modes, whereas energy dissipation is often assumed to be controlled by serum flow properties. This latter factor is directly determined by the porosity of the gel as set-up by the microstructure and the pore deformation propensity when the system is put under strain. Van den Berg and colleagues (2008) showed that an attribute like 'spreadability' is directly related to directing as much energy as possible to fracture. When energy can be efficiently stored in the protein matrix, this directs the attribute 'crumbliness'. The effective interaction between protein-building blocks that make up the product matrix can be engineered. Strengthening this interaction, using for example transglutaminase (e.g. Dondero et al., 2006), will direct the energy flow from fracture to storage and gels will become less spreadable and more crumbly. Sala and co-workers (2008) showed that by modulating the interaction between a filler (like a fat-particle) and the protein matrix the texture of these protein gels could be strongly affected. This balance between active-inactive property of the filler could be delicately controlled by means of chemical modification.

Types of chemical protein engineering -Exploring potential functionality
Proteins can be chemically modified in vitro by covalently cross-linking the protein with a molecule of interest. Many of these reactions will also take place post-translationally in the strictly regulated environment of the cell, such as phosphorylation of cAMP-dependent protein kinases which plays a role in the enhancement of glycogen degradation (Soderling, 1975). However, this paragraph will entirely focus on the kind of chemical engineering intentionally brought about to link specific molecules to proteins which act as functional ingredients in food. These molecules change the behavior of the protein and are largely hypothesized to infer characteristics to the protein which are little present in the unmodified protein, such as improved foaming properties, inhibition of aggregation or enhanced surface activity. Rationale for chemical modification of proteins is multiple but can be categorized into four main reasons.
i. Waste control: For example the re-use of fish gelatin from waste requires less natural resources for their production. Other examples include the production of a peptide with anti-oxidant activity from algae (Sheih et al., 2009). Protein rich by-products are also recovered upon electrocoagulation of wastewater resulting from egg processing (Xu et al., 2002). Chemical modification can be used to increase the functional properties of these proteins. ii. Health considerations: An example of this is the replacement of meat or soy proteins by (other) vegetable proteins (O'Kane et al., 2004;Pedroche et al., 2004;Vioque et al., 1999;reviewed in Moure et al., 2006). Similar to proteins obtained as by-product from waste material, unmodified plant proteins often have limited functional applicability. Hence, chemical modification of plant proteins can be used to improve molecular functionality. iii. Cost effectivity by extending molecular functionality of a protein. Chemical modification can be employed to enhance the functional properties of a protein, such that less material is required to obtain a product with similar structural characteristics. iv. Structure-function relationships: Chemical modification is often used to investigate the contribution of specific molecular parameters, such as surface hydrophobicity, to functionality of the protein at ingredient level. Ample examples of protein functional extension are discussed in the rest of this paragraph.
The types of modification described in literature are extensive and include phosphorylation (attachment of a phosphate to serine, tyrosine or threonine), methylation (attachment attachment of a methyl group to arginine or the N-terminus of the protein), glycosylation (attachment of carbohydrates to lysine or the N-terminus), acetylation (attachment of acetyl to an amino group such as lysine or the N-terminus), and many more. The various types of modification and their impact on molecular behavior of proteins have been reviewed in a vast number of publications and book chapters (Feeney et al., 1982;Means & Feeney, 1971;Tawfik, 2002). This paragraph aims to shortly discuss the current state-of-the-art of the various reactive groups in proteins which can be targeted by chemical engineering. The chemical basis of these modifications will be discussed and applications from literature will be reviewed merely to illustrate the wide variety of applications of the chemical modification procedure.

Glycosylation and deglycosylation
Food storage and preparation processes such as heating by pasteurization or sterilization often provide for conditions which induce spontaneous and uncontrolled reaction of a reducing carbohydrate present in the food matrix with proteins. This reaction, termed the Maillard reaction, which is actually a complex cascade of reactions, is responsible for the formation of browning products and can have substantial impact on the flavor and color of food products. The Maillard reaction is initiated by a condensation reaction between the -amino group of lysine and the reducing group of a sugar to form Amadori or Heyn's rearrangement products via N-substituted glycosylamine. During the advanced stages of this reaction, the Amadori and Heyn's rearrangement products are degraded via a number of pathways (Mossine et al., 1994;Röper et al., 1983). The last stages of the Maillard reaction involve extensive protein cross-linking reactions and the formation of so-called melanoidins (Pellegrino et al., 1999).  . Also prolonged storage of high-protein nutrition bars showed nonenzymatic Maillard browning as a result of interaction between whey protein isolate and high-fructose corn syrup or sorbitol syrup (McMahon et al., 2009). The baking process of bread was found to affect color formation determined by furosine and hydroxymethylfurfural concentrations (Ramírez-Jiménez et al., 2000). Variation of baking temperature and dough composition determines the extent of furosine formation and loss, acid-released lysine, and carboxymethyllysine formation of cookies (Charissou et al., 2007). Glycosylation via the Maillard reaction has also been brought about intentionally to study the effects of covalent sugar linkage to proteins in terms of e.g. protein stability and aggregation (Feeney et al., 1975). Glycosylation of proteins by means of the Maillard reaction has been observed to both induce and protect against aggregation. Incubation of hazelnut proteins with glucose resulted in the formation of high molecular weight protein aggregates detected by SDS-PAGE (Cucu et al., 2011). At the same time, others have shown that Maillardation with glucose inhibited the aggregation of cod fish parvalbumin (de Jongh et al., 2011). As the Maillard reaction is a reaction involving many steps, one possibility which can be raised to explain the discrepancy between these observations is that the various intermediates may display differences in resistance against aggregation, some of which may be protective, others which may be inducing aggregation. Increased thermal stability upon glycosylation was found for many proteins including the apple allergen Mal d3 upon reaction with glucose (Sancho et al., 2005), and -lactoglobulin reaction with glucose and fructose (Broersen et al., 2005). Alternatively, deglycosylation was shown to induce denaturation and aggregation of ovalbumin (de Groot et al., 2007). Apart from the Maillard reaction, protein glycosylation can be achieved by several other routes. Nglycosylation takes place by modification of the side chains of asparagine or arginine (Kornfeld & Kornfeld, 1985). O-glycosylation is brought about by modification of serine, threonine, or tyrosine (Hart, 1992). Many different glycan structures have been identified to be involved in these two types of modification and they are often necessarily involved in biological function of the protein (Rudd et al., 2001). These last two types of glycosylation take place mainly as a result of post-translational processing of proteins in vivo and are not used as means to induce glycosylation of proteins applied in the food industry. Hence, Nand O-glycosylation will not be discussed in this paragraph.

Lipophilization
Covalent linkage of lipids to proteins results in increased hydrophobic exposure of a protein with interesting applications related to altered surface properties. Increased exposed hydrophobicity of proteins has for example been related to an improved capacity to form and stabilize emulsions and foams which is the result of improved potential to interact with hydrophobic surfaces, both the air-water and oil-water interface, and including (model)membranes (Nakai, 1983;Wierenga et al., 2003;reviewed in Wilde, 2000;Wilde et al., 2004). Various saturated and unsaturated fatty acids have been employed to induce lipophilization of proteins including caproic acid (Liu et al., 2000), capric acid (Aewsiri et al., 2010;Liu et al., 2000), lauric acid (Aewsiri et al., 2010), myristic acid (Aewsiri et al., 2010;Ibrahim et al., 1993;Liu et al., 2000), palmitic acid (Haque et al., 1982;Haque & Kito, 1983a, 1983bIbrahim et al., 1991), stearic acid (Djagny et al., 2001;Ibrahim et al., 1993), and oxidized forms of linoleic acid (Aewsiri et al., 2011a(Aewsiri et al., , 2011b, and the efficiency of the lipophilization reaction was found to be inversely proportional to the length of the lipid chains used (Liu et al., 2000). Reaction of 28% of the available free amino groups of ovalbumin with activated capric acid was shown to result in retained secondary structure while inducing oligomerization and destabilization of the protein structure as a result of lowering the enthalpy for unfolding . The presence of acyl chains was thought to cause significant dehydration of the protein. In another study, hen egg white lysozyme was lipophilized with short and middle chain saturated fatty acids including caproic (C6:0), capric (C10:0), and myristic (C14:0) acid (Liu et al., 2000). Lipophilization of lysozyme was reported to decrease the thermal stability of lysozyme as a result of partial loss of -helical content of the protein, and this molecular destabilization appeared to be proportionally related to the chain length and the number of bound fatty acids. The lysine residues involved in the modification were thought to be located in the helical region and to subsequently induce partial unfolding of the -helical region surrounding these residues (Liu et al., 2000). Lysozyme has also been chemically modified using palmitic acid (Ibrahim et al., 1991(Ibrahim et al., , 1993 with the primary aim to study the effect of lipophilization on the antimicrobial effect of the protein. Even though increasing extents of covalent linkage with palmitoyl residues lead to insoluble protein, as spectrophotometrically determined by solution turbidity at 500 nm, foaming stability and emulsifying activity were progressively improved by linkage of palmitic acid to the protein molecule. More groups showed that the foaming or emulsifying activities of a wide range of proteins could be improved upon lipophilization, including soybean glycinin (Haque et al., 1982),  s1 -casein (Haque & Kito, 1983b), and cuttlefish skin gelatin (Awesiri et al., 2011a(Awesiri et al., , 2011b. A further effect resulting from the incorporation of myristic and stearic acids into lysozyme was related to antimicrobial activity and stearic and palmitic acid conjugation resulted in more effective antimicrobial agents against E. coli, than the attachment of myristic acid or the unmodified protein (Ibrahim, 1993). Myristoylation was found to induce lysozyme aggregation resulting in concurrent loss of antimicrobial function. The effects of palmitoylation on the structural and functional properties of  s1 -casein have also been explored (Haque & Kito, 1983a, 1983b. The conjugation of the -amino groups of  s1 -casein with palmitic acid lead to micelle formation as a result of increased hydrophobicity while negative net charge was increased (Haque & Kito 1983a). Further work by this group showed that palmitic acid linkage to  s1 -casein did not lead to large scale structural rearrangement of the molecule, both at a secondary and a tertiary structure level using circular dichroism. Interestingly, Aewsiri and colleagues have also investigated the antioxidative activity of cuttlefish skin gelatin modified with a combination of oxidized linoleic acid and oxidized tannic acid, a potent antioxidant (Aewsiri et al., 2011a). Oxidation of lipids and proteins in foams primarily takes place at the air-water interface and the addition of a hydrophilic antioxidant alone reduces surface activity (Aewsiri et al., 2011a). Co-conjugation of tannic acid and linoleic acid to gelatin both improves migration of the protein to the air-water or oilwater interface improving foaming and emulsifying activity, respectively, while retaining anti-oxidant activity (Aewsiri et al., 2011a).

Chemical-reactive groups
Sulfhydryl groups play an important role in regulating the self-assembly of proteins as well as their stability driven by disulfide interchange reactions (Sawyer, 1968). Hence, the presence of these groups has substantial impact on the aggregation and gelation behavior of a wide variety of proteins which has been confirmed by many researchers (Arntfield et al., 1991;Broersen et al., 2006;Graña-Montes et al., 2011;Hayakawa & Nakai, 1985;Hoffmann & van Mil, 1997;Margoshes, 1990;Mine, 1992;Sawyer, 1968;Shimada & Cheftel, 1989). A variety of modifications can be performed targeting sulfhydryl groups, which are part of the cysteine residues. Sulfhydryl groups are highly reactive against various reactants and are thus suitable targets for modification. Sulfhydryls can be blocked to prevent cross-linking by S-methyl methanethiosulfonate (MMTS), N-ethylmaleimide (NEM) , or iodoacetamide (Anson, 1940;Huggins & Jensen, 1949;Smythe, 1936), or additional sulfhydryl-groups can be attached to primary amines (SATA) of proteins. Further, Nhydroxysuccinimide esters can react irreversibly with primary amines releasing Nhydroxysuccinimide.
Aggregates and gel networks are often the result of combined action of hydrophobic and electrostatic interactions and covalent interactions, in the form of disulfide bonds, are sometimes present (Kato et al., 1983;Koseki et al., 1989;Sun & Hayakawa, 2002). Thiolation of ovalbumin mediated through the reaction of S-acetylmercaptosuccinic anhydride (S-AMSA) with primary amines results in the formation of acetylthio groups and the acetyl group can be cleaved off to yield reactive sulfhydryl groups by the addition of hydroxylamine (Klotz & Heiney, 1962). A range of modification degrees can be obtained by varying the S-AMSA:lysine ratio (Broersen et al., 2006). Next to the linkage of a sulfhydryl group, additional carboxyl groups are conjugated through this reaction introducing additional charge variation which can lead to an additional parameter which can induce variation in aggregation, gelation or stability of a protein. To circumvent this variation, proteins with activated sulfhydryl groups are best compared with similarly modified proteins with blocked (i.e. not reacted with hydroxylamine) acetylthio groups, rather than directly with the unmodified protein (Broersen et al., 2006). Thiolation of ovalbumin in this way lead to limited changes at a secondary and tertiary structure level at high degrees of modification suggesting that the original molecular fold was largely retained upon modification. High degrees of thiolation resulted in a decrease of thermal stability of ovalbumin while fibril morphology was affected. Interestingly, the rate of aggregate formation was not modified by the presence of additional sulfhydryl groups available for disulfide formation. It was concluded that disulfide formation does not represent the prime driving force for aggregation of ovalbumin which was further illustrated by the finding that at room temperature, where significant sulfhydryl groups are already exposed upon modification, no disulfide bond aggregation was observed. Final aggregate morphology, gel formation and stability are affected as a result of rapid covalent network formation which does not allow rearrangement into more stable networks, as illustrated by lower gel Young's moduli obtained upon thiolation (Broersen et al., 2006).

Charge modification by methylation and succinylation
Net charge and local charge densities of proteins have been implicated in the regulation of protein stability, aggregation, and aggregate morphology affecting the visual appearance of food products. These hypotheses have been substantiated by a range of observations which involved charge introduction, removal or reversal through succinylation and methylation reactions (Broersen et al., 2007a;Weijers et al., 2008). The reactions of succinylation and acetylation both lead to blockage of the reactive amino groups of proteins with an acyl residue and are hence collectively termed acylation reactions. The rate of acylation reaction depends on the rate of nucleophilic attack. Succinylation leads to increased net negative charge by the covalent linkage of succinate anions to the cationic amino groups of a protein thereby converting a cationic group into an anionic residue having implications for the distribution of net charge of a protein. Upon acetylation, ammonium cations are replaced by neutral acetyl groups resulting in electrostatically neutral groups. Large extents of succinylation have been reported to affect the integrity of secondary and tertiary structure of soy protein hydrolysate as shown by intrinsic tryptophan fluorescence and circular dichroism (Achouri & Zhang, 2001). Similar conformational rearrangements have been reported upon succinylation of whey protein isolate (Gruener & Ismond, 1997), bovine serum albumin (Jonas & Weber, 1970), canola protein (Lakkis & Villota, 1992), Faba bean legumin (Schwenke et al., 1998), rapeseed 12S globulin (Gueguen et al., 1990), and winged bean protein (Narayana & Rao, 1991). As a result of co-incubation of soy protein hydrolysate with succinic anhydride, which is a common compound used to succinylate proteins, heterogeneous reaction mixtures were obtained. Next to the aimed amine groups, this method of succinylation also commonly results in O-succinylation, involving threonine or serine hydroxyl groups or tyrosine succinylation (Achouri & Zhang, 2001;Chang & Sun, 1978;Schwenke et al., 1998). This latter reaction was found to be reversible upon treatment with hydroxylamine (Habeeb & Atassi, 1969), but, when present, to induce substantial expansion of Faba bean legumin as observed by viscometric studies (Schwenke et al., 1998). It was postulated that the high accumulation of negative charge upon extensive succinylation leads to dissociation and expansion of the individual subunits legumin is composed of (Schwenke et al., 1998). Other functional properties are equally affected by succinylation. For example, protein solubility has been reported to increase upon succinylation as has been demonstrated for rapeseed preparations (Dua et al., 1996), flax protein isolate (Wanasundara & Shahidi, 1997), oat protein isolate (Mirmoghatadaie et al., 2009), and soy protein isolate (Franzen & Kinsella, 1976a). Improved solubility has been related to the ability of proteins to perform more efficiently as stabilizers in emulsions and foams (Nakai & Li-Chan, 1988;Waniska & Kinsella, 1979), which, in turn, is greatly affected by their ability to absorb at the air-water interface (Wierenga et al., 2005). It has indeed been shown that treatment of a variety of proteins with succinic anhydride leads to increased foam capacity (Dua et al., 1996;Franzen & Kinsella, 1976a;Mirmoghatadaie et al., 2009), although others suggest that succinylation leads to decreased foam expansion capacity (Wanasundara & Shahidi, 1997). These seemingly contradictive findings may be explained by the demonstration by Wierenga and colleagues (2005) that the likelihood of a protein molecule to adsorb at an interface is the result of a balance between hydrophobic and steric effects: highly charged molecules may be adsorbing to the interface as a result of hydrophobic interaction, but the density at which protein molecules continue adsorbing to the interface is mainly determined by the repulsive nature of the charged proteins. Other effects observed upon charge modification of proteins are related to emulsification properties: methylation increased while succinylation decreased the emulsifying activity of rapeseed preparations (Dua et al., 1996). Other studies show an increased emulsifying activity and stability upon succinylation of soy protein (Franzen & Kinsella, 1976a), and oat protein isolate (Mirmoghatadaie et al., 2009). In terms of gelation, an increase in net charge lead to more transparent gels upon gelation of ovalbumin which is related to the morphology of the aggregated network making up the gel structure (Weijers et al., 2008). Overall, many and detailed efforts have been made employing net charge modification of proteins in the field of food science. These studies have lead to in-depth knowledge of the role of electrostatics to common protein functionalities such as emulsification, foaming, and aggregation propensity.

PEGylation
The covalent attachment of a polyethylene glycol (PEG) polymer chain to a protein, also termed 'PEGylation' is mostly applied in the field of pharmaceutics as the conjugation of non-toxic PEG imparts substantial advantages to support drug delivery (reviewed in Damodaran & Fee, 2010;DeSantis & Jones, 1999;Francis et al., 1998). The protein targets for PEG modification are regarded as non-specific and include the -amino groups of lysine and other nucleophilic groups such as glutamic acid, aspartic acid, threonine, serine or tyrosine on the surface of the protein resulting in highly heterogeneous protein-PEG conjugates upon modification (Losso & Nakai, 2002). Commercially available PEG is available as mixtures of different oligomer sizes in various molecular weight ranges enabling the variation of exposed hydrophobicity of proteins. Conjugation of hydrophilic PEG to a hydrophobic protein generally results in an increase in hydrodynamic size and water solubility (Damodaran & Fee, 2010). From a pharmaceutical viewpoint, PEGylation has been reported to enhance circulation life of bovine liver catalase in the blood of mice while the presence of PEG does not induce an immune response upon injection (Abuchowski et al., 1977). Modification of peroxidase from turnip was shown to enhance catalytic activity of the enzyme with increased stability in organic solvents as well as increased temperature resistance (Quintanilla-Guerrero et al., 2008). Similar results were found upon PEGylation of trichosanthin, which showed prolonged plasma half-life and reduced immunogenicity (He et al., 1999). PEGylation of lysozyme similarly lead to stabilization of the protein against pH and temperature variation as well as resistance against proteolysis (Silva Freitas & Abrahão-Neto, 2010). Most of these studies have been carried out in the context of pharmaceutical application. The study of PEGylation as a potential route to bring about modification of physicochemical parameters of proteins applied in food products has been less well explored. The only known effort in the field of food science combined oxidative sulfitolysis with conjugation of 5000 dalton activated PEG to investigate the impact on the emulsifying properties of -lactoglobulin A (Losso & Nakai, 2002). PEG molecules were found to cover the entire surface of unmodified and sulfitolyzed -lactoglobulin and PEGylation alone did not improve emulsifying activity or emulsion stability. The combination of sulfitolysis and PEGylation however lead to more stable emulsions and improved emulsion activity index as a result of better stabilization of individual droplets against coalescence by the absorption of PEG onto the surface of the sulfitolyzed -lactoglobulin (Losso & Nakai, 2002). From the limited number of studies available it is not possible to derive direct conclusions on the applicability of PEGylation on the advancement of functionality of proteins in food products.

Deamidation
Deamidation involves the hydrolysis of the amino acids glutamine and asparagine into glutamic and aspartic acid and is achieved by acid, alkaline, or enzymatic treatment (Liao et al., 2009;Shih, 1990;reviewed in Wright & Urry, 1991). Consequently, deamidated protein is often obtained as a by-product of food processing. For example, the extrusion of wheat flour induces deamidation of wheat proteins (Izzo et al., 1993). Deamidation has been shown to affect protein functionality. Functional properties, such as solubility, emulsifying and foaming properties, of gluten have been reported to improve upon low levels of deamidation brought about by mild acid hydrolysis (Hamada & Marshall, 1989;Matsudomi et al., 1982Matsudomi et al., , 1985. Deamidation also was reported to increase exposed hydrophobicity of gluten induced by a conformational change and, subsequently, to increase surface activity (Matsudomi et al., 1982). A number of studies investigated the molecular mechanism for protein structural destabilization upon deamidation. For example, the deamidation treatment has direct implications for charge density and, in turn, affects electrostatic interactions the protein may undergo by interacting with water or upon self-assembly (Finley, 1975;reviewed in Riha et al., 1996). This role of electrostatics to deamidationinduced disruption of protein structure was supported by further observations on wheat gluten: both acetic acid and HCl induced deamidation had substantial consequences for the secondary structure of wheat gluten. It was thus postulated that strong deamidation induced protein unfolding as a result of electrostatic repulsion (Liao et al., 2010). Acetic acid induced deamidation of wheat gluten was further found to inhibit SDS-stable aggregate formation whilst largely retaining its ability to form disulfide bonds (Liao et al., 2010). The rate of the deamidation reaction has been found to depend on primary sequence and pH under which the reaction takes place, but was independent of ionic strength for model peptides (Patel & Borchardt, 1990a, 1990bRobinson & Rudd, 1974;Tyler-Cross & Schirch, 1991), soy protein and egg white lysozyme (Zhang et al., 1993).

Cross-linking
A variety of aldehydes, including gluteraldehyde, formaldehyde, and hydroxyadipaldehyde, have been used to induce chemical cross-linking of proteins. Between these reagents, gluteraldehyde was found to cross-link bovine serum albumin most efficiently forming large insoluble networks (Hopwood, 1969). Also for other proteins gluteraldehyde has been reported as the most efficient cross-linking agent yielding thermally and chemically stable cross-links (Bowes & Cater, 1968;Nimni et al., 1987). Primary amino groups have been reported to act as prime target to initiate the aldehydeinduced cross-linking reaction (Quiocho & Richards, 1966), next to aromatic amino acids (Hopwood et al., 1970). Depending on environmental conditions, gluteraldehyde can bring about cross-linking through a wide variety of reaction mechanisms (reviewed in Migneault et al., 2004). This is caused by the large number of different molecular structures gluteraldehyde can assume in solution (Hardy et al., 1969;Korn et al., 1972;Richard & Knowles, 1968;reviewed in Migneault et al., 2004) although the mechanistic details for this is unknown. As a result of its reported toxicity, gluteraldehyde-induced cross-linking has not been employed in the field of food technology, other than as a tool to enable the investigation of intermediates in the aggregation pathway or to immobilize proteins onto a surface to allow further investigation.

Measuring the degree of modification
To evaluate the effect of a specific type of modification on the functional aspects under study, the success of the chemical engineering process on proteins is evaluated by most researchers. To this end, targeted chemical and biophysical assays have been developed which are now widely used. These quantitative assays are mostly based on the formation of a chromogenic or fluorogenic product upon specific interaction with reactive groups of a protein. Some assays will provide information on the average degree of modification in the entire ensemble of protein molecules in a solution. Examples of such assays are Ellman's reagent (Ellman, 1959) or the sulfhydryl-disulfide exchange (SEI) index (Owusu Apenten et al., 2003). Others are also useful to obtain information on the distribution of the degree of modification obtained, such as mass spectrometry. The type of assays developed can be categorized by the type of aimed conjugated chemical group of the protein they probe, such as amine groups, thiol groups or carboxyl groups. Some other researchers use methods which rather probe for the attached molecule, such as the use of gas liquid chromatography (GLC) to determine the degree of lipid incorporation (Haque et al., 1982).

Amine groups
Amine groups in proteins originate either from free amino groups of proteins or from Nterminal residues of proteins (Skraup & Kaas, 1906;Chibnall, 1942). From 1906 it was recognized that lysines in proteins were largely responsible for the free amino groups present in proteins (Skraup & Kaas, 1906). This finding triggered the development of various assays to study accessible -amino groups of lysines in proteins (Gurin & Clarke, 1934;Sanger, 1945). Three of the most commonly used assays in the field of protein engineering to determine the number of available amino groups of a chemically modified protein involve the use of chemicals 2,4,6-trinitrobenzenesulphonic acid (TNBS) (Fields, 1971), ninhydrin (Yemm & Cocking, 1955;Schilling et al.;1963;Samejima et al., 1971), and ortho-phtaldialdehyde (OPA) (Roth, 1971). The latter assay is based on the reaction of the OPA compound with free amino groups in proteins in the presence of -mercaptoethanol under alkaline conditions. This reaction results in the formation of highly fluorescent alkyliso-indole derivatives which emit at a wavelength of 455 nm upon excitation at 340 nm. High concentrations (i.e. 10%) of sodium dodecyl sulfate (SDS) are often added to the protein solution to aid the exposure of all amino groups which are sometimes buried in the folded protein. The extinction coefficient for the formed adducts of both and -amino groups are similar with an absorptivity of 6000 M -1 cm -1 . The OPA assay was further developed to evaluate the degree of proteolysis of dairy proteins by determining the number of -amino groups released upon hydrolysis (Church et al., 1983). The assay can also be used to quantify all types of reactions involving the modification of lysine and, hence, includes lipidation of ovalbumin Wierenga et al., 2003), lysozyme (Liu et al., 2000), and gelatin (Aewsiri et al., 2010(Aewsiri et al., , 2011a(Aewsiri et al., , 2011b. Other modifications which can be quantified using the OPA assay are succinylation Wierenga et al., 2005), methylation , glycosylation (Broersen et al., 2004;, and thiolation of proteins (Broersen et al., 2006).
The second frequently used method to quantify available amino groups upon modification involving the chemical TNBS was developed by Okuyama and Satake (1960) and Satake and colleagues (1960). This method was first employed to study free amino acid groups in trypsin and chymotrypsin inhibitors (Haynes et al., 1967) and for routine screening of protein concentrates for animal feeds (Hall et al., 1973). The reaction product can be quantified spectrophotometrically at 335 nm. A disadvantage of this method is that TNBS also reacts with free sulfhydryl groups, albeit at a slower rate than with amino groups and to form a labile product (Kotaki et al., 1964). The TNBS assay has been used for various types of protein modification including fatty acid incorporation (Andersson et al., 1971;Ibrahim et al., 1991Ibrahim et al., , 1993, glycosylation of proteins as a result of the Maillard reaction (Sun et al., 2004), succinylation (El-Adawy, 2000;Schwenke et al., 1998;Zhao et al., 2004aZhao et al., , 2004b, and acetylation of faba bean legumin (Krause et al., 1996).
Comparison of TNBS, OPA or ninhydrin to determine -amino groups in pea protein isolates and hydrolysates thereof lead to the conclusion that TNBS and OPA produced comparable results while ninhydrin detected only half of the amino groups that were detected by TNBS and OPA (Panasiuk et al., 1998).

Sulfhydryl groups
Two commonly used assays are available to evaluate the successful conjugation or blockage of sulfhydryl groups in proteins. The assay which was developed first by Ellman (1959) involves the reaction of 5,5'-dithiobis(2-nitrobenzoic acid) (DTNB) or Ellman's reagent with free sulfhydryl groups yielding colored 3-carboxylato-4-nitrothiophenolate (CNT). Spectrophotometric absorbance intensity at 412 nm provides for a direct measure of the concentration of CNT in solution and cysteine is commonly used as calibration standard. Ellman's assay has been used before to determine the extent of N-ethylmaleimide modification of -lactoglobulin A , thiolation of ovalbumin (Broersen et al., 2006), or acylation of soy protein sulfhydryl groups (Franzen & Kinsella, 1976a). However, the sizeable DTNB at 400 dalton may not be able to detect sulfhydryl groups which are buried inside the folded structure of intact proteins or aggregated proteins (reviewed by Visschers & de Jongh, 2005). A useful alternative for the Ellman's assay has been developed by Owusu Apenten and colleagues (2003) for -lactoglobulin and bovine serum albumin. Chemical reactivity of thiol groups is an absolute requirement to enable covalent cross-linking through disulfide bond formation (Hillier et al., 1980). This assay, called the sulfhydryl-disulfide exchange (SEI) index, therefore provides a direct measure of the chemical reactivity of thiol groups as it determines the conversion of substrate in time and kinetically relates the conversion to that of fully exposed thiol groups (Owusu Apenten et al., 2003). Chemical activity of introduced sulfhydryl groups by modification of ovalbumin using the SEI index has been verified for example by Wierenga and colleagues (2006).

Carboxylic acid groups
The carboxylic acid content of proteins is primarily investigated using the compound 2ethyl-5-phenylisoxazolium-3'-sulfonate or Woodward's reagent K Sinha & Brewer, 1985). The activity of Woodward's reagent K is the result of a multistep process. First, Woodward's reagent K is converted into ketoketenimine at neutral pH. The intermediate compound is then either further disintegrated to form ketoamide or interacts with carboxylic acid groups of a protein. The latter interaction results in the formation of an enol ester (Pétra, 1971) which absorbs at 340 nm with a molar extinction coefficient of 7000 M -1 cm -1 (Sinha & Brewer, 1985). At a later stage,  revised the extinction coefficient of the product to 3150 M -1 cm -1 at 269 nm to improve specificity of the reaction and eliminate the substantial contribution of side reactions with other nucleophiles in proteins (Llamas et al., 1986), and histidine and cysteine (Bustos et al., 1996;Johnson & Dekker, 1996) to the absorbance at 269 nm. This revised version of the assay employing Woodward's reagent K to estimate the number of carboxylic acid groups of chemically modified proteins has been used by a number of researchers. Wierenga and colleagues (2005) used Woodward's reagent K to estimate the degree of modification of succinylated ovalbumin to study the relation between protein net charge and adsorption to air-water interfaces. Similar net charge variation induced by succinylation was used to investigate colloidal versus conformational stability of ovalbumin to aggregation (Broersen et al., 2007a). To investigate the stability of ovalbumin, the protein was modified by succinylation, methylation, glycosylation, and lipophilization and the degrees of modification were validated using Woodward's reagent K and the OPA assay (paragraph 3.8.1).

Conjugated groups
An alternative route to obtain information on the degree of protein modification is to selectively probe the conjugated group. This can be achieved for example by incorporating an isotopically labeled reagent or inclusion of a chromophore or fluorophore which can then be quantified by read-out of fluorescence intensity using a fluorimeter or simple absorbance measurements using a standard spectrophotometer. Raman spectroscopy was shown to provide direct insight into degrees of succinylation and acetylation of a range of proteins originating from soy, egg white or whey by distinct contributions of the conjugated groups at 1737 cm -1 and 1420 cm -1 (Zhao et al., 2004a(Zhao et al., , 2004b. The peak intensities at these wavelengths could be directly converted to obtain information on the degree of modification. Degrees of palmitoylation of soybean glycinin (Haque et al., 1982) and  s1casein (Haque & Kito, 1983a) have been determined using gas liquid chromatography.
Chromatography has been further explored in the shape of cation exchange chromatography to validate the degree of methylation of ovalbumin which was found to provide comparable read-outs as the revised version of Woodward's reagent K method to determine carboxylic acid groups . All these techniques provide insight in the ensemble average degree of modification. Chemical engineering inherently implies the rise of heterogeneous species of proteins. Mass spectrometry, often employed as Matrix-assisted laser desorption/ionization-time of flight (MALDI-tof) mass spectrometry, has proven a powerful method to specifically obtain insight into the distribution of the modification reactions. This method has been employed to derive information on modification distributions of N-ethylmaleimide modified -lactoglobulin A , and glycosylation of -lactoglobulin (Broersen et al., 2004;van Teeffelen et al., 2005). These studies demonstrated that degrees of modification obtained upon chemical engineering of proteins are rather broad and show a Gauss distribution profile rather than a single well-defined modification degree (Broersen et al., 2004;van Teeffelen et al., 2005). Some attempts have been made to isolate modified protein fractions with more defined degrees of modification, for example by using ion exchange chromatography of succinylated ovalbumin (Wierenga et al., 2005).

Application potential of food protein modification
From the above it is clear that in the past decades a lot of effort has been spend on better understanding and controlling protein behavior and protein-based microstructure formation by making use of chemical engineering approaches. But how much impact have these insights had on the development of new food applications? There are a number of well-known product categories where engineered protein functionality has led to improved product properties. In the early seventies Unilever produced new lines of margarines that showed better performance in aspects like spreadability, prolonged storage stability and during baking, caused by acetylation of milk proteins leading to better fat emulsification (Evans & Irons, 1970). Also for mayonnaises and salad dressings modification of egg yolk proteins (via N-succinylation) provided improved product quality (Evans & Irons 1970). Another example of an application of improved protein functionality is that of the use of succinylation to improve the solubility/dispersability properties of soy proteins in the extraction and refinery process (Melnychyn & Stapley, 1973). It is interesting to evaluate via a patent-literature screening how frequent the wide variety of technological possibilities as described in section 3 to better control protein behavior have led to unique marketpropositions. Figure 1 illustrates a landscape representation of patents (worldwide), filed in the last decade, in the area of food and feed where protein modification has played a crucial role in deriving a new type of product functionality. The height of the contour indicates the activity in that particular area. The distance between patents reflects their commonality. In total only 445 relevant patents (grouped in 157 families) can be found. In comparison, a search on any biobased-product (including products with protein-based technical polymers as in coatings, paints, paper, etc.) revealed more than 8600 patents. Clearly the role of protein engineering to derive new food product specifications are very limited, especially in view of the potential commercial impact. The observation that the patent-families found are spread rather constant over the plot illustrates that these patents are not directly linked to each other in terms of engineering approach or application area. When evaluating the patent filings of the last ten years five areas can be distinguished with a relatively high patenting activity and these will be discussed below in more detail. These are numbered 1 to 5 in the figure.

Current application of protein modifications in food/feed-related products
1. Refinery of seed storage proteins (12 patents). There is an increasing interest to use readily abundant and relatively cheap seed storage proteins as nutritional component in food and feed. The major difficulty of this protein-source is to obtain functional proteins after the refinery steps. Loss of functionality occurs especially when the protein is used as powdered ingredient, because of difficulties in resolubilization and unpredictable caking of the powder during storage/transport. A number of patents have been filed that use mild engineering tools, like Maillardation, or enzymetreatments to preserve functional proteins during refinery steps (e.g. patent EP1370157B1: "Highly soluble, high molecular weight soy protein"). Interestingly, all patents filed in this area pay attention to the in principle reversibility of the modification applied. 2. Nano-particles (23 patents). To better direct properties of protein-based nano-particles enzymatic introduction of lipidic groups like small fatty acids or PEGylation has been employed to encapsulate bioactives, typically as microemulsions (e.g. patent US20070154907A1: "Microemulsions as precursors to solid nanoparticles"). These applications are considered food-grade or have passed medical-ethical approval in their testing/and or usage. There are no indications that these application have found their way into the food/feed product market. 3. Nutritional availability in feeds (13 patents). A considerable number of patents can be found in the application area to increase the nutritional availability of amino acids in feed (14 patents). Most patent-positions are dealing with destabilizing proteinstructures to promote their digestibility and their (proposed) nutritional uptake. Typical patented approaches are de-amidation and Maillardation (e.g. patent WO2004020977A3: "De-amidation to promote nutritional value of rumen in feed"). No patents can be found that link nutritional value and protein modification in foods. 4. Protein-based emulsion-stabilizers (two times 7 patents). A few patents are found where spe c if i c prot e i n modi fi c at io n is us ed t o impro ve emu l si f ic at ion o f ( oi l in wa ter ) products. Especially enzymatic glycosylation and lipidation approaches have been used (e.g. patent US7126042B1: "Recombinant oleosins from cacao and their use as flavoring or emulsifying agents"). In view of the small scale examples provided in these patents, it is not likely that these inventions have been implemented in a commercial product. 5. Edible coatings (12 patents). A whole family of patents is present on enzymatic protein cross-linking (mainly by transglutaminase) in relation to the production of edible coatings. Modifications typically act on a microstructural level, and not so much on the protein molecular level, to strengthen the spatial network formed (e.g. patent EP963704B1: "Food containing proteinaceous material treated with a transglutaminase and an oxidoreductase"). As these interventions in a product do not necessarily need to be labeled on the product, it is difficult to evaluate whether they have resulted in product development.

Food legislation
Whereas in paragraph 3 it was demonstrated how active and progressing the understanding of protein functionality in complex systems has been in the past decades, it is striking to see how minor the contributions to new applications these insights apparently have been. To understand this better one needs to take into account that the technological developments coincided with an increasing level of legislation. In the United States this is embodied in the Food and Drug Administration (FDA). The European Food Safety Authority (EFSA) is its European counterpart. The FDA is responsible for protecting and promoting public health through the regulation and supervision of food safety. It does so by formulating acts that set the boundaries for implementation of new ingredients, processes or compositions related to foods. There are a few acts that have had a strong impact on the food sector. In 1990 the Nutrition Labeling and Education Act was launched. This required food products to be labeled in terms of composition, allowing traceability of its ingredients to their source. This act also amended that all nutrient content claims (e.g. 'high fiber', 'low fat', etc.) would meet the standards set by the FDA. In the end it meant that every engineered protein would require a new label and would need to be recognized and approved first by the FDA.

2 3 4
Exceptions were those modified ingredients that could be considered as 'occurring from a natural process' or that were 'reversibly modified' (so temporary). The FDA Modernization Act of 1997 was designed to reduce the time for the approval of new pharmaceutical drugs, but also had an impact on food technology by the acknowledgment of the advancement of technological, trade, and public health complexities. Basically, a 'new' food ingredient needed to be seen and evaluated in the complex role it had considering its production up to its digestion in the food. This act was further refined by the Amendments Act of 2007, leading to much sharper defined criteria in what was considered as safe in food products. Recently, in 2010, The Food Safety Modernization Act was signed. Sections of this act require food producers to enable tracking and tracing of all ingredients used. The use of engineered proteins (either genetically or chemically) requires a separate approval for market-clearance. Summarizing, the FDA (and EFSA) have acknowledged that (future) food production requires innovations at the ingredient level, but also via processing routes, and they are in principle open to protein engineering routes. At the same time it enforces that functionally improved ingredients are checked along the full chain from refinery to nutritional value and human health within the complexity of the product.

Potentials in the area of food product sustainability
With increasing world-population and welfare the demand for protein as food-nutritional component is rising sharply. Also the identification of proteins as building blocks in nonfood applications in view of a more sustainable economy, has led to increasing pressure on innovations in production, refinery, and application of proteins from wider sources than in the current economy is provided. For the development of a vision on food quality and especially the role of nutritional impact needs to be seen in the context of the basic requirements set not only by consumer demands, but also by participating industries. This can be presented by a so-called Pyramid of Food Innovation, as shown in figure 2. On top one finds the foods that need to be developed in a most sustainable way; to achieve this prospect one needs to comply with lower levels of restrictions, limitations and concomitant scientific challenges.
Food safety forms the most fundamental aspect here. Terminology like for example 'nanotechnology' cannot count on consumer acceptance and also the inclusion of genetically modified ingredients gives rise within Europe to hesitance in applicability by foodproducing companies. Especially for traceability and chemical characterization the demands become exceedingly higher. This is the level though where new protein engineering routes could be contributing most. A major bottleneck today in reformulation-strategies is the occurrence of (sensory) differentiations relative to the original product. Exceptions are reduced sugar or salt products, but from a marketing-technical perspective moderations of structuring components are preferably performed within the frame as novel food. A continuously on-going drive to elucidate structure-texture-product acceptance relations is essential, just like innovations in the area of optimized processing tools to deliver products that are acceptable with retained food safety. Typically the efforts on applying protein modifications are focusing in deriving food structures more efficiently or to provide new functional building blocks to create new food structures. In an economically global society there is a strong pressure on both ingredient-prices and commercial acceptable processing. Industrial entrepreneurship is essential and the availability of second-line ingredients for products hampers the implementation of new food production strategies. Especially within Consumers are susceptible to additional health-aspects when safe and acceptable good quality products are available for reasonable prices. In this information technology era consumers are capable to evaluate the added value of products far better, setting higher demands for health marks of food products. The use of added value to products, like bioactives, has not led to major winners in the food sector yet. The desire (and need) to innovate in food product developments could be fed by the wealth of information on how product functionality relates to microstructural morphologies and how these in turn are dictated by molecular properties of specific proteins. On the other hand there are robust approaches in the field of protein engineering that allows us to direct protein behaviour and their propensities to (self)assemble into spatial networks. Still, one has to recognize that these two aspects have not come together yet to contribute to a more sustainable production and use of proteins in foods. This will be further discussed in section 5.

Considerations of chemical modification of proteins
While genetic modification of proteins generally results in a homogeneous product, chemical engineering approaches to modify proteins are well-known to result in heterogeneity. A good example is the application of the Maillard reaction to -lactoglobulin (Broersen et al., 2004) or fish parvalbumin (de Jongh et al., 2011). On average one may find that for example 6 of the 12 available lysine residues has become glycosylated, where mass analysis will demonstrate that a significant population of protein molecules is present with 5 or 7 sugar groups, some molecules may contain 4 or 8 groups and even traces of molecules with 3 and 9 groups are

Nutritional/ Health
present. This is commonly the result of variation in reactivity between the different amino groups present in a specific protein structure: i.e. some amino acids are more exposed and hence are more prone to rapid modification. Buried amino acids do not present a straight forward target for modification as most chemicals are unable to access the folded or aggregated protein structure. Another effect which likely contributes to the extent of heterogeneity is the location of target amino acids in the primary sequence. For example, modification of an amino acid located next to a negatively charged amino acid by succinic anhydride will be hampered as a result of steric hindrance. Some attempts have been made to fractionate protein molecules with different degrees of modification. Five ovalbumin preparations with different degrees of modification have been purified using ion exchange chromatography (Wierenga et al., 2005). The heterogeneous nature of chemically modified proteins has marked implications on the acceptance of these novel ingredients under current food law. Chemical heterogeneity would imply that each obtained modified species would have to be tested separately on toxicity in order to become regarded as safe to use as novel food ingredient.

Health risks
In this paragraph we will discuss potential health implications induced by chemically modified proteins used in food products including allergenic response, the presence of antinutritional factors linked to proteins and the development of potential toxic compounds as a side-effect of the modification reaction. Concerns can be raised regarding the impact of protein modification on the potential health risks from products containing modified protein ingredients. Bernstein and colleagues (2003) identified three possible modes for novel food ingredients to result in adverse health effects. These include toxicity, impaired nutrition and food allergy.

Toxicity
One of the known toxic compounds to be formed upon chemical engineering of proteins is acrylamide, which results from the Maillard reaction between reducing sugars and asparagine or methionine (Mottram et al., 2002;Stadler et al., 2002). The high solubility of this compound induces rapid absorption and metabolism in the body (reviewed in Dearfield et al., 1988). Following absorption, acrylamide can bind to DNA aiding a genotoxic and carcinogenic response which has been demonstrated in animals (Rudén, 2004). The occurrence of carcinogenic acrylamide in foods has been related to frying and cooking of food products (Rosen & Hellenäs, 2002;Tareke et al., 2002). Two communications published in Nature in 2002 demonstrated the requirement of the reaction between asparagine or methionine and a reducing sugar to intermediate formation of the dicarbonyl reactant followed by Strecker degradation (Mottram et al., 2002;Stadler et al., 2002). Particularly plant proteins are rich in asparagine suggesting that glycosylation of proteins from plant-origin using the Maillard reaction should be carefully considered in terms of the known toxic effects of acrylamide.
Other toxic side effects of chemical modification of proteins have not been reported.

Impaired nutrition
Chemical engineering can theoretically have far-reaching consequences for the nutritional value of proteins. In a limited number of cases protein engineering has been intentionally employed to modify the nutritional quality, for example to increase the nutritional value of plant proteins (Liao et al., 2010). However, the main objective of most studies employing protein engineering is to investigate the consequences of the modification procedure for a range of functional properties. Only a small number of studies investigates nutritional aspects of chemical protein engineering which is mostly regarded as a convenient or inconvenient side effect of the modification procedure. Nutritional aspects covered in literature are exclusively based on in vitro studies and cover protein digestibility, availability of essential amino acids or the presence of anti-nutritional factors.

Anti-nutritional factors
Anti-nutritional factors are related to reduced protein digestibility and amino acid availability (reviewed by Gilani et al., 2005;Salunkhe et al., 1982) and are commonly present in large concentrations in plant products (Kay, 1979;Liener, 1980). One of the studies in this field reports on the nutritional quality of mung bean isolate following the exposure to varying concentrations of acetic or succinic anhydride to induce acylation (El-Adawy, 2000). The concentrations of anti-nutritional factors tannin, phytic acid, and trypsin inhibitor showed a significant loss with increasing degrees of modification suggesting that this type of modification can positively impact the effect of anti-nutritional factors. The concentration of trypsin inhibitor even decreased with 70% of the original level of trypsin inhibitor in unmodified protein. The introduction of bulky and/or negatively charged side groups was postulated to affect the extent of protein-tannin (El-Adawy, 2000), protein-mineral-phytic acid (Dua et al., 1996;El-Adawy, 2000), or protein-phenol (Loomis (1974) interactions. Loomis (1974) further showed that the flour and protein production processes provide for optimal conditions for the conversion of polyphenols into quinone oxidation products which, in turn, may bind covalently with sulfhydryl groups of cysteine and -amino groups of lysine and N-terminal amino groups. Further support was provided by Dua and colleagues (1996) who showed that acylation and methylation of rapeseed meal and its water-soluble fraction resulted in loss of anti-nutritional factors polyphenol, glucosinolates and phytic acid. However, the methylation procedure employed by Dua and colleagues (1996) resulted in very limited degrees of modification compared to the succinylation and acetylation process suggesting that other factors than the chemical conjugation itself may play a role in the loss of anti-nutritional factors upon chemical modification. As El-Adawy (2000) comments, the extensive dialysis of the protein following the acylation procedure may well be primarily responsible for the loss of water-soluble anti-nutritional factors upon modification. No studies are known to date that report on the loss of anti-nutritional factors upon dialysis.

In vitro digestibility
In vitro digestibility of modified proteins is often assayed through exposure of the proteins to a single or a mixture of enzymes including trypsin and pancreatin (Salgó et al., 1984), a combination of trypsin, chymotrypsin and peptidase (Hsu et al., 1977), or pepsin-pancreatin mixtures (Haque et al., 1982) to simulate (post)gastrointestinal digestion of food proteins. The small increase in digestibility of acylated mung bean protein isolate reported by El-Adawy (2000) was primarily correlated to the concurrent loss of tannin; tannins have been shown to play an important role in the reduction of protein digestibility (Barroga et al., 1985). Alternative factors proposed to induce increased digestibility of modified proteins include the improved access of sites susceptible to enzymatic cleavage as a result of the dissociation of quaternary complexes of proteins or partial unfolding induced by the modification procedure (Achouri & Zhang, 2001). Protein unfolding as a result of modification has been shown for soy protein hydrolysate using techniques to study the secondary and tertiary structure content of the protein upon succinylation (Achouri & Zhang, 2001). However, the acylation of cotton seed flour did not improve in vitro protein digestibility (Rahma & Narasinga Rao, 1983) suggesting that results in this area are somewhat controversial and perhaps other factors play a role (see paragraph 4.5.2.1). In vitro protein digestibility using a multienzyme system containing trypsin, chymotrypsin and peptidase was not impaired for pea proteins upon acetylation (Johnson & Brekke, 1983) while Ma (1984) reported increased digestibility for acylated pea protein, similar to low degrees of succinylation of soy protein hydrolysate (Achouri & Zhang, 2001). Loss of in vitro digestibility has also been reported upon succinylation of a variety of proteins, particularly affecting the release of lysine (Matoba & Doi, 1979;Siu & Thompson, 1982;Wanasundara & Shahidi, 1997) or lipophilization of soy bean glycinin with palmitic acid (Haque et al., 1982). Data on in vivo digestibility of ingested proteins upon modification have not been reported.

Availability of essential amino acids
A commonly used target for protein modification are the lysine or cysteine residues, lysine being classified as an essential amino acid, i.e. this amino acid cannot be synthesized de novo by humans and should therefore be ingested. Extensive modification of these amino acids can therefore result in a lower availability. Few studies report on the impact of lysine or cysteine modification on the availability of these amino acids, usually assayed through amino acid analysis. Overall only small decreases in lysine were reported upon succinylation of soy protein hydrolysate (Achouri & Zhang, 2001). Similar findings were reported for acylated soy proteins (Franzen & Kinsella, 1976a), acylated sunflower proteins (Kabirullah & Wills, 1982), and succinylation of oat proteins (Ma & Wood, 1987). It is not clear whether these reported effects of chemical engineering inducing the loss of lysine availability ultimately result in a noticeable and substantial loss of nutritional quality for the human population.

Food allergy and intolerance
An estimated 3-4% of the children and 1-2% of adults in the industrialized world exert an allergenic response to one or more ingested food proteins (Baral & Hourihane, 2005;Jansen et al., 1994). Not all ingested proteins behave as allergens but proteins implicated in allergenic response often share features such as unusual resilience against heat, acid or protease digestion, propensity to bind to lipids and are glycosylated to some degree (Lehrer et al., 2002;Metcalfe et al., 1996). Also posttranslational modifications including Nglycosylation, and hydroxylation of proline residues have been postulated to affect IgE reactivity to Phl p 1 present in timothy grass pollen (Petersen et al., 1998). The precise mechanism of the effects of glycosylation on the allergenic response are not clear. Attempts have been made to develop and evaluate algorithms which predict protein allergenic response based on sequence homology (Aalberse & Stapel, 2001;Jameson & Wolf, 1988), structural identity, and evolutionary relationship (Jenkins et al., 2007), albeit with limited predicting power. The Food and Agriculture Organization (FAO) of the United Nations together with the World Health Organization (WHO) developed a decision tree to assess the allergenic potential of genetically engineered food ingredients based on sequence homology followed by serum screens (FAO/WHO 2001, reviewed by Bernstein et al., 2003). Based on these findings and the fact that to date there is no indication for a common sequence motif of linear IgE epitopes, it can be concluded that the underlying mechanisms of immunological sensitization to food proteins remain elusive. To be able to assess the potential of chemically modified food proteins to induce allergenic reactions implies extensive knowledge of the underlying mechanical aspects of the allergic response. An early paper in the late 1970s reported that, upon covalent attachment of methoxypolyethylene glycols of 1900 daltons or 5000 daltons to bovine liver catalase, no evidence was observed of a modulated immune response following repeated injection of the modified enzymes into mice (Abuchowski et al., 1977).

Future prospects
There is a clear ambition to implement fundamental insights obtained on protein behavior in complex systems more effectively in sustainable food production in the future. There is also the need, considering that one should become more flexible in protein sources and more effective in utilizing the proteins structuring and nutritional potential. The technology to enable this is largely present, as demonstrated by the diversity of literature presented in paragraph 3. The mechanistic insight that has been derived in underlying principles on the relation between (engineered) protein functionality is impressive. And still, the development of new applications using chemical protein engineering to make a better usage of protein functionality or to promote protein source exchangeability is poor. There are a number of reasons to identify. The use of food protein modification has been limited to the domain of food chemistry or food physics. There is no significant literature available where the fine-tuning of microstructure formation as part of product development has been studied at a sensory or (human) physiological level. Compared to genetic modification, chemically engineered proteins pose an inherent heterogeneity. Instead of a single ingredient with an altered functionality, the FDA needs to consider all levels of heterogeneity in their approvals. This is an enormous laborious task as it requires fractionation of derived materials and sufficient stability of the formed products. Moreover, many attempts to reduce heterogeneity in the protein material will lead to non-food grade materials that cannot be studied in relation to e.g. sensory or human physiological aspects. The food safety authorities have amended (see paragraph 4.2) that the protein needs to be seen in the complex context of the food product. This is not just chemical analysis, but also includes product stability, consumer acceptance (sensory) and human physiological data. Especially in these latter two domains there is no public literature readily available to help these authorities to establish objective views on tolerance. This is where the scientific community has a highly needed role in bringing different disciplines together to produce literature relevant for authorities to base their legislations on.