DESIGNING THE FUNCTIONALITY OF MILK PROTEIN BY ENZYMATIC MODIFICATION

Harold E. Swaisgood, George L. Catignani, Xiaolin L. Huang, Val W. Valentine and Debra A. Clare

Department of Food Science, Southeast Dairy Foods Research Center , North Carolina State University, Raleigh NC 27695-7624

HYPERLINKED LIST OF CONTENTS

INTRODUCTION LIMITED PROTEOLYSIS OF BETA-LACTOGLOBULIN LIMITED PROTEOLYSIS OF WHEY PROTEIN DESIGN OF RECOMBINANT ENZYMES CONCLUSIONS REFERENCES

INTRODUCTION

Increasingly, proteins are used as ingredients in foods to provide specific functional characteristics such as emulsification, foaming, gelation, viscosity and water-holding capability (Phillips et al., 1994; Kinsella et al ., 1994). With the more recent interest in non-fat foods, proteins have also been used as a fat mimetic and b-lactoglobulin ( b- Lg) has been proposed as a carrier for lipophilic nutrients (Wang et al., 1997a, 1997b). The functionalities of proteins in foods depend upon their structure and structural stability. Thus, the functional behavior is influenced by the surface topology as characterized by pockets or protrusions that can interact with other molecules, the chemical characteristics of the surface as determined by the charge and polarity and their distribution, the flexibility of the structure and the structural properties of partially unfolded intermediates between the native and random coil states.

Various means to modify the structure of proteins have been studied, including chemical, enzymatic and genetic methods (Phillips et al., 1994; Kester and Richardson, 1983). Use of enzymes offers many advantages, including the ability to perform modifications under physiological conditions with great specificity and stereoselectivity without undesirable side reactions. They also provide an aesthetic advantage because enzymes are natural products. However, the supply of many enzymes is limited, their activity is labile or they are difficult to purify; hence, their cost prohibits their use in the food industry. For this reason most of the enzymes used commercially are hydrolases obtained by fermentation. Thus, numerous studies have examined the use of proteinases for modification of protein functionality ( Kilara , 1986). However, improvement of the functional characteristics is not always observed (Kuehler and Stine, 1974; Kilara, 1986; Turgeon et al., 1991). Changes in the substrate protein functionality depend upon the protein’s structure, the type of enzyme used and the conditions of the enzyme treatment. For example, tryptic peptides from whey proteins appeared to have better emulsifying properties than did the chymotryptic peptides (Jost and Monti , 1982; Turgeon et al., 1991).

The size of the peptides formed and their secondary and tertiary structures are also extremely important. For example, Jost and Monti (1982) observed improved emulsifying activity and greater emulsion stability as compared with untreated protein using an oligopeptide fraction obtained from a tryptic digest of whey protein by ultrafiltration. Use of an unfractionated tryptic hydrolysate of whey proteins resulted in a decreased emulsifying activity in comparison to the untreated protein; however, when the oligopeptides were isolated and used the emulsifying activity was greater than that of intact protein (Turgeon et al., 1991). Furthermore, the interfacial properties of the oligopeptide b- Lg(f41-60) were much superior to the intact protein and all other tryptic peptides from b- Lg that were examined (Turgeon et al., 1992). It has been known for some time that denatured proteins have inferior functional characteristics (Kilara, 1986). We have shown that large oligopeptide portions of the b-barrel domain of b- Lg can be prepared with retention of secondary and tertiary structure by limited proteolysis with immobilized trypsin (Chen et al., 1993; Huang et al ., 1994). Moreover, the functional properties of this oligopeptide are superior to those of the intact protein (Huang et al., 1996; Swaisgood et al ., 1996).

Thus, it appears that a critical factor that limits the use of proteolysis to improve functionality in food proteins results from the hydrolysis being too extensive (Kester and Richardson, 1984). Of course, the enzymes could be heat-inactivated but this would also denature the oligopeptides released resulting in a loss of functionality. Perhaps this is the reason many studies have found little improvement or an actual decrease in functionality after enzyme treatment. On the other hand, use of immobilized enzymes eliminates the inactivation requirement, prevents enzyme autolysis and provides for precise control of the extent of hydrolysis. This review will summarize our studies of the functional properties of limited proteolysates of whey proteins prepared with immobilized trypsin and the characteristics of protein crosslinking with immobilized transglutaminase. We will also describe a genetic approach for expression of recombinant enzymes designed for one-step purification and immobilization.

LIMITED PROTEOLYSIS OF b-LACTOGLOBULIN

Use of immobilized trypsin allows precise control of the extent of proteolysis of b-Lg. This protein is a member of the lipocalyn family with an eight-stranded antiparallel b-barrel (Papiz et al., 1986; Monaco et al ., 1987) capable of binding many lipophilic ligands in the calyx (Wang et al., 1996). A limited proteolysis results in release of large oligopeptide fractions of the central b-barrel domain of this protein (Chen et al., 1993; Huang et al ., 1994). A major peptide obtained by limited proteolysis is an 8.6 kDa fragment of the b-barrel, b-Lg (f41-100 + 149-162), that retains secondary and tertiary structure which in turn may be reversibly unfolded in urea. This fragment also displays an endothermic peak upon DSC analysis, although the structure is less stable than that of the intact protein (Chen et al., 1993).

Emulsification properties

A 100-mL bioreactor containing trypsin covalently immobilized on succinamidopropyl-Celite (32 TAME units of activity/mL) was used to prepare limited proteolysates. b-Lactoglobulin (5 mg/mL) in 10 mM ammonium acetate, pH 7.5, was recirculated through the bioreactor for 60 min at 24 °C, and the hydrolysate was membrane-fractionated to obtain the oligopeptide fraction as the permeate of a 30-kDa membrane and the retentate of a 3-kDa membrane (Huang et al., 1996). These conditions resulted in a degree of hydrolysis (DH) of 2.4% and a 50% reduction of the original native b-Lg. The 8.6-kDa domain fragment comprised 50% of the oligopeptide fraction.

Figure 1 . Comparison of the emulsifying activities of the domain fraction with that of intact b-lactoglobulin and egg white at various pHs. Taken from data reported by Huang et al. (1996a). Peanut oil (0.5 mL) was dispersed in 1.5 mL of 0.5% protein solution.

The emulsification characteristics of the oligopeptides were determined with peanut oil using the method of Pearce and Kinsella (1978) and the stability of the emulsions were evaluated by the method of Britten and Giroux (1991). As indicated by the data in Figure 1, the emulsifying activity index (EAI) was 2- to 3-fold larger for the oligopeptide fraction as compared to the parent protein throughout the pH range of 3 to 9 that was examined (Huang et al., 1996). The activity index was also considerably larger on a weight basis than that of egg white protein in the higher pH range. Furthermore, the stability of the emulsion formed with the oligopeptide fraction was better than that for the parent protein. After standing for one week at ambient temperature, the emulsion formed with intact b-Lg exhibited two distinct phases; whereas, that formed with the oligopeptide fraction had not separated. Greater stability of the oligopeptide emulsion also was indicated by the formation of smaller and more numerous droplets as viewed by scanning electron microscopy (Huang et al ., 1996).

Ideal emulsifying characteristics of a protein result from a delicate balance of net charge and charge density, the proper distribution of polar and nonpolar residues, the existence of tertiary structure and good structural flexibility (Huang et al., 1996). Excellent solubility and small size, conferred by net charge and charge density and a tertiary structure that buries hydrophobic residues thus preventing aggregation, allow the molecule to diffuse rapidly to the interface. However, structural flexibility is extremely important because the molecule must unfold readily to allow orientation at the interface. Periodic distribution of clusters of hydrophilic and hydrophobic residues allows a stable orientation of the residues in the aqueous and oil phases, respectively (Turgeon et al., 1992). Finally, a remaining flexible tertiary structure in the aqueous phase allows molecular interaction between surface molecules thus giving the emulsion stability. For these reasons, denatured proteins or small peptides resulting from hydrolysis that is too extensive are not good emulsifiers ( Kuehler and Stine, 1974; Puski, 1975; Lakkis and Villota , 1990; Turgeon et al., 1992; Waniska et al., 1981). On the other hand, the 8.6-kDa domain fragment has a net charge of approximately –5.6 and a charge density of 0.39 at pH 6.6 (compared to –10 (Basch and Timasheff, 1967) and 0.30, respectively for the parent protein), a periodic distribution of clusters of polar and nonpolar residues (Swaisgood, 1982), and a tertiary structure that is considerably less stable and therefore more flexible than that of native b-Lg.

Table 1. Comparison of the rheological properties of gelation at 60 ° C for a 15%

b -lactoglobulin solution and a limited proteolysate of b -lactoglobulin (35% reduction

of intact protein). The solutions were prepared in 50 mM TES, pH 7.0, containing 20 mM CaCl_2.¹

b-Lactoglobulin	t_{50% G’} (min)	t_{90% G’} (min)	G’ (Pa)
b-Lactoglobulin	t_{50% G’} (min)	t_{90% G’} (min)	60 °C ²	25 °C ³
Intact protein	527 ± 7	688 ± 4	68.4 ± 15	140 ± 30
Limited proteolysate	388 ± 5	639 ± 5	4913 ± 103	8523 ± 212

¹ Solutions were heated from 25 °C to 60 °C at a rate of 1 °C/min and held for 12 h before cooling at a rate of 1 °C/min to 25 °C.

² Value obtained after holding for 12 h at 60 °C. ³ Value obtained after returning to 25 °C and holding for 80 min.

Gelling Properties

A limited proteolysate of b-Lg, prepared in a similar manner with immobilized trypsin resulting in a 35% reduction in the intact protein, but without membrane fractionation, was used to examine the effect on gelling properties (Chen et al., 1994). Solutions of the limited proteolysate form strong gels at 60 °C in the presence of 20 mM CaCl₂; whereas, the intact protein formed only very weak gels under the same conditions after cooling to 25 °C (Table 1). Furthermore, the rate of gelation for the treated protein was greater as indicated by the times to reach 50% or 90% of the final value for the storage modulus, G’. A 7% solution of the hydrolyzed protein also gelled at 60 °C, although the G’ was 29-fold lower. Gelation of the untreated protein under these conditions is questionable (Chen et al., 1994; Ziegler and Foegeding, 1990).

The gelling characteristics of these limited proteolysates are consistent with the structural stability of the domain fragments. Thus, the unfolding transitions, as measured by T_ms determined from DSC, are in the range of 55 ° to 65 °C; whereas, that for the intact protein under similar conditions is 75 ° to 79 °C (Chen et al ., 1994; Foegeding et al., 1992). Thus, unfolded domain fragments begin to associate at 60 °C, either with themselves or with native or partially unfolded b-Lg, leading to the network formation associated with gelation.

LIMITED PROTEOLYSIS OF WHEY PROTEIN ISOLATE (WPI)

A 10 mg/mL solution of WPI in 10 mM ammonium acetate, pH 7.6, was hydrolyzed by circulation through a fluidized bed bioreactor containing 100 mL of immobilized trypsin for various times at ambient temperature. The percentage composition of the major whey proteins in the original WPI as compared to that of the

Figure 2. Protein composition of WPI and enzyme-treated WPI. Percentages were estimated by size-exclusion chromatography. Taken from data presented by Huang et al. (1999).

total limited proteolysate resulting from a 90-min treatment is shown in Figure 2. Under these conditions, b-Lg is more susceptible to hydrolysis than a-lactalbumin ( a-LA), with a 32% reduction in the amount of intact b-Lg remaining (Huang et al., 1999). Hence, the concentration of domain fragment oligopeptides produced should be similar to that in the studies previously described for the total hydrolysate of b-Lg.

Emulsification characteristics

The emulsifying activity index and emulsion stabilities were determined as previously described (Huang et al . , 1996). As shown by the data in Figure 3, limited proteolysis increased the emulsifying activity throughout the pH range investigated (4.0

Figure 3 . Comparison of the emulsifying activities of b-lactoglobulin with that of limited proteolysates having varying degrees of hydrolysis (DH). Peanut oil (0.5 mL) was dispersed in 1.5 mL of 0.5% protein solution adjusted to the designated pH.

to 8.5). Using varying treatment times gave a range of DH from 2.1 to 4.3. Within this range, the emulsifying activities were similar, although the activities tended to be higher at the higher pH values. The stabilities of the peanut oil emulsions were also improved by limited proteolysis. Upon storage of the emulsions at pH 5.5 for one week at 26 °C, the emulsion formed with untreated WPI had obviously separated into two phases; however, that formed with the hydrolysates, particularly that formed with the DH 2.1 hydrolysate, still exhibited emulsified droplets. These results are similar to those previously described for the emulsifying properties of limited proteolysates of b-Lg. Although the improvement may have been slightly attenuated, perhaps because of the presence of other proteins or peptides, it appears that the release of domain oligopeptides from b-Lg enhances the emulsifying properties of enzyme-treated WPI (EWPI).

Foaming properties

The foaming properties of WPI and EWPI were determined using a micro-scale method (Huang et al., 1997) based on the standard method of Phillips et al.( 1990). This method requires only 4 mL of a 5% protein solution. The foam overrun and foam stability for WPI and limited proteolysates with DHs of 2.8 and 4.3 are compared in Figure 5. The foam overrun of a proteolysate with a DH of 2.8 increased 1.4-fold, while the stability was essentially unchanged (Huang et al., 1997). However, continued

Figure 5 . Effect of the degree of hydrolysis (DH) on the foam overrun and stability of foams formed with WPI and EWPI. Foams were prepared at pH 7.0 with 5% protein solutions.

proteolysis resulting in a DH of 4.3 clearly decreased the overrun and the stability as compared to untreated WPI. These results suggest that protein or peptide size and structure are important factors with regard to foaming characteristics. The importance of molecular size has been suggested previously ( Halling, 1981; Kinsella, 1984). We suggest that a flexible structure is also important because it allows interaction between surface molecules thus providing mechanical strength to the film.

Gelling characteristics

The gelling properties of a limited proteolysate produced by a 90-min treatment of WPI as described above were examined by small strain dynamic rheometry and by texture profile analysis (TPA) of the gels formed (Huang et al., 1999). Solutions of 10% protein (w/v) in 50 mM TES buffer, pH 7.0, containing 50 mM NaCl were treated to form heat-induced gels by heating at a rate of 1 °C/min from 25 °C to 80 °C, holding at 80 °C for 180 min and cooling at the same rate to 25 °C. Values of the rheological parameters are compared for WPI and EWPI in Table 2. These results indicate that WPI formed a more rigid and elastic gel as shown by the larger storage modulus and the lower

Figure 6 . Comparison of the thermal stabilities of WPI, EWPI and domain peptides in 50 mM TES buffer, pH 7.0, containing 50 mM Na⁺ or 20 mM Ca²⁺, as determined by differential scanning calorimetry. Taken from data presented by Huang et al. (1999).

phase angle. Also, the EWPI solution started to gel at 77 °C; whereas, the solution of WPI did not form a gel until it had been held at 80 °C for 1.4 min but the initial gelling rate of the WPI solution was 8-fold larger.

Gels were also formed for visual, microscopic and TPA analysis by heating at 80 °C for 30 min. The appearance of the two gels formed with WPI and EWPI were distinctly different. EWPI gels were opaque while the WPI gels were transparent. Also, the EWPI gels had a more sponge-like structure and expressed water more readily than the WPI gels. Examination of the microstructure by scanning electron microscopy revealed a more ordered-compacted, fine-stranded protein matrix for the WPI gels; whereas, the EWPI gels were more porous with an irregular compact protein network, typical of a particulate gel. Results of texture profile analysis (Table 3) indicate that the EWPI gels exhibited significantly greater hardness, cohesiveness and gumminess in comparison to the WPI gels. Furthermore, the EWPI gels did not fracture at 75% deformation.

Table 2. Rheological properties of WPI and enzyme-treated WPI. Gels were formed with 10% protein in 50 mM TES, 50 mM NaCl, pH 7.0, at 80 ° C. Values given are those obtained after cooling to 25 ° C. ¹

Protein

G’

(kPa)

G’’

(kPa)

(degree) ²

WPI

7.4^a ± 0.1

0.71^a ± 0.01

5.4^a ± 0.1

EWPI

2.0^b ± 0.2

0.26^b ± 0.03

7.3^b ± 0.2

¹Means in a column not followed by the same letter are significantly different (P<0.05). ² It should be noted that tan d = G”/G’.

Considering protein gelation to be a two-stage process of unfolding followed by network formation resulting from protein-protein interactions, it is apparent that interaction of b-Lg with the domain fragments affects one or both of these processes. Recalling that the domain fragments have lower thermal stability than the intact protein (Chen et al., 1993; also see Figure 6), they should unfold prior to the unfolding of b-Lg. The subsequent interaction of the unfolded fragments with intact protein results in a lower gelling point (see above). Several observations indicate an interaction between domain fragments and native or partially unfolded b-Lg. The T_m for b-Lg determined by DSC is increased (Figure 6; also see Barbeau et al., 1996). Also, turbidimetric studies of the thermal aggregation rate showed that dilute solutions of WPI in 50 mM NaCl rapidly aggregated at 80 °C but solutions of EWPI only weakly associated under these conditions (Huang et al., 1999). Furthermore, solutions of the domain fragments in 20 mM CaCl₂ exhibited a maximum aggregation rate at 45 °C while EWPI solutions did not rapidly aggregate until the temperature reached 80 °C. A lower surface hydrophobicity and higher charge density for the domains (Huang et al., 1996) would account for these observations upon their association with intact b-Lg.

Table 3. Results of texture analysis of gels formed with WPI and enzyme-treated WPI.¹

Protein	Springiness	Hardness	Fracturability	Cohesiveness	Gumminess
WPI	0.54^a	200^a	420	0.17^a	30^a
EWPI	0.42^a	1600^b	No fracture	0.42^b	720^b

¹ Gels were formed with 10% protein in50 mM TES, pH 7.0, containing 50 mM NaCl by heating at 80 °C for 30 min. Gels were

cooled to ambient temperature, stored overnight at 4 °C and equilibrated at ambient temperature prior to TPA analysis. Values

are the means of six replications. Means in a column not followed by the same letter are significantly different (P<0.05).

The type of gel structure formed is influenced by the relative rates of protein unfolding and protein aggregation, with a particulate type of structure being favored when the rate of aggregation exceeds that of unfolding. Thus, it appears that at protein concentrations required for gel formation, association of the domain fragments with intact b-Lg leads to rapid association of the unfolded proteins to form large aggregates. An inhomogeneous distribution of protein aggregates and the resulting refractive index dispersion would cause increased light scattering producing an opaque appearance. This hypothesis is substantiated by the observation that addition of 29% domain fragments to 71% WPI to produce a simulated EWPI solution also resulted in a particulate gel (Huang et al., 1999).

CROSSLINKING OF MILK PROTEINS USING IMMOBILIZED TRANSGLUTAMINASE

Proteins can be crosslinked enzymatically using transglutaminase to catalyze the acyl transfer between the g-carboxamide group of glutaminyl residues in proteins and the e-amino group of lysyl residues forming an isopeptide bond. Such reactions can be more extensive than crosslinking by sulfhydryl-disulfide interchange due to the higher frequencies of the former residues in milk proteins (Swaisgood, 1982). As a preliminary study to the development of streptavidin-transglutaminase fusion proteins, we examined the activities of guinea pig liver transglutaminase either immobilized covalently or by adsorption of the biotinylated enzyme on immobilized avidin (Oh et al., 1993; Huang et al ., 1995). The covalently immobilized enzyme was attached to polylysyl- a_S-casein that had been previously immobilized on porous glass (Oh et al., 1993). For adsorption, the enzyme was biotinylated and then adsorbed on avidin that had been previously adsorbed on biotinylated porous glass (Huang et al., 1995). Thus, in both cases, a protein spacer was placed between the enzyme and the surface. Without the spacer, very little activity was observed.

Figure 7 . Progress curves showing the disappearance of native b-lactoglobulin and the formation of polymers (>100 kDa) in b-lactoglobulin solutions (A) or in WPI solutions (B) in 25 mM imidazole buffer, pH 7.5, containing 10 mM DTT and 5 mM CaCl₂. Crosslinking was catalyzed with biotinylated transglutaminase immobilized by selective adsorption on immobilized avidin. The solutions were analyzed by size-exclusion chromatography.

Kinetics of the crosslinking of various milk proteins were determined at pH 7.5 in 25 or 50 mM imidazole buffer containing 5 mM CaCl₂ and 10 mM DTT. The parameters obtained are listed in Table 4. These data indicate that immobilization caused some loss of activity. This result is not surprising since the residues involved with attachment are randomly distributed over the enzyme surface and thus some interactions would cause steric hindrance. In the case of the adsorbed enzyme, biotinylation alone caused a 37% loss of activity (Huang et al. , 1995). Nevertheless, the immobilized enzyme retained a substantial amount of the original activity. Specific activities

Figure 8 . Progress curve showing the hydrolysis of TAME by the solubilized periplasmic space fraction from recombinant E. coli cells harboring the plasmid containing the trypsin- streptavidin gene construct. Cells without this plasmid do not hydrolyze TAME.

measured with the small substrates CBZ-Gln-Gly and hydroxylamine indicated that the adsorbed enzyme retained 30% of the original activity (Huang et al ., 1995).

Flexibility of the structure is an important feature of the substrate. Thus, a_S-casein is a very good substrate, whereas b-Lg is a good substrate only if the disulfide bonds are reduced. The domain fragments from b-Lg are also reasonably good substrates without reduction because of their structural flexibility (Chen et al., 1993). Treatment of a_S-casein with immobilized transglutaminase resulted in the rapid formation of dimers, tetramers and very large polymers; within 40 min, monomer casein was completely absent as determined by SDS-PAGE (Huang et al., 1995). Crosslinking of pure b-Lg and whey protein isolate (WPI) catalyzed by immobilized transglutaminase under reducing conditions was monitored by size-exclusion chromatography (Huang et al., 1997). Progress curves showing the disappearance of native protein and the appearance of polymers >100 kDa in size are given in Figure 7 for solutions of b-Lg and WPI, respectively. In these experiments the disappearance of b-Lg in WPI was measured; however, the composition of the polymer is unknown and may contain other proteins besides b-Lg. These results illustrate the rapid appearance of large polymers.

Table 4. Comparison of the kinetic parameters for crosslinking of various milk proteins as catalyzed by adsorbed and covalently immobilized guinea pig liver transglutaminase. Substrates were prepared in 25 mM imidazole buffer, pH 7.5, containing 5 mM CaCl₂ and 10 mM DTT.

Substrate	K_M ( mM)	k_cat (min^-1)	k_cat/K_M (mM min^-1)^-1
*Adsorbed enzyme:*¹
Macropeptide	46	0.14	3
Domain fragments	60	0.52	9
b-Lactoglobulin	8	0.89	111
a_S -Casein²	13	0.13	10
*Covalently immobilized* :³
b-Lactoglobulin⁴	15	0.26	18
a_S -Casein⁴	19	1.99	106
Soluble enzyme:
a_S -Casein^1,2	3.7	3.15	855
a_S -Casein^3,4	5.7	4.86	853

¹ Measurements were made at 37 °C. ² Taken from Huang et al., 1995, measured in 50 mM imidazole. ³ Measurements were made

at ambient temperature. ⁴ Taken from Oh et al., 1993.

DESIGN OF RECOMBINANT ENZYMES FOR BIOREACTOR PREPARATION

The major factors contributing to the costs of enzyme technology are the cost of purification and the costs of bioreactor preparation and regeneration (Swaisgood, 1991). The design of bifunctional fusion proteins that include an affinity domain allow one-step purification and immobilization by bioselective adsorption (Ong et al., 1991; Walsh and Swaisgood, 1994; Huang et al., 1996). We have developed the use of streptavidin as an affinity

domain in bifunctional fusion proteins (Walsh and Swaisgood, 1994; Walsh and Swaisgood, 1996; Lee and Swaisgood, 1998). Our studies have shown that streptavidin- b-galactosidase can be purified and immobilized in one step from a crude cell lysate by bioselective adsorption on biotinyl moieties covalently attached to supports (Walsh and Swaisgood, 1994). Because of the high affinity of the interaction between streptavidin and biotin, bioreactors prepared in this manner do not exhibit leaching problems associated with other adsorption methods. Furthermore, the biotinyl moiety is extremely robust so the surface can be used repeatedly over long periods of time. After the enzyme has lost activity it can be desorbed with 6 M guanidinium chloride and the bioreactor regenerated by exposure to fresh recombinant fusion protein (Huang et al., 1996). Thus, this technology minimizes the costs of enzyme purification, bioreactor preparation and regeneration and maximizes the biocatalyst specific activity and bioreactor operational stability.

Design of a trypsin-streptavidin bifunctional fusion protein

Our current studies are focused on the design of a trypsin-streptavidin fusion protein that would allow preparation of a trypsin bioreactor for modification of whey protein functionality as described above. A plasmid containing the gene for rat anionic trypsin was obtained from C. S. Craik, University of California, San Francisco. The gene was excised using PCR and cloned in-frame to the 5’-end of the streptavidin gene that previously had been inserted into plasmid vector pET26b. This vector was used to transform Escherichia coli cells and expression of the fusion protein was induced by addition of isopropylthiogalactoside (IPTG). The construct included a signal sequence that should cause its secretion into the periplasmic space. Various cellular fractions were assayed using the tryptic substrate p-tosyl-L-arginine methyl ester (TAME). A progress curve showing the activity in the solubilized particulate fraction from the periplasmic space is given in Figure 8. The activity expressed corresponded to 5 U/ L (TAME) or 288 U/L (BAEE) of cell culture. We have also shown that the activity is bioselectively adsorbed on biotinylated porous glass.

Design of a streptavidin-transglutaminase fusion protein

Figure 9 . Specific activities of standard guinea pig liver transglutaminse and that of recombinant bifunctional fusion protein streptavidin- transglutaminase. The results are for the construct containing the full length streptavidin gene or the core streptavidin formed by residues 23-157. Activities were measured with the small molecular substrates carbobenzoxy-L-glutaminylglycine and hydroxylamine.

The gene for an endothelial cell tissue transglutaminase was obtained from P. J. A. Davies at the University of Texas Medical School. Using PCR, this gene was modified to include new restriction sites and then excised from the plasmid obtained and cloned in-frame to the 3’-end of the streptavidin gene which had been previously inserted into plasmid vector pET26b. Two forms of the streptavidin gene were used and were obtained by PCR modification of the genetic construct described by Walsh and Swaisgood (1994). One construct contained the full length of mature streptavidin while the other represented the “core” structure with biotin-binding capability corresponding to residues 23 to 157. The resulting plasmids, designated pSATG9 and pCSATG3, respectively, were used to transform Escherichia coli cells. Expression of the bifunctional fusion protein by IPTG-induced cells was demonstrated by Western blotting of SDS-polyacrylamide gels of cell lysates using anti- transglutaminase antibodies. There was some evidence of proteolytic cleavage of the fusion protein or translational truncation in the region between the two domains, especially for the fusion containing full-length streptavidin. Nevertheless, the levels of expression of active fusion protein were quite good, approaching 1-2 mg per liter of culture. The cells were lysed and the crude lysate was passed through a column of biotinylated porous glass beads. The specific activities of the immobilized fusion protein were very good as shown by the data in Figure 9. For example, the specific activity of immobilized core-streptavidin-transglutaminase was nearly 50% of that for the soluble guinea pig liver enzyme. These results, as well as SDS-PAGE patterns for the protein released from the beads by treatment with 6 M guanidinium chloride, indicate that the fusion protein was selectively immobilized; i.e., the enzyme activity was purified and immobilized in one step.

CONCLUSIONS

Characteristics of food proteins or peptides that provide good functionality include a flexible tertiary structure, an adequate size, and a net charge and polar/nonpolar residue distribution that gives good solubility and allows orientation at interfaces. Limited proteolysis can release structural domains of proteins that retain tertiary structure but have decreased stability and thus increased flexibility. Use of immobilized proteinases for this purpose allows precise control of the degree of proteolysis and eliminates the requirement of a downstream heat inactivation step that may also destroy the tertiary structure of the oligopeptides released. Limited proteolysis of whey proteins with immobilized trypsin yields large b-barrel domain fragments of b-Lg that retain their tertiary structure. Such treatment of whey protein isolate yields an ingredient that has improved emulsifying activity and emulsion stabilities, improved foaming properties, and altered gelling characteristics.

Crosslinking of proteins is another means for dramatically changing the functional behavior of food proteins. Sulfhydryl-disulfide interchange has long been known to be an important means of crosslinking to alter functionality. Recently, however, the catalytic activity of transglutaminase has been recognized as an important tool for changing the functional characteristics of proteins in foods. Addition of this enzyme to protein solutions of sufficient concentration eventually results in the formation of gels. Once again, to control precisely the degree of reaction and to avoid an inactivation step, it may be desirable to use an immobilized form of the enzyme. We have shown that inclusion of a protein spacer between the surface and the enzyme permits immobilization with retention of activity. Crosslinking of casein or whey proteins with immobilized transglutaminase bioreactors results in rapid production of large soluble polymers.

The major deterrents of industrial use of immobilized enzymes are the costs associated with bioreactor preparation and regeneration. We are developing technology that will minimize these costs and at the same time minimize the costs of enzyme purification. Genetic construction of fusion proteins with streptavidin as an affinity domain allows one-step concentration, purification and simultaneous immobilization of enzyme activity. Biotinylated surfaces are very robust and can be used over long periods of time. Bioreactors can be regenerated by desorbing the inactive enzyme with a strong protein denaturant, 6 M guanidinium chloride, followed by bioselective adsorption of freshly prepared recombinant cell lysate or culture medium.

REFERENCES

Barbeau, J., Gauthier, S. F., and Pouliot , Y. 1996. “Thermal Stabilization of b-Lactoglobulin by Whey Peptide Fractions”, J. Agric. Food Chem., 44:3939-3945.

Basch, J. J. and Timasheff , S. N. 1967. “Hydrogen Ion Equilibria of the Genetic Variants of Bovine b-Lactoglobulin”, Arch. Biochem . Biophys. ,118:37-47.

Britten, M. and Giroux, H. J. 1991. “Coalescence Index of Protein-stabilized Emulsions”, J. Food Sci., 56:792-795.

Chen, S. X., Hardin, C. C. and Swaisgood, H. E. 1993. “Purification and Characterization of b-Structural Domains of b-Lactoglobulin Liberated by Limited Proteolysis”, J. Protein Chem., 12:613-625.

Chen, S. X., Swaisgood, H. E. and Foegeding, E. A. 1994. “Gelation of b-Lactoglobulin Treated with Limited Proteolysis by Immobilized Trypsin”, J. Agric. Food Chem., 42:234-239.

Foegeding, E. A., Kuhn, P. R. and Hardin, C. C. 1992. “Specific Divalent Cation-induced Changes during Gelation of b-Lactoglobulin”, J. Agric. Food Chem., 40:2092-2097.

Halling, P. J. 1981. “Protein-stabilized Foam and Emulsions”, CRC Crit. Rev. Food Sci. Nutr., 15:155-203.

Huang, X. L., Catignani, G. L. and Swaisgood, H. E. 1994. “Comparison of the Size and Rate of Formation of Peptides Released by Limited Proteolysis of b-Lactoglobulins A and B with Immobilized Trypsin”, J. Agric. Food Chem., 42:1281-1284.

Huang, X. L., Catignani, G. L. and Swaisgood, H. E. 1995. “Immobilization of Biotinylated Transglutaminase by Bioselective Adsorption to Immobilized Avidin and Characterization of the Immobilized Activity”, J. Agric. Food Chem., 43:895-901.

Huang, X. L., Catignani, G. L. and Swaisgood, H. E. 1996a. “Improved Emulsifying Properties of the b-Barrel Domain Peptides Obtained by Membrane-fractionation of a Limited Tryptic Hydrolysate of b-Lactoglobulin”, J. Agric. Food Chem., 44:3437-3443.

Huang, X. L., Walsh, M. K. and Swaisgood, H. E. 1996b. “Simultaneous Isolation and Immobilization of Streptavidin - b-Galactosidase: Some Kinetic Characteristics of the Immobilized Enzyme and Regeneration of Bioreactors”, Enzyme Microbial Technol., 19:378-383.

Huang, X. L., Catignani, G. L. and Swaisgood, H. E. 1997. “Micro-scale Method for Determining Foaming Properties of Protein”, J. Food Sci., 62(5):1028-1030 & 1060.

Huang, X. L., Catignani, G. L. and Swaisgood, H. E. 1999. “Modification of the Rheological Properties of Whey Protein Isolates by Limited Proteolysis”, Nahrung – Food, 43(2): (In Press).

Jost, R. and Monti , J. C. 1977. “Partial Hydrolysis of Whey Protein by Trypsin”, J. Dairy Sci., 60:1387-1393.

Kester, J. J. and Richardson, T. 1984. “Modification of Whey Proteins to Improve Functionality”, J. Dairy Sci., 67:2757-2774.

Kilara, A. 1986. “Enzyme-modified Protein Food Ingredients”, Process Biochem., 20:149-157.

Kinsella, J. E. 1984. “A Review of Milk Proteins: Physicochemical and Functional Properties”, CRC Crit . Rev. Food Sci. Nutr ., 21:197-262.

Kinsella, J. E., Rector, D. J. and Phillips, L. G. 1994. “Physicochemical Properties of Proteins: Texturization via Gelation, Glass and Film Formation”, in Protein Structure-function Relationships in Foods, eds., R. Y. Yada, R. L. Jackson and J. L. Smith, London: Blackie Academic & Professional, pp. 1-21.

Kuehler, C. A. and Stine, C. M. 1974. “Effect of Enzymatic Hydrolysis on Some Functional Properties of Whey Protein”, J. Food Sci., 39:379-382.

Lakkis, J. and Villota, R. 1990. “A Study on the Foaming and Emulsifying Properties of Whey Protein Hydrolysates”, in Food Emulsions and Foams, AIChE Symposium Series 277, Vol. 86, New York , NY , pp. 87-101.

Lee, P. and Swaisgood, H. E. 1998. “Cloning and Expression of a Streptavidin-lipase Fusion Gene in Escherichia coli and Characterization of the Immobilized Fusion Protein”, Enzyme Microb. Technol . 22:246-254.

Monaco, H. L., Zanotti , G., Spandon , P., Bolognsi, M., Sawyer, L., and Eliopoulos, E. E. 1987. “Crystal Structure of the Trigonal Form of Bovine b-Lactoglobulin and of Its Complex with Retinol at 2.5 Å Resolution”, J. Mol. Biol., 197:695-706.

Oh, S., Catignani, G. L. and Swaisgood, H. E. 1993. “Characteristics of an Immobilized Form of Transglutaminase: A Possible Increase in Substrate Specificity by Selective Interaction with a Protein Spacer”, J. Agric. Food Chem., 41:1337-1342.

Ong, E., Gilkes, N. R., Miller, R. C. Jr., Warren, R. A. J., and Kilburn, D. G. 1991. “Enzyme Immobilization Using a Cellulose-binding Domain of a Cellulomonas fimi exoglucanase”, Enzyme Microb. Technol., 13:59-65.

Papiz, M. Z., Sawyer, L., Eliopoulos, E. E., North, A. C. T., Findlay, J. B. C., Sivaprasadarao, R., Jones, T. A., Newcomer, M. E. and Kraulis, P. J. 1986. “The Structure of b-Lactoglobulin and Its Similarity to Plasma Retinol-binding Protein”, Nature, 324:383-385.

Phillips, L. C., German, J. B., O’Neill, T. E., Foegeding , E. A., Harwalkar, V. R., Kilara, A., Lewis, B. A., Mangino, M. E., Morr, C. V., Regenstein, J. M., Smith, D. M. and Kinsella, J. E. 1990. “Standardized Procedure for Measuring Foaming Properties of Three Proteins, a Collaborative Study”, J. Food Sci., 55:1444-1453.

Phillips, L.C., Whitehead, D. M., and Kinsella , J. E. 1994. Structure-Function Properties of Food Proteins. San Diego, CA : Academic Press, Inc., pp. 107-249.

Pulski, G. 1975. “Modification of Functional Properties of Soy Proteins by Proteolytic Enzyme Treatment”, Cereal Chem., 52:655-664.

Swaisgood, H. E. 1982. “Chemistry of Milk Proteins”, in Developments in Dairy Chemistry-1, ed., P. F. Fox, London : Applied Science, pp. 1-59.

Swaisgood, H. E. 1991. “Immobilized Enzymes: Applications to Bioprocessing of Food”, in Food Enzymology, Vol. 2, ed., P. F. Fox, New York, NY : Elsevier Applied Science, pp. 309-341.

Turgeon, S. L., Gauthier, S. F. and Paquin , P. 1991. “Interfacial and Emulsifying Properties of Whey Peptide Fractions Obtained with a Two-step Ultrafiltration Process”, J. Agric. Food Chem., 39:673-676.

Turgeon, S. L., Gauthier, S. F., Mollé , D. and Léonil, J. 1992. “Interfacial Properties of Tryptic Peptides of b-Lactoglobulin”, J. Agric. Food Chem., 40:669-675.

Walsh, M. K. and Swaisgood, H. E. 1994. “An Escherichia coli Plasmid Vector System for Production of Streptavidin Fusion Proteins: Expression and Bioselective Adsorption of Streptavidin- b-galactosidase”, Biotechnol. Bioengr ., 44:1348-1354.

Walsh, M. K. and Swaisgood, H. E. 1996. “Investigating the Use of the Chymosin-sensitive Sequence of k-Casein as a Cleavable Linker Site in Fusion Proteins”, J. Biotechnol., 45:235-241.

Wang, Q., Allen, J. C. and Swaisgood, H. E. 1997a. “Binding of Retinoids to b-Lactoglobulin Isolated by Bioselective Adsorption”, J. Dairy Sci., 80:1047-1053.

Wang, Q., Allen, J. C. and Swaisgood, H. E. 1997b. “Binding of Vitamin D and Cholesterol to b-Lactoglobulin”, J. Dairy Sci ., 80:1054-1059.

Waniska, R. D., Shetty , J. K. and Kinsella, J. E. 1981. “Protein-stabilized Emulsions: Effects of Modification on the Emulsifying Activity of Bovine Serum Albumin in a Model System”, J. Agric. Food Chem., 29:826-831.

Ziegler, G. R. and Foegeding, E. A. 1990. “The Gelation of Proteins”, Adv. Food Nutr. Res., 34:203-298.