Food functionalities and bioactivities of protein isolates recovered from skipjack tuna roe by isoelectric solubilization and precipitation.

Four roe protein isolates (RPIs) from skipjack tuna were prepared using isoelectric solubilization (pH 11 and 12) and precipitation (pH 4.5 and 5.5) (ISP) at different pH points to evaluate their physicochemical and functional properties and in vitro bioactivities. Moisture (<6.3%) and protein (71%-77%) content were maintained. Sulfur, sodium, phosphorus, and potassium were the major elements, and glutamic acid and leucine were the prevalent amino acids (12.2-12.8 and 9.6-9.8 g/100 g protein, respectively) in RPIs. RPI-1 showed the highest buffering capacity at pH 7-12. RPIs and casein showed similar water-holding capacities. At pH 12, RPI-1(pH 11/4.5) showed the highest solubility, followed by RPI-3(pH 12/4.5), RPI-2(pH 11/5.5), and RPI-4(pH 12/5.5) (p < .05). Oil-in-water emulsifying activity indices of RPI-1 and RPI-3 significantly differed. At pH 2 and 7-12, pH-shift treatment improved the food functionality of RPIs, which was superior to positive controls (casein and hemoglobin). RPI-1 showed ABTS+ radical scavenging (102.7 μg/ml) and angiotensin-converting enzyme inhibitory activities (44.0%).


| INTRODUC TI ON
Skipjack tuna (Katsuwonus pelamis) is consumed worldwide because of its abundance, nutritional value, and sensorial attributes (firm texture and flavorful flesh). This tuna is widely used as a raw material for sashimi and sushi in Korea and Japan . In 2015, 51,201 metric tons of skipjack tuna were canned, accounting for 57% of all canned products in Korea (KOSIS, 2016). The annual discard rate of fish processing byproducts has increased with increasing fish production. The fish processing industry generates large quantities of byproducts such as heads, skin, frames, roe, viscera, and scales (Narsing Rao, Prabhakara Rao, Satyanarayana, & Balaswamy, 2012), most of which are not used as sources of high value-added products for human or animal consumption and are disposed of as waste. However, these byproducts are excellent sources of proteins . Upcycling or reuse of byproducts has gained wide attention in the seafood processing industry owing to its economic effectiveness . In recent years, numerous studies have examined the use of protein-rich seafood byproducts as nutraceuticals or protein and lipid supplements (Kim, Yoon, Shim, & Lim, 2016).
Among fish byproducts, roes are nutritionally rich in essential fatty acids, minerals, and amino acids (Heu et al., 2006;Park et al., 2016). The production of caviar and roe products from other fish species has been examined previously (Bledsoe, G., Bledsoe, C., & Rasco, 2003).
Roe, a byproduct of fish processing (3.0%-30.0% depending on species), contains 11% albumin, 75% ovoglobulin, and 13% collagen (Sikorski, 1994) and is generally used as animal feed or in pet food preparation. The underutilized skipjack tuna roe (STR) requires processing methods for conversion to more marketable and acceptable forms, such as protein concentrates, isolates, and extracts. Previous studies characterized the ingredients, chemical composition (Heu et al., 2006;, fractions, and characteristics of the protease inhibitors (Kim et al., 2013), concentrates Yoon, Lee, et al., 2018), and protein isolates  in fish roes.
Isoelectric solubilization and precipitation (ISP) promotes the selective pH-induced water solubility of muscle proteins, segregates lipids, and removes materials unfit for human consumption, such as bones, scales, and skin (Hultin & Kelleher, 1999). ISP is used for protein recovery from inaccessible sources such as fish (Chen & Jaczynski, 2007), chicken (Tahergorabi, Beamer, Matak, & Jaczynski, 2011), and beef (Mireles, Gomez, & James, 2002) processing byproducts; additionally, the fat content in the recovered protein isolate is significantly reduced. A major technology was developed for feasible extraction of functional protein isolates from low-value raw materials (dark muscle fish and fatty fish) and seafood processing byproducts (fish cutoffs and fish frame) (Nolsoe & Undeland, 2009).
Proteins recovered from the ISP process exhibit desirable functional properties and may have better gelling properties than proteins obtained by conventional surimi processing (Kristinsson, Theodore, Demir, & Ingadottir, 2005). Proteins are the basic functional components of processed foods and determine the textural and nutritional properties and contribute to the quality and sensorial attributes of foods (Mohamed, Xia, Issoufou, & Qixing, 2009;Park et al., 2016).
No studies have examined the protein isolate preparation of skipjack tuna roe or its food functional characteristics and bioactivity. The purpose of this study was to investigate the physicochemical and functional characteristics of protein isolates recovered from skipjack tuna roe using the ISP process, as well as their antioxidative and antihypertensive bioactivities.

| Sample
Skipjack tuna roe was obtained from Dongwon F & B Co., Ltd.
(Changwon, Korea), stored at −70°C in sealed polyethylene bags, and transferred to the laboratory. Frozen roe was partially thawed for 24 hr at 4°C and then cut into small pieces to 2.0 cm thickness and minced with a food grinder. Minced roe was frozen at −20°C until use.

| Preparation of roe protein isolates (RPIs)
Roe protein isolates were prepared as described by . The frozen minced roe was homogenized with deionized distilled water (DDW) at a 1:6 (w/v) ratio using a homogenizer. The homogenate was adjusted to pH 11 and 12 with 2N NaOH. Solubilization was allowed to occur at 4°C for 1 hr, followed by centrifugation at 12,000 × g and 4°C for 30 min. After centrifugation, two alkali-soluble (pH 11 and 12) supernatant fractions were collected.
To prepare the RPIs from the alkali-soluble fractions by acid precipitation, the pH of the fractions was readjusted to pH 4.5 and 5.5 (similar to the isoelectric point of fish proteins) by adding 2N HCl. The suspensions were centrifuged at 12,000 × g and 4°C for 30 min. Precipitates obtained after ISP were washed with DDW and centrifuged at 12,000 × g and 4°C for 30 min to remove the NaCl.

| Proximate composition
The proximate composition was determined by the AOAC method (AOAC, 2005). Moisture content (950.46) was determined by placing a 0.5-g sample in an aluminum pan and drying in a forced air convection oven at 105°C until a constant mass was reached. The ash content (920.153) was determined by charring approximately 0.2 g sample in a ceramic crucible over a hot plate and then heating in a muffle furnace at 550°C until a constant mass was achieved. The total crude protein content (928.08) was determined by the semimicro Kjeldahl method. Total lipid content (960.39) was determined by the Soxhlet extraction method. Sample (5 g) was extracted with dimethyl ether for 30 min at a drip rate of 10 ml/min. The total lipid content was determined by gravimetry and expressed as a percentage.

| Protein concentration
The protein concentration of the samples was determined as described by Lowry, Rosebrough, Farr, and Randall (1951) using bovine serum albumin as a standard.

| Total amino acid content
The samples were hydrolyzed by 6 N HCl at 110°C for 24 hr on a heating block and filtered using a vacuum filtration device. Amino acids were quantified with an amino acid analyzer (Biochrom 30, Biochrom Ltd.) using sodium citrate buffers (pH 2.2) in step gradients.
The results are reported in milligrams of amino acid per 100 g protein.

| Mineral analysis
The mineral content of the samples was determined by inductively coupled plasma optical emission spectrophotometry (Optima 4300 DV, PerkinElmer). The samples were dissolved in 10 ml of 70% (v/v) nitric acid and heated on a hot plate until digestion was complete. The volume of the samples (in duplicates) was made up to 100 ml with 2% nitric acid in a volumetric flask. The mineral concentration was calculated and expressed as milligrams per 100 g sample.

| Color value
The samples were equilibrated to 20 ± 2°C for 2 hr prior to Hunter color measurement. Color values were determined using a color meter (ZE-2000 Nippon Denshoku Inc.

| SDS-polyacrylamide gel electrophoresis (SDS-PAGE)
The molecular weight distribution of proteins was investigated by SDS-PAGE, which was performed according to the method of Laemmli (1970). Briefly, 20 mg sample was solubilized in 5 ml of 5% SDS. The protein solution was then mixed at a 4:1 (v/v) ratio with SDS-PAGE sample treatment buffer (pH 6.8) and boiled at 100°C for 3 min. Samples (20 µg protein) were loaded into a 10% Mini-PROTEAN ® TGX™ precast gel and electrophoresed at a constant current of 10 mA using a Mini-PROTEAN ® tetra cell (Bio-Rad).

| Buffer capacity
Buffer capacity was measured as described by Park et al. (2016).
The sample (300 mg) was dispersed in 30 ml distilled water, and measured volumes of 0.5 M NaOH and HCl were added in small increments. The corresponding changes in pH in alkaline and acidic ranges were recorded. The amounts of base and acid added were plotted against the pH, and the buffer capacity in each range was expressed as the mean value of NaOH or HCl (in millimolar) required per gram sample to induce a unit change in pH.

| Water-holding capacity (WHC)
The WHC of samples was measured as described by Park et al. (2016). The sample (300 mg) was mixed with 30 ml DDW in a 50-ml centrifuge tube. The mixture was thoroughly vortexed for 10 min at 20 ± 2°C and centrifuged at 12,000 × g for 20 min at 4°C. WHC was determined from the difference in masses and expressed as grams of water absorbed per gram protein.
where C is protein concentration (%).

| Protein solubility
The protein solubility of samples was determined according the method of Park et al. (2016). First, 300 mg sample was dispersed in 30 ml DDW and the pH was adjusted to 2-12 with 2N HCl and 2N NaOH. The mixture was stabilized at room temperature for 30 min prior to centrifugation at 12,000 g for 20 min. Protein content in the supernatant and total protein content in the sample were determined by Lowry's method after solubilizing the sample in 2N NaOH.
Protein solubility was calculated as follows: Each measurement was replicated at least five times, and the results were expressed as the means ± SD. Foaming capacity (FC) and foam stability (FS) of the sample solution (1%, w/v) were determined as described by Park et al. (2016) with slight modifications. Ten milliliters of 1% (w/v) sample solution was transferred to a 25-ml volumetric cylinder. The solution was homogenized (Polytron ® PT 1200E, Kinematica AG) at 12,500 rpm for 1 min at 20 ± 2°C. The sample was allowed to stand for 0, 15, 30, and 60 min, and FC and FS were calculated using the following equations: where VT is the total volume after whipping, V 0 is the initial volume, and Ft and Vt are the total foam and total volume, respectively, after standing at room temperature for different durations (t = 15, 30, and 60 min).

| Emulsifying properties
The emulsifying activity index (EAI) and emulsion stability index (ESI) were determined as described by Park et al. (2016). Soybean oil (Ottogi Co., Ltd.) and 1% (w/v) sample at a 1:3 (v/v) ratio were homogenized at a rate of 12,500 rpm for 1 min. An aliquot of the emulsion (50 μl) was pipetted from the bottom of the volumetric cylinder 0 and 10 min after homogenization and mixed with 5 ml of 0.1% SDS solution. The absorbance of the mixture was measured at 500 nm (UV-2900, Hitachi).
The absorbances measured immediately (A 0min ) and at 10 min (A 10min ) after emulsification were used to calculate EAI and ESI as follows: where A is the absorbance at 500 nm, DF is the dilution factor (100), l is the path length of the cuvette (1 cm), φ is the oil volume fraction (0.25), and C is the protein concentration in the aqueous phase (g/ml).
where ΔA = A 0min -A 10min and Δt = 10 min. A 0min and A 10min are the absorbance values measured at 0 and 10 min after emulsification, respectively.
2.7 | Antioxidative and antihypertensive activity 2.7.1 | ABTS + radical scavenging activity ABTS + radical scavenging activity was determined using the ABTS assay with slight modification based on the method of Binsan et al. (2008). Fresh ABTS + solutions in ethanol were prepared for each assay.
Samples (1 ml) were mixed with 3 ml ABTS + solution and incubated at room temperature for 30 min in the dark. Absorbance was measured at 734 nm with a spectrophotometer. The IC 50 value was defined as the concentration required to scavenge 50% of ABTS + radical.
Absorbance was measured immediately (A 734 ), and ABTS + radical scavenging activity was calculated as follows: where control 734 is the absorbance of the same reaction system without the sample.

| Tyrosinase inhibitory activity
The tyrosinase inhibitory activity was determined as described by Iida et al. (1995) with some modifications. Briefly, 900 μl (50 U/ml reaction mixture) mushroom tyrosinase was preincubated with the sample in 50 mM phosphate buffer (pH 6.8) for 30 min at room temperature.
Next, 300 μl 10 mM L-DOPA was added to the reaction mixture, and enzyme activity was monitored at room temperature by measuring the change in absorbance at 475 nm (UV-2900, Hitachi) for 30 min at 1-min intervals, which corresponded to the formation of dopachrome.
Samples without inhibitor were used as controls. The percentage inhibition of enzyme activity by the active compounds was calculated as follows: where control 475 is the absorbance of a reaction system without the sample.

| Angiotensin-converting enzyme (ACE) inhibitory activity
Angiotensin-converting enzyme inhibitory activity was estimated as described by Cushman and Cheung (1971) Tyrosinase inhibitory activity (%) = (Control 475 − Sample 475) Control 475 × 100 spectrophotometer. Absorbance was measured immediately (A 228 ) and used to calculate the ACE inhibitory activity as follows: where the sample blank is the absorbance of the inactivated sample before adding HHL and control blank is the absorbance of the inactivated control before adding HHL.

| Statistical analysis
All analyses were performed in triplicate. Data were averaged, and the SD was calculated. Data were analyzed by analysis of variance using SPSS 12.0 K software (SPSS, Inc.) for Windows. Mean comparisons were performed using the multiple range Duncan's test (p < .05).
However, these values were lower than those of the positive controls (85.5 ± 0.0% for casein, 94.4 ± 0.4% for hemoglobin). The protein content of RPIs was higher than those reported for fish protein powders from arrowtooth flounder and herring (Sathivel et al., 2004). The protein content in fish eggs (Sathivel, Yin, Bechtel, & King, 2009) and surimi (Huda, Abdullah, & Babji, 2001) was similar to those in the RPIs. Lipid (13.8 ± 2.5-15.6 ± 1.0%) and ash content (1.8 ± 0.6-3.6 ± 0.1%) in the RPIs were lower than those in FDSTR (p < .05). This reduction in lipid and mineral content occurred because of mineral and fat migration into the processed water during ISP (Lee et al., 2017;. The protein content of RPI-2 and RPI-4 precipitated at pH 5.5 after alkaline solubilization was significantly higher than that of RPI-1 and RPI-3 precipitated at pH 4.5 (p < .05). In contrast, the total yield (10.3-10.9 g/100 g) and protein yield (7.8-7.9 g/100 g) of RPI-1 and RPI-2 were significantly lower than those of RPI-3 and RPI-4 (p < .05). Thus, alkali solubilization at pH 12 was better than that at pH 11, while acid precipitation did not affect protein yield (p < .05).

| Color values
The color properties of FDSTR and the RPIs are shown in FDSTR, freeze-dried skipjack tuna roe; Hb, hemoglobin. ND: not determined; RPI-1 and RPI-2, roe protein isolate adjusted to pH 4.5 and 5.5, respectively, after alkaline solubilization at pH 11; RPI-3 and RPI-4, roe protein isolate adjusted to pH 4.5 and 5.5, respectively, after alkaline solubilization at pH 12. † Yield is weight (g) of each sample obtained from 100 g of raw STR. ‡ Protein yield (g) = yield × protein (%). differences in color values may have resulted from the fractionation effect of the pigment induced by ISP. The relatively low L* and high a* values indicate that the product was brown. The color and whiteness of the fish protein isolates may be partially affected by the presence of connective tissues, which increase brightness, whereas lipid retention can affect yellowness. Additionally, the yellow-brownish color of the product may be attributed to the deposition of heme proteins, which may affect redness, or to the denaturation and oxidation of hemoglobin (Kristinsson & Rasco, 2000).

| Total amino acids
The total amino acid composition (g/100 g protein, %) of FDSTR, RPIs, and positive controls is shown in Table 2. The protein content of all samples ranged from 75.8% to 96.3% based on the dry weight.
Intarasirisawat, Benjakul, and Visessanguan (2011) reported that the major nonessential amino acids (NEAA) in defatted tuna roes were glutamic acid and aspartic acid, while leucine and lysine were the predominant essential amino acids (EAA), which agrees with our results. The lysine content in the RPIs was higher than that in Channa striatus (6.94%) and Lates calcarifer (6.86%) (Narsing Rao et al., 2012).
The EAAs in the RPIs ranged from 52.9% to 53.8%, which was higher than that in FDSTR (50.1%). The EAA/NEAA ratio in the RPIs ranged from 1.12 to 1.17, which was higher than that in casein (0.92) and slightly lower than that in hemoglobin (1.33). The protein nutritional value of the RPIs was improved by 12%-17% compared with that of FDSTR because of the decrease in NEAA content during ISP.
The amino acid compositions of FDSTR and the RPIs were similar to those of mullet, cod, pollock, chinook salmon roe, and yellowfin tuna roe isolates (Bledsoe et al., 2003;. The hydrophobic amino acid content in the RPIs and positive controls was 45.0%-45.9% and 44.6%-49.5%, respectively. The significantly lower content of proline and glycine in the RPIs compared with in FDSTR may be because of the precipitation of collagenous material (alkaline insolubles) during alkali solubilization . However, the hydrophobic amino acid content, apart from glycine and proline, in the RPIs was significantly higher than that in FDSTR (p < .05). Overall, the EAA content in FDSTR and the RPIs were higher than those in casein (47.9%) but lower than those in hemoglobin (57.1%). Therefore, the concen-

| SDS-PAGE
The SDS-PAGE profiles of the STR and RPIs are shown in Figure 1.
The major protein fractions of STR were observed at 100-75, 50-37, 25, 15, and 15-10 kDa. A dense protein band at 103 kDa was detected in the soluble and insoluble fractions of Alaska walleye pollock roe (Bechtel, Chantarachoti, Oliveira, & Sathivel, 2007), and four dense protein bands of 40-100-kDa were observed in spraydried catfish protein powder (Sathivel et al., 2009). In the RPI protein bands, the 42-kDa protein band of STR was absent and the 75-100-kDa bands were weaker than those of STR. The disappearance of the 42-kDa protein band indicates that this protein was not readily soluble during alkaline solubilization. The protein bands of acidprecipitated RPIs after alkaline solubilization were 75-100, 25, and 10-15 kDa.

| Buffer capacity
Buffer capacity is defined as the volume (ml) or amount (mmol) of HCl or NaOH required to induce a unit change in pH. The buffer capacities of the RPIs and positive controls are shown in Figure 2a.
Over a pH range of 2-6, RPI-4 required 23.2 mM HCl (average) to cause unit pH change, which is higher than that required by the other RPIs (19.9-22.8 mM HCl). In contrast, in the pH range of 6-12, RPI-1 required a higher alkali concentration (63.9 mM NaOH) to cause a unit pH change compared with the other RPIs (39.9-51.8 mM NaOH).

| WHC
Water-holding capacity depends on protein-water interactions that affect protein function. Mohamed, Xia, Issoufou, and Qixing (2012) reported that interactions between proteins and water or oil are important in food systems, as they affect the flavor and texture of food. WHCs (g per g protein) with or without pH-shift treatment (pH 2.0-12.0) of the RPIs and positive controls are shown in Figure 2b. In the control samples, the WHCs of RPI-1 and RPI-3 (5.1 and 5.4 g/g protein, respectively) were higher than those of RPI-2, RPI-4, and casein (3.3-3.8 g/g protein). After alkaline solubilization, the WHCs of RPI-1 and RPI-3 (recovered at pH 4.5) were significantly higher than those of RPI-2 and RPI-4 (recovered at pH 5.5). This may be because of the fractionation effect between the recovered RPIs during ISP.
The WHC of solubilized hemoglobin (0.9 g/g protein) after pH-shift treatment was significantly lower than those of the RPIs and casein (p < .05).
TA B L E 2 Amino acid composition (g/100 g protein) of protein isolates recovered from skipjack tuna roe (RPIs) using isoelectric solubilization and precipitation process reported that WHCs of protein isolates from tilapia were 2.63-2.51 g/g protein, which are lower than those of the RPIs found in this study. Azadian, Nasab, and Abedi (2012) reported that the WHC of minced fish reached a minimum near the isoelectric point (pH 4-6). Chalamaiah et al. (2013) reported that the lack of polar amino groups on the surface of protein molecules reduced the WHC because these polar groups are responsible for protein-water interactions.
The WHC of RPIs from STR was superior to that of yellowfin tuna roe , Labeo rohita fish egg (Balaswamy, Jyothirmayi, & Rao, 2007), and mrigal egg protein concentrates (Chalamaiah et al., 2013). Particularly, our results showed that structural changes in proteins were actively induced at extreme pH (pH 2 and pH 10-12), which increased the WHC by more than 5-8-fold compared with the controls. This is because acidic and alkaline pH shifts cause protein conformational changes in the RPIs, allowing hydrophilic amino acids to easily access the surrounding water and increase the WHC.

| Protein solubility
Solubility is an important functional property of proteins, as it affects rheological, hydrodynamic, and surface activity characteristics. Protein solubility is also important in many protein-based formulations . The solubilities (%) of the RPIs with and without pH-shift treatment are shown in Figure 2c. The solubilities of the control RPIs (1.7 ± 0.0-6.7 ± 0.3%) were significantly lower (p < .05) than that of hemoglobin (71.3 ± 16.8%). However, the solubility of casein was the lowest (0.4%), and the protein was nearly insoluble in a 1% dispersion. The solubility of pH-shifted RPIs F I G U R E 1 SDS-PAGE patterns of protein isolates isolated from skipjack tuna roe (RPIs) using isoelectric solubilization and precipitation. M, protein maker; STR, skipjack tuna roe; RPI-1, RPI-2, RPI-3, and RPI-4, roe protein isolates adjusted to pH 4.5 and 5.5 after alkaline solubilization at pH 11 and 12 F I G U R E 2 Buffer capacity (a), water-holding capacity (b), and solubility (c) of protein isolates recovered from skipjack tuna roe (RPIs) using isoelectric solubilization and precipitation without (control) and with pH-shift treatment. Values are expressed as the mean ± SD of triplicate determinations. Means with different letters within the sample and pH are significantly different at p < .05 by Duncan's multiple range test. Hb, hemoglobin increased significantly at pH 2 (17.8 ± 0.3-33.1 ± 12.4%) compared with the controls. At pH 12, their solubilities ranged from 69.1 ± 25.8 to 92.9 ± 18.7%, indicating a higher rate of solubility increase at alkaline pH. After alkaline solubilization, the solubilities of RPI-1 and RPI-3 after acid precipitation at pH 4.5 were significantly higher than those of RPI-2 and RPI-4 recovered at pH 5.5. However, the solubility of hemoglobin (82.3 ± 4.0-99.9 ± 17.9%) was not affected by pH-shift treatment (pH 2-12). Casein at pH 2 (42.7 ± 4.9%) and pH 7-12 (83.8 ± 7.2-90.3 ± 9.0%) showed significantly higher solubility than the controls. The RPIs and casein showed the lowest solubilities near the isoelectric point of pH 4-6 because of acid-and alkali-limited protein solubilization. These results indicate that extreme pH changes (such as pH 2 and 12) affected protein solubility by exposing more charged and polar groups to the surrounding water (Kristinsson & Rasco, 2000). The pH-dependent solubility of proteins is important in food functional properties and food system applications, particularly when the pH is below 4 or over 7, and is affected by protein-protein and protein-solvent interactions and the surface hydrophilic-hydrophobic balance (Gbogouri, Linder, Fanni, & Parmentier, 2004). The high solubility of fish proteins is important in food applications because it affects other functional properties such as foam and emulsification characteristics (Kristinsson & Rasco, 2000).

| EAI and ESI
The oil-in-water EAI and ESI were determined to evaluate the ability of the samples to emulsify in foods such as soups, sauces, confectionery breads, and dairy products. The EAI (m 2 /g protein) and ESI (min) of the RPIs and positive controls are shown in Table 3. The EAI of the dispersed RPIs (control 1) was 1.8 ± 0.1-3.9 ± 0.5 m 2 /g protein, and RPI-3 showed the highest EAI (3.9 ± 0.5 m 2 /g protein).
However, the EAI of the RPIs was significantly lower than that of hemoglobin (18.4 ± 2.1 m 2 /g protein) (p < .05). After centrifugation (control 2), the EAI of the supernatant of the dispersed samples was generally increased, except for that of RPI-4 (2.3 ± 0.5 m 2 /g protein).
The increase in EAI and ESI after centrifugation may be related to the presence of insoluble particles in the dispersion that interfere with emulsion layer formation. EAI assessment of RPIs and positive controls after pH-shift (pH 2-12) treatment showed that RPI-2, RPI-4, casein, and hemoglobin had the lowest EAIs (2.3 ± 0.5-12.9 ± 1.7 m 2 /g protein) at pH 4 and RPI-1 and RPI-3 had the lowest EAIs (13.3 ± 0.8 and 9.2 ± 0.4 m 2 /g protein, respectively) at pH 6 ( Table 3). These results suggest that the FC and emulsifying activity are closely related to protein solubility in food, which was minimal in the RPIs and positive controls at pH 4-6. At pH 2 and 7-12, the EAIs of all samples, except for hemoglobin, were higher than that of control 2. At all pH ranges except for pH 2, the EAIs of RPI-1 and RPI-3 were higher than those of RPI-2 and RPI-4. The EAI of the RPIs showed no significant differences (p > .05) at pH 2 and 7-12; however, the emulsifying activity significantly increased with pH (p < .05). The emulsion stability did not increase in proportion to the increase in emulsifying activity with pH. The large deviation in emulsion stability after pH-shift treatment may have occurred because of the nonuniformity of emulsified particles. Mutilangi et al. (1996) reported that a higher content of high-MW peptides or TA B L E 3 Foaming capacity (FC, %), foam stability (FS, %, 60 min), emulsifying activity index (EAI, m 2 /g of protein), and emulsion stability (ESI, min) with pH-shift of protein isolates recovered from skipjack tuna roe (RPIs)  Table 4 shows the ABTS + radical scavenging activity (IC 50 , μg/ml), tyrosinase inhibitory activity (%), and ACE inhibitory activity (%)

| Antioxidative and antihypertensive activity
of RPI-1. The ABTS + radical scavenging assay can be performed on both lipophilic and hydrophilic compounds and has been widely used as an antioxidant activity assay. The ABTS + radical scavenging activity (IC 50 ) of RPI-1 (1.5 ± 0.1 mg protein/ml) was 102.7 ± 1.0 μg/ml, which was better than those of alcalase and Protamex hydrolysates from shrimp processing byproducts (160 and 170 µg/ml, respectively) . UV irradiation can generate reactive oxygen species that influence skin pigmentation. Tyrosinase inhibitors have recently gained attention in the medical and cosmetic industries because of their association with hyper-pigmentation (Choi, Kim, & Lee, 2011). The tyrosinase inhibitory activity of RPI-1 was 13.5 ± 1.7%, and some whitening effects were predicted. Choi et al. (2011) reported a tyrosinase inhibitory activity of 31% for tuna cooking drip, but this value increased with the dose of gamma irradiation. Choi et al. (2017) reported that anchovy muscles not subjected to subcritical water hydrolysis inhibited tyrosinase activity by 14.65%. In these experimental results and reports, tyrosinase inhibitory activity was also observed in proteinous materials containing proteins or amino acids, but its inhibitory activity was low. The inhibition of ACE, a key enzyme that regulates blood pressure, has been recognized as an effective therapy for treating hypertension. The ACE inhibitory activity of RPI-1 was 44.0 ± 4.9%. The current search for natural ACE-inhibiting peptides has extended into seafood protein sources, particularly seafood byproducts. The hydrolysates of the byproducts of skin (Ngo, Ryu, & Kim, 2014) and yellow sole frame  showed 35%-86% ACE inhibitory activities, which are consistent with our results. These results suggest that RPIs from STR possess antioxidative and antihypertensive activities, which can be improved by enzymatic hydrolysis.

| CON CLUS ION
The aims of this study were to investigate the physicochemical and functional properties of protein isolates recovered from skipjack tuna roe (STR) by ISP. The roe protein isolates (RPIs) were similar or superior to the positive controls (casein and hemoglobin) and other fish protein isolates in terms of buffering capacity, foaming ability, and emulsifying ability. The RPIs were rich in proteins containing essential amino acids and exhibited suitable functional characteristics for supplementing surimi-based products and traditional foods. They also showed in vitro antioxidant and antihypertensive activities. This study suggests that high value-added products can be developed from fish roes, which are currently underutilized in the seafood processing industry.

ACK N OWLED G M ENTS
This research was a part of the projected titled "Development and commercialization of traditional seafood products based on the Korean coastal marine resources (2016)," funded by the Ministry of Oceans and Fisheries, Korea.

CO N FLI C T O F I NTE R E S T
The authors claim no conflicts of interest.

E TH I C A L A PPROVA L
This study did not involve human or animal testing.