T:Tox

icolog

y&Ch

emica

lFoo

dSafe

ty

Detection of Soybean Proteins in FermentedSoybean Products by Using Heating ExtractionNaoki Morishita, Takashi Matsumoto, Fumiki Morimatsu, and Masatake Toyoda

Abstract: Soybean is used in processed foods worldwide. Because soybean can cause adverse reactions in some atopicpatients, appropriate labeling regarding its content in processed foods is needed to better protect consumers. In theprevious study, we developed a reliable sandwich Enzyme Linked Immunosorbent Assay (ELISA) method with highsensitivity and specificity for detecting soybean proteins by using antibody to Gly m Bd 30K, which was originallycharacterized as a vacuolar protein with a molecular mass of 34 kDa in soybean. The ELISA displayed satisfactoryrepeatability and reproducibility in an interlaboratory evaluation. However, it could not detect soybean protein infermented soybean products. We therefore developed an extraction method combined with a heating process to inhibitsoybean protein degradation by microbial proteolytic enzymes in fermented soybean products. This extraction methodenables the sensitive detection of soybean protein in fermented soybean products such as natto and miso. It was able todetect with high-sensitivity soybean protein present at 10 μg/g levels in model processed foods. This method is suitablefor quantifying soybean protein in processed foods without the degrading effects of microbial proteolytic enzymes. Thepresent extraction method can be used sensitively to monitor labeling systems in a reliable manner and should be usefulfor the mandatory inspections required under Japanese regulations.

Keywords: Enzyme Linked Immunosorbent Assay (ELISA) kit, fermentation, food allergen, food safety, soybean

Practical Application: The extraction and ELISA methods that we developed enable sensitive detection of soybean proteinin soybean products, including fermented foods. These methods should be useful for reliable and sensitive monitoring ofproduct labeling systems and should help to solve the problem of insensitive in soybean labeling of processed foods.

IntroductionFood allergies are important diseases. The clinical manifesta-

tions of food allergies vary from mild symptoms such as oral al-lergy syndrome or mild urticaria to severe anaphylactic reactionswith fatal consequences. Approximately, 8% of children and 2%of adults in industrialized countries have food allergies (Jansenand others 1994; Ebisawa and others 2003). To prevent possiblelife-threatening reactions, the only effective treatment is to strictlyavoid the consumption of these allergenic foods. We thereforeneed sufficient information about potentially allergenic ingredi-ents in food products. In 1999, the Joint FAO–WHO Codex Al-imentary Commission agreed to label 8 kinds of food ingredientthat have components known to be allergenic; these include soy-bean (FAO of the United Nations 1995). In Japan, the Ministry ofHealth, Labor, and Welfare (MHLW) has enforced a labeling sys-tem for allergenic food materials since April 2002 (MHLW 2002).In this system, labeling for 7 food ingredients (egg, milk, wheat,buckwheat, peanuts, shrimp, and crab) is mandatory, and labelingis recommended for 20 other food materials, including soybeanand walnut. Because of the almost unlimited uses for soybean andthe fact that the number of patients with soybean allergy has beenincreasing, the most important of these recommended labelings is

MS 20131805 Submitted 12/3/2013, Accepted 3/12/2014. Authors Morishita,Matsumoto, and Morimatsu are with Nippon Meat Packers. Inc., 3-3 Midorigahara,Tsukuba, Ibaraki, 300-2646, Japan. Author Toyoda is with Jissen Women’s Univ.,1-1 Oosakaue 1-chome, Hino, Tokyo, 191-8510, Japan. Direct inquiries to authorMorishita (E-mail: n.morishita@nipponham.co.jp).

Author disclosures: None.

that for soybean (Savage and others 2010). The high functionalityand good processing properties of soybean have promoted its usein processed foods (for example, tofu, soybean milk, meat alter-natives, and fermented soybean products) worldwide. Fermentedsoybean products such as soy sauce, miso, natto, and tempeh areproduced by fermentation processes using microorganisms. Thesefoods are traditional in Japan and other Asian countries (Golb-itz 1995; Murooka and Yamshita 2008; Phromraksa and others2008; Namgung and others 2010). Fermented soybean productsretain their allergenicity (Astwood and others 1996; Tsuji and oth-ers 1997; Ito and others 2005; Kobayashi 2005; Frias and others2008; Phromraksa and others 2008; Inomata and others 2012).Accordingly, to better protect the consumer, we need adequatelabeling information about the presence of soybean in processedfoods. In the previous study, we developed a reliable sandwich En-zyme Linked Immunosorbent Assay (ELISA) method with highsensitivity and specificity for detecting soybean proteins by us-ing antibody to Gly m Bd 30K, which has been characterizedas a vacuolar protein with a molecular mass of 34 kDa (Mor-ishita and others 2008). The ELISA displayed sufficient repeata-bility and reproducibility in an interlaboratory evaluation (Sakaiand others 2010). However, it cannot detect soybean protein infermented soybean products (Morishita and others 2008). Otherstudies aimed at detecting soybean protein by using ELISA havebeen reported (Brandon and others 1991; Tsuji and others 1997;Bando and others 1998; Moriyama and others 2005; Hei and oth-ers 2012). However, reliable measurement using these methods isconsidered difficult in processed foods because of cross-reactivityand low repeatability and reproducibility. In addition, these meth-ods cannot detect soybean protein in fermented soybean products.

C© 2014 Institute of Food Technologists R©doi: 10.1111/1750-3841.12461 Vol. 79, Nr. 5, 2014 � Journal of Food Science T1049Further reproduction without permission is prohibited

T:Toxicology&ChemicalFoodSafety

Detection of soybean in fermented food . . .

We therefore developed a present extraction method that includeda heating process to detect soybean protein in fermented soybeanproducts. We showed that the present extraction method couldbe used to detect soybean protein in fermented soybean productsbecause the microbial proteolytic enzymes in these products weredeactivated by the heating process.

Materials and Methods

Food materialsSoybeans and commercial processed foods were purchased at

local supermarkets (Ibaraki, Japan) between 2009 and 2012.

Chemicals and reagentsAcetone, 2-amino-2-hydroxymethyl-1,3-propanediol (Tris),

hexane, 2-mercaptoehanol (2-ME), methanol, polyoxyethylene-sorbitan monolaurate (Tween 20), sodium chloride (NaCl),sodium dodecyl sulfate (SDS), sodium hydrogen carbonate(NaHCO3), sucrose, and sulfuric acid (H2SO4) were purchasedfrom Wako Pure Chemical Industries, Ltd. (Osaka, Japan). Bovineserum albumin (BSA) and phosphate-buffered salts (PBS) werepurchased from Sigma-Aldrich Co. (St. Louis, Mo., U.S.A.) andTakara Bio, Inc. (Shiga, Japan), respectively.

Protein extraction from food samplesProtein was extracted from food samples in accordance with a

previously reported method (MHLW 2002; Morishita and others2008; Sakai and others 2010). Food samples were homogenizedin a food processor (IFM-700G, Iwatani Intl. Co., Osaka, Japan).Nineteen milliliters of extraction buffer (120 mM Tris-HCl [pH7.4], 0.1% BSA, 0.05% Tween 20, 2% SDS, 2% 2-ME) was addedto 1 g of food homogenate. The mixture was then extractedfor 16 h by shaking at 90 to 110 rpm. This extraction methodwas defined as the “previous extraction method.” In contrast,with the “present extraction method” we used heating instead ofshaking. The present method was as follows. Nineteen millilitersof extraction buffer was added to 1 g of food homogenate and 5glass beads 5 mm in diameter to make the good dispersibility of thesample in the extraction buffer for heating process. The mixturewas extracted by heating at different temperatures (25, 40, 60, 80,or 100 °C) in a water bath for different times (5, 15, or 60 min),with vortexing every 5 min. Food extracts produced by using theprevious and the present methods were centrifuged at 3000g for20 min, and the supernatant was filtered through a filter paper (5AFilter Paper, Toyo Roshi Kaisya, Ltd., Tokyo, Japan). The filtratewas collected as the food sample extract and used immediately foranalysis.

Preparation of calibration standard solutionsTo detect soybean proteins by using ELISA, a calibration stan-

dard solution needs to be prepared. This solution was prepared inaccordance with the official Japanese guideline (MHLW 2002). A300-mg sample of soybean powder was added to a 20-mL mixtureof 20 mM Tris-HCl (pH 7.5), 0.5 M NaCl, 0.5% SDS, and 2%2-ME. The mixture was then shaken for 16 h at room temper-ature for extraction. The extract was centrifuged at 10000g for30 min, and the supernatant was filtered through a 0.8-μm mi-crofilter paper (DISMIC-25CS, Toyo Roshi Kaisya Ltd.). Theprotein content of the initial extract was assayed with a 2-DQuant Kit (GE Healthcare UK Ltd., Amersham, England). Theinitial extract was diluted to 50 ng/mL with 0.1 M PBS (pH7.4), 0.1% SDS, 0.1% 2-ME, 0.1% BSA, and 0.1% Tween 20

and stored as the calibration standard solution for ELISA at 4 °C.All experiments were performed using this calibration standardsolution.

To prepare a standard solution for western blot, the initial extractwas mixed 1:1 with Laemmli buffer (Bio-Rad Laboratories, Inc.,Hercules, Calif., U.S.A.) containing 2.5% 2-ME. The sample wasboiled at 100 °C for 5 min and then cooled in flowing water. Itwas stored at 4 °C.

Preparation of model processed foodsThe best sources of information on the performance of meth-

ods used to detect allergens are model processed foods. Threekinds of model processed food (hamburger steak, rice gruel, andsweet adzuki bean soup) were prepared in accordance with themethod used in a previous report (Morishita and others 2008).To give a final protein concentration of 10 μg/g (soybean pro-tein weight/sample weight) in the model processed foods, theamount of soybean powder used to spike each model processedfood was calculated. The individual ingredients were spiked withthe soybean powder and then cooked according to the followingmethod. Hamburger steak consists of pork, salt, sugar, ice water,and soybean powder. The hamburger steak was packed in a retortpouch (SP-1000K, Mita Rika Kogyo Co., Ltd., Osaka, Japan) andcooked at 80 °C for 20 min. Rice gruel consists of rice, sugar,water, and soybean powder. The rice gruel was packed in a retortpouch and cooked at 121 °C for 10 min. Sweet adzuki bean soupconsists of adzuki beans, sugar, water, and soybean powder. Thesweet adzuki bean soup was packed in a retort pouch and cookedat 100 °C for 10 min. The cooked model processed foods werecooled by passing the retort pouch under flowing water and thenstored at −20 °C.

ELISAPolyclonal antibody against p34, the allergenic protein Gly m

Bd 30K, was prepared according to the method used in a pre-vious report (Morishita and others 2008). Three hundred gramsof soybean was homogenized in 2 L of a 0.1-M Tris-HCl (pH8.6), 10 mM 2-ME mixture with a juicer (MX-X6, MatsushitaElectric Industrial Co., Ltd., Osaka, Japan). The soybean sameas the soybean used for to make the standard for the calibrationcurve was used. The homogenate was centrifuged at 20000g tocollect the oil-body pad. The oil-body pad was resuspended ina 5-time volume of 0.1 M Na2CO3 for 1 h on ice, and the su-pernatant containing the protein was collected by centrifugationat 20000g. The p34 was purified by gel filtration chromatography(Superdex 75, 26/90 cm, GE Healthcare UK Ltd.) equilibratedwith 0.1 M Na2CO3 and then used as an antigen to obtain theantibody. The antigen was mixed 1:1 with Freund’s complete ad-juvant and injected into Japanese white female rabbits as a 1stinjection. After the 1st injection, injections of antigen and Fre-und’s incomplete adjuvant were performed 4 more times at 2-wkintervals over a total period of 8 wk. One week after the finalinjection, whole blood was collected. The antibody was purifiedagainst the affinity-column-fixed antigen. The care and use of theexperimental animals followed the “Ethical Guidelines of Ani-mal Care, Handling, and Termination” prepared by Nippon MeatPackers.

A microtiter plate (F8 Maxisorp Nunc-Immuno module, Nunc,Roskilde, Denmark) was coated with the diluted polyclonal anti-body against p34 in 1 M NaHCO3 (pH 8.5) (100 μL/well) andincubated for 16 h at 4 °C. After 2 washings with a washing buffer(100 mM PBS [pH 7.4], 1% Tween 20, 250 μL/well), the plate

T1050 Journal of Food Science � Vol. 79, Nr. 5, 2014

T:Tox

icolog

y&Ch

emica

lFoo

dSafe

ty

Detection of soybean in fermented food . . .

was blocked with a blocking buffer (100 mM PBS [pH 7.4], 1%BSA, 5% sucrose, 200 μL/well) for 3 h at 25 °C. The blockingbuffer was removed, and then the plate was dried for 16 h at 25 °C.The food sample extract was diluted 1:19 in dilution buffer A (100mM PBS [pH 7.4], 0.1% BSA, 0.1% Tween 20, 0.1% SDS, 0.1%2-ME). The calibration standard solution was diluted to concen-trations of 0.78, 1.56, 3.13, 6.25, 12.5, 25, and 50 ng/mL withdilution buffer B (a mixture of dilution buffer A and extractionbuffer at 19:1). The diluted food sample extract and the calibra-tion standard solution were added to the plate (100 μL/well) andincubated for 1 h at 25 °C. All experiments were performed intriplicate. After 5 washings, diluted biotin-labeled polyclonal an-tibody against p34 (100 μL/well) was added to the plate and themixture was then incubated for 1 h at 25 °C. After the plate hasbeen washed, a diluted horseradish peroxidase-conjugated strep-tavidin (Poly-HRP streptavidin N200, Thermo Fisher ScientificInc., Rockford, Ill., U.S.A.; 100 μL/well) was added to the plateand the mixture was incubated for 30 min at 25 °C. After the platehad been washed again, 3,3′-5,5′-tetramethylbenzidine (TMBE-1000, Moss, Inc., Pasadena, Md., U.S.A.; 100 μL/well) was addedand the plate was incubated for 20 min at 25 °C. The reaction wasstopped by the addition 100 μL of 0.25 M H2SO4. Absorbance wasmeasured with a plate reader (Sunrise 2000, Wako Pure Chemi-cal Industries, Ltd.) at dominant and subdominant wavelengths of450 and 620 nm, respectively. A 4-parameter logistic analysis wasused to calculate the concentration of the soybean soluble pro-tein in the diluted food sample extract. The value of the soybeansoluble protein in the foods was converted by using a dilutionfactor (×400).

Western blotWestern blot using monoclonal antibody against p34 was per-

formed to detect p34 in fermented soybean products. Mono-clonal antibody against p34 was obtained from Kansai Univ. ofWelfare Sciences (Ogawa and others 1991, 1993). The samplesolution was mixed 1:1 with Laemmli buffer (Bio-Rad Labora-tories) containing 2.5% 2-ME. The sample was boiled at 100 °Cfor 5 min and then cooled in flowing water. The sample and amolecular weight marker (Precision Plus Protein All Blue Stan-dard, Bio-Rad Laboratories) were applied at 10 μL/lane to 15%separation gel (E-T15L, Atto Co., Tokyo, Japan) in a runningbuffer (10× Tris–glycine–SDS buffer, pH 8.3, Bio-Rad Laborato-ries). Electrophoresis was performed at a constant electric currentof 20 mA/gel. The protein was blotted onto a PolyVinylideneDiFluoride (PVDF) membrane (Hypond-P, Wako Pure ChemicalIndustries, Ltd.) in a blotting buffer [a mixture of 10× Tris/glycinebuffer, pH 8.3 (Bio-Rad Laboratories), methanol, and distilled wa-ter at 1:2:7] by using a blotting system (Trans-Blot SD Cell, Bio-Rad Laboratories). Blotting was performed at a constant electriccurrent of 2 mA/cm2 for 1 h. The membrane was blocked for 30min at 25 °C with 1× TBS (Bio-Rad Laboratories) containing0.1% BSA and 0.05% Tween 20. After 3 washings with a wash-ing buffer (1× TBS containing 0.05% Tween 20), the membranewas shaken with a diluted monoclonal antibody against p34 inblocking buffer for 1 h at 25 °C. After 3 washings, the mem-brane was shaken with a diluted peroxidase-labeled anti-mouseIgG antibody (Envision System, Dako Japan, Inc., Tokyo, Japan)in blocking buffer for 30 min at 25 °C. After 3 washings, themembrane was detected with 3,3′-5,5′-tetramethylbenzidine (EzWestblue, Atto Co.). The reaction was stopped by adding distilledwater.

Results and Discussion

Development of extraction method with a heating processIn previous research, we developed a reliable sandwich ELISA

method with high sensitivity and specificity for detecting soybeanproteins by using antibody to Gly m Bd 30K (Morishita and others2008; Sakai and others 2010). The limit of detection of the ELISAwas 1 μg/g. However, the ELISA could not detect soybean pro-teins in fermented soybean products, likely because of degradationof the soybean proteins by microbial proteolytic enzymes remain-ing in the fermented products. Figure 1 shows the results of sodiumdodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE)analyses of natto and miso from which proteins were extractedby using the previous extraction method (extraction for 16 h withshaking at 90 to 110 rpm). A band of high molecular mass, includ-ing BSA with a molecular mass of 66 kDa, was not observed ineither the natto or the miso extract after application of the previousextraction method (lane 2). Thus the soybean protein and BSA inthe extraction buffer were degraded into low-molecular-weightpeptides by the microbial proteolytic enzymes remaining in thenatto or miso after the 16 h extraction. The microbial proteolyticenzymes likely inhibited the detection of soybean protein in thefermented soybean products.

We developed the present extraction method to inhibit pro-tein degradation by the microbial proteolytic enzymes. Generally,heating, pH, and protease inhibitors are used to inhibit microbialproteolytic enzymes. Because of utility and cost considerations,we selected an extraction method that uses heating. To determinethe best heating temperature and time for inhibiting microbialproteolytic enzymes, we examined different heating temperaturesand extraction times (Figure 1 and 2).

Bands of high molecular mass were observed in both natto andmiso upon extraction using heating at 80 or 100 °C (Figure 1).On the other hand, no high-molecular-mass bands were detected

Figure 1–SDS–PAGE analyses of natto and miso subjected to extraction byusing the previous method or to heating at different temperatures (25, 40,60, 80, or 100 °C) in a water bath for 15 min. (A) SDS–PAGE of natto; (B)SDS–PAGE of miso. Lanes: 1, molecular weight markers (250, 150, 100,75, 50, 37, 25, 20, 15, 10 kDa); 2, previous method (nonheating extractionfor 16 h); 3, heating extraction at 25 °C; 4, heating extraction at 40 °C;5, heating extraction at 60 °C; 6, heating extraction at 80 °C; 7, heatingextraction at 100 °C. Arrow = BSA band (molecular weight 66 kDa).

Vol. 79, Nr. 5, 2014 � Journal of Food Science T1051

T:Toxicology&ChemicalFoodSafety

Detection of soybean in fermented food . . .

Table 1–Detection of soybean protein in model processed foods.

Previous extraction method Present extraction method

Food Mean (μg/g) Recovery (%) Mean (μg/g) Recovery (%) Present method / previous method (%)

Hamburger steak 10.3 103.0 9.8 98.0 95.1Rice gruel 10.5 104.8 9.5 94.8 90.5Sweet adzuki bean soup 10.4 103.7 8.9 89.0 85.8

Detection of soybean protein in 3 kinds of model processed food (hamburger steak, rice gruel, and sweet adzuki bean soup) by using the present and previous extraction methods.The model processed foods contained soybean protein at approximately 10 μg/g. Mean = average concentration (soybean protein weight/food weight) of soybean proteins.Recovery = mean/10 (μg/g).

in natto or miso upon extraction using heating at 25 or 40 °C.Temperatures of 80 °C or more therefore likely inhibited thedegradation of soybean protein by microbial proteolytic enzymes.Degradation of soybean protein in natto was stronger than in miso.We speculated that this difference in degradation was due to dif-ferences in the fermentation process used to make natto and miso.Microbial proteolytic enzymes remain in natto, because heating isnot performed to destroy microorganisms after fermentation.

We performed SDS–PAGE analyses of natto and miso afterextraction by heating at 80 °C in a water bath for different times(5, 15, or 60 min) (Figure 2). Bands of high molecular mass weredetected in both natto and miso after extraction for 15 or 60 min.These results suggested that the suitable heating time for extractionwas 15 min or more. Accordingly, the heating conditions chosento optimize the inhibition of microbial proteolytic enzymes were80 °C for 15 min. We therefore set the heating temperature forthe extraction to 80 °C and the time to 15 min. This protocol wascalled the “present extraction method.”

Figure 2–SDS–PAGE analyses of natto and miso subjected to extractionby heating at 80 °C in a water bath for different times (5, 15, or 60 min).(A) SDS–PAGE of natto; (B) SDS–PAGE of miso. Lanes: 1, molecular weightmarkers (250, 150, 100, 75, 50, 37, 25, 20, 15, 10 kDa); 2, heating for5 min; 3, heating for 15 min; 4, heating for 60 min. Arrow = BSA band(molecular weight 66 kDa).

Table 2–Detection of soybean protein in commercial processedfoods.

Previous Present Presentextraction extraction method / previousmethod method method

Food Mean (mg/g) Mean (mg/g) (%)

Roasted soybean flour 104.6 104.8 100.2Soybean milk 41.7 42.5 101.9Yuba (bean curd skin) 58.6 50.2 85.6Soybean boiled in water 51.4 48.0 93.4Tofu 43.5 34.0 78.2Freeze-dried tofu 330.0 318.2 96.4Fried tofu 105.8 98.1 92.7Soy pulp 19.6 23.8 121.2

Detection of soybean proteins in 8 kinds of commercial food by using the present andprevious extraction methods. Mean = average concentration (soybean proteinweight/food weight) of soybean proteins. Recovery = mean/10 (μg/g).

Detection of soybean proteins in model processed foodsand commercial processed foods by using the presentextraction method

We developed a present extraction method that used heating toinhibit the degradation of soybean protein in fermented soybeanproducts by microbial proteolytic enzymes. The conditions of thispresent method were 80 °C for 15 min. In contrast, in previousresearch, we developed a highly sensitive and specific ELISA us-ing the previous extraction method, namely extraction for 16 hwith shaking at 90 to 110 rpm. The extraction time used in thepresent method was therefore much shorter than that used in theprevious one. In Japan, the threshold for food allergen labeling inprocessed foods has been set at 10 μg/g (MHLW 2002). There-fore, the present extraction method needed to be able to sensitivelydetect the relevant soybean protein at 10 μg/g levels in processedfoods (Matsuda and others 2006). To evaluate the capacity of thepresent extraction method, we performed a recovery study us-ing incurred model processed foods (Table 1). The recoveries ofsoybean protein by using the previous extraction method were103.0%, 104.8%, and 103.7% in hamburger steak, rice gruel, andsweet adzuki bean soup, respectively. The recoveries of soybeanprotein using the present extraction method were 98.0%, 94.8%,and 89.0% in hamburger steak, rice gruel, and sweet adzuki beansoup, respectively. The correlation between the results of the pre-vious and present extraction methods was 85.8% to 95.1%. Thusthe present extraction method had good extraction ability thatwas equal to that of the previous method. In addition, the presentextraction method shortened the extraction time from 16 h to15 min.

To evaluate the usefulness of the present method, 8 kinds ofcommercial processed food that included highly concentratedsoybean were subjected to extraction by using the previous andpresent methods; the extracts were then analyzed by using ELISA(Table 2). The correlation between the results of the previous and

T1052 Journal of Food Science � Vol. 79, Nr. 5, 2014

T:Tox

icolog

y&Ch

emica

lFoo

dSafe

ty

Detection of soybean in fermented food . . .

Table 3–Detection of soybean protein in fermented soybeanproducts.

Previous Present Presentextraction extraction method/previousmethod method method

Mean Mean (%)Product (μg/g) (μg/g)

Soybean miso 322.2 4022.4 >1000.0Rice miso 2550.8 6795.6 266.4Barley miso 474.5 2347.6 494.8Raw miso 743.3 4223.2 568.2Soy sauce nd nd –Natto 0.8 1402.0 >1000.0Crushed natto 0.7 835.6 >1000.0Tempeh 46436.0 67420.0 145.2Tianmianjiang 2056.8 1987.6 96.6Douchi 2.5 13079.2 >1000.0Tofu-yoh 4.9 11294.0 >1000.0Soybean milk yoghurt 26184.0 26856.0 102.6

Detection of soybean protein in 12 kinds of fermented soybean product by using theprevious and present extraction methods. Mean = average concentration (soybeanprotein weight/food weight) of soybean proteins; Recovery = mean/10 (μg/g); nd =not detected (<0.03 μg/g);Soybean miso = miso fermented with malted soybean; rice miso = miso fermentedwith malted rice; barley miso = miso fermented with malted barley; raw miso =unheated miso; crushed natto = natto made from ground soybeans; tianmianjiang =sweet soybean paste; douchi = fermented black soybeans; tofu-yoh = fermented tofu.

present extraction methods was 78.2% to 121.2%. In commercialprocessed foods consisting mainly of soybean, the present extrac-tion method had good sensitivity that was equal to that of theprevious method.

Detection of soybean protein in fermented soybeanproducts by using the present extraction method

Previous studies have detected soybean proteins by using ELISAbased on antibodies to soybean proteins such as the trypsin in-hibitor 7S-conglycinin and 11S glycinin (Hitchcock and others1981; Ravestein and Driedonks 1986; Brandon and others 1991;Koppelman and others 2004). However, to our knowledge therehave been no reports of the successful detection of soybean proteinsin fermented soybean products. Hei and others (2012) constructeda sandwich ELISA based an antibody to β-conglycinin, but thismethod could not detect sufficient soybean protein in fermentedsoybean products. Tsuji and others (1995, 1997) constructed asandwich ELISA method based on a monoclonal antibody to p34,but this method could not detect soybean protein in miso, natto,or soy sauce. We speculated that the microbial proteolytic en-zymes in fermented soybean products cause the degradation ofsoybean proteins, because neither of these methods used heatingas part of the extraction process. On the other hand, Moriyamaand others (2005) developed a sandwich ELISA based an antibodyto β-conglycinin and heating extraction for 5 min. Their methodwas able slightly to detect soybean proteins in natto. We considerthat this poor detection ability was due to the short heating timeused in the extraction.

We developed our present method so as to inhibit the degra-dation of soybean proteins by the microbial proteolytic en-zymes in fermented soybean products. To evaluate the use-fulness of the method, we subjected 12 kinds of fermentedsoybean product to extraction by using the previous andpresent methods, and we analyzed the results by using ELISA(Table 3) and western blot analysis with a monoclonal antibodyagainst p34 (Figure 3). The concentrations of soybean proteinsdetected in 4 kinds of miso and in natto, crushed natto, tem-

peh, douchi, and tofu-yoh were higher with the present methodthan with the previous one (Table 3). The results with the presentmethod were equal to those with the previous one in the case oftianmianjiang and soybean milk yogurt. Soybean proteins couldbe not detected in soy sauce with either method. These resultsshowed that the present method was able to detect soybean pro-teins sensitively in most fermented soybean products.

We examined the results of western blot analyses of fer-mented soybean products using monoclonal antibody against p34(Figure 3). With the present method, bands of p34 were detectedin all fermented soybean products except soy sauce, though thatof p34 were detected very slightly in natto and crashed natto.In contrast, with the previous extraction method the western

Figure 3–Western blot analyses of fermented soybean products subjectedto extraction by using the previous and present methods. (A) Western blotresults after application of the previous method; (B) Western blot resultsafter application of the present method. Lanes: 1, molecular weight markers(250, 150, 100, 75, 50, 37, 25, 20, 15, 10 kDa); 2, standard solution ofsoybean proteins; 3, soybean miso; 4, rice miso; 5, barley miso; 6, raw miso;7, soy sauce; 8, natto; 9, crushed natto; 10, tempeh; 11, tianmianjiang; 12,douchi; 13, tofu-yoh; 14, soybean milk yoghurt. Arrow = p34 band.

Vol. 79, Nr. 5, 2014 � Journal of Food Science T1053

T:Toxicology&ChemicalFoodSafety

Detection of soybean in fermented food . . .

blot did not detect p34 in any of the fermented soybean prod-ucts except tempeh, tianmianjiang, and soybean milk yoghurt.We consider that the soy sauce band was not detected witheither method because the soybean proteins in the soy saucewere degraded completely into amino acids and peptides dur-ing fermentation (Kobayashi 2005). These results showed thatthe present extraction method inhibits the degradation of soy-bean proteins by microbial proteolytic enzymes and thus enablesthe detection of soybean proteins in most fermented soybeanproducts.

ConclusionThe present extraction method combined with heating is a

sensitive, and useful way of detecting soybean protein in pro-cessed foods and fermented soybean products. This methodis suitable for quantifying soybean protein in processed foodswithout the effects of microbial proteolytic enzymes. The pro-posed extraction and ELISA method should be useful in sen-sitively monitoring labeling systems in a reliable manner andthus for the mandatory inspections required under Japaneseregulations.

AcknowledgmentsWe thank Dr. Tadashi Ogawa of Kansai Univ. of Welfare Science

for providing the monoclonal antibody against p34. We also thankAiko Yagi, Shiori Ito, Aiko Kubota, Yuki Suwa, Nanako Yanai,Sonoko Wada, Erika Naito, Kaori Ito, Yuki Nogami, and EmiMatsunaga for their valuable discussions.

Author ContributionsN. Morishita collected test data and drafted the manuscript. T.

Matsumoto and F. Morimatsu discussed the results. M. Toyodadesigned the study, collected test data, and discussed the results.

ReferencesAstwood JD, Leach JN, Fuchs RL. 1996. Stability of food allergens to digestion in vitro. Nat

Biotechnol 14:1269–73.Bando N, Tsuji H, Hiemori M, Yoshizumi K, Yamanishi R, Kimoto M, Ogawa T. 1998.

Quantitative analysis of Gly m Bd 28K in soybean products by a sandwich enzyme-linkedimmunosorbent assay. J Nutr Sci Vitaminol 44:655–64.

Brandon DL, Bates AH, Friedman M. 1991. ELISA analysis of soybean trypsin inhibitors inprocessed foods. Adv Exp Med Biol 289:321–37.

Ebisawa M, Ikematsu K, Imai T, Tachimoto H. 2003. Food allergy in Japan. Allergy ClinImmunol Int 15:214–7.

Food and Agriculture Organization of the United Nations, World Health Organization (FAO).1995. Report of FAO technical consultation on food allergens. November 13–14, Rome(Italy).

Frias J, Song YS, Martinez-Villaluenga C, Gonzalez de Mejia E, Vidal-Valverde C. 2008.Immunoreactivity and amino acid content of fermented soybean products. J Agric FoodChem 9:99–105.

Golbitz P. 1995. Traditional soyfoods: processing and products. J Nutr 125:570S–2S.Hei W, Li Z, Ma X, He P. 2012. Determination of beta-conglycinin in soybean and soybean

products using a sandwich enzyme-linked immunosorbent assay. Anal Chim Acta 13:62–8.Hitchcock CHS, Bailey FJ, Crimes AA, Dean DAG, Davis PJ. 1981. Determination of soya

proteins in food using an enzyme-linked immunosorbent assay procedure. J Sci Food Agric32:157–65.

Inomata N, Nomura Y, Ikezawa Z. 2012. Involvement of poly (γ -glutamic acid) as an allergenin late-onset anaphylaxis due to fermented soybeans (natto). J Dermatol 39:409–12.

Ito M, Kato T, Matsuda T. 2005. Rice allergenic proteins, 14–16 kDa albumin and alpha-globulin, remain insoluble in rice grains recovered from rice miso (rice-containing fermentedsoybean paste). Biosci Biotechnol Biochem 69:1137–44.

Jansen JJ, Kardinaal AF, Huijbers G, Vlieg-Boerstra BJ, Martens BP, Ockhuizen T. 1994.Prevalence of food allergy and intolerance in the adult Dutch population. J Allergy ClinImmunol 93:446–56.

Kobayashi M. 2005. Immunological functions of soy sauce: hypoallergenicity and antiallergicactivity of soy sauce. J Biosci Bioeng 100:144–51.

Koppelman SJ, Lakemond CM, Vlooswijk R, Hefle SL. 2004. Detection of soy proteinsin processed foods: literature overview and new experimental work. J AOAC Int 87:1398–407.

Matsuda R, Yoshioka Y, Akiyama H, Aburatani K, Watanabe Y, Matsumoto T, Morishita N,Sato H, Mishima T, Gamo R, Kihira Y, Maitani T. 2006. Interlaboratory evaluation of twoenzyme-linked immunosorbent assay kits for the detection of egg, milk, wheat, buckwheat,and peanuts in foods. J AOAC Int 89:1600–8.

Ministry of Health, Labor and Welfare of Japan. 2002. Notification. Nr. 106001.Morishita N, Kamiya K, Matsumoto T, Sakai S, Teshima R, Urisu A, Moriyama T, Ogawa

T, Akiyama H, Morimatsu F. 2008. Reliable enzyme-linked immunosorbent assay for thedetermination of soybean proteins in processed foods. J Agric Food Chem 27:6818–24.

Moriyama T, Machidori M, Ozawa S, Maebuchi M, Urade R, Takahashi K, Ogawa T,Maruyama, N. 2005. A novel enzyme-linked immunosorbent assay for quantification of soy-bean β-conglycinin, a major soybean storage protein, in soybean and soybean food products.J Nutr Sci Vitaminol 51:34–9.

Murooka Y, Yamshita M. 2008. Traditional healthful fermented products of Japan. J Ind Micro-biol Biotechnol 35:791–8.

Namgung HJ, Park HJ, Cho IH, Choi HK, Kwon DY, Shim SM, Kim YS. 2010. Metabo-lite profiling of doenjang, fermented soybean paste, during fermentation. J Sci Food Agric30:1926–35.

Ogawa T, Bando N, Tsuji H, Okajima H, Nishikawa K, Sasaoka K. 1991. Investigation of theIgE-binding proteins in soybeans by immunoblotting with the sera of the soybean-sensitivepatients with atopic dermatitis. J Nutr Sci Vitaminol 37:555–65.

Ogawa T, Tsuji H, Bando N, Kitamura K, Zhu YL, Hirano H, Nishikawa K. 1993. Identificationof the soybean allergenic protein, Gly m Bd 30K, with the soybean seed 34-kDa Oil-body-associated protein. Biosci Biotech Biochem 57:1030–3.

Phromraksa P, Nagano H, Boonmars T, Kamboonruang C. 2008. Identification of proteolyticbacteria from Thai traditional fermented foods and their allergenic reducing potentials. J FoodSci 73:M189–95.

Ravestein P, Driedonks RA. 1986. Quantitative immunoassay for soya protein in raw andsterilized meat products. J Food Technol 21:19–32.

Sakai S, Adachi R, Akiyama H, Teshima R, Morishita N, Matsumoto T, Urisu A. 2010. Enzyme-linked immunosorbent assay kit for the determination of soybean protein in processed foods:interlaboratory evaluation. J AOAC Int 93:243–8.

Savage JH, Kaeding AJ, Matsui EC, Wood RA. 2010. The natural history of soy allergy. J AllergyClin Immunol 125:683–6.

Tsuji H, Okada N, Yamanishi R, Bando N, Kimoto M, Ogawa T. 1995. Measurement ofGly m Bd 30K, a major soybean allergen, in soybean products by a sandwich enzyme-linkedimmunosorbent assay. Biosci Biotech Biochem 59:150–1.

Tsuji H, Okada N, Yamanishi R, Bando N, Ebine H, Ogawa T. 1997. Fate of major soybeanallergen, Gly m Bd 30K, in rice-, barley, and soybean-koji miso (fermented soybean paste)during fermentation. Food Sci Technol Int Tokyo 3:145–9.

T1054 Journal of Food Science � Vol. 79, Nr. 5, 2014