UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Elucidation of the interaction between soy protein isolate and simulated beef flavour Moon, Soo Yeun 2007

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_2007-267660.pdf [ 10.03MB ]
Metadata
JSON: 831-1.0100487.json
JSON-LD: 831-1.0100487-ld.json
RDF/XML (Pretty): 831-1.0100487-rdf.xml
RDF/JSON: 831-1.0100487-rdf.json
Turtle: 831-1.0100487-turtle.txt
N-Triples: 831-1.0100487-rdf-ntriples.txt
Original Record: 831-1.0100487-source.json
Full Text
831-1.0100487-fulltext.txt
Citation
831-1.0100487.ris

Full Text

ELUCIDATION OF THE INTERACTION BETWEEN SOY PROTEIN ISOLATE AND SIMULATED BEEF FLAVOUR by SOO YEUN MOON  B. Sc., Seoul National University, Korea, 1992 M . Sc., Seoul National University, Korea, 1994  A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF  DOCTOR OF PHILOSOPHY  in  THE FACULTY OF GRADUATE STUDIES  (Food Science)  THE UNIVERSITY OF BRITISH COLUMBIA April 2007  © Soo Yeun Moon  Abstract  The objective of this research was to explore interactions between aroma compounds in simulated beef flavour (SBF) and soy protein isolate (SPI) that may be involved in the suppression of beefy notes in S B F by SPI. A sensitive and reproducible headspace solid phase microextraction (HSS P M E ) was established to isolate volatile compounds in S B F for analysis by gas chromatography (GC). Volatile and odour-active compounds in S B F were qualitatively evaluated by gas chromatography-mass spectrometry ( G C - M S ) and gas chromatography-olfactometry (GC-O) using detection frequency method, respectively. A total of 70 compounds were tentatively identified including three furans, six S-heterocyclic compounds, ten N-heterocyclic compounds, six aldehydes, three alcohols, and two esters.  O f 49 volatile compounds detected in the sniffing  port of G C - O , the most odour-active included 2-methyl-3-furanthiol, delta-3-carene, alphaterpinene, 2-ethyl-3,6-dimethylpyrazine, and several unidentified odourants. Descriptive analysis (DA) along with G C analysis was conducted to investigate changes in S B F aroma characteristics upon addition of SPI. Five attributes (beefy, roasted, yeasty, soymilk-like and cereal) were selected to assess various mixtures of S B F and SPI. The results from D A confirmed that "roasted", "beefy" and "yeasty" notes were highly positively correlated with S B F concentration, and the beefy related notes were substantially suppressed by increasing SPI content. Fifteen peaks from G C analysis were selected as indicator peaks to represent beef attribute in the mixtures of SPI and S B F . Changes in the release of beefy aroma components o f S B F by addition of ingredients (glucosamine, sucrose, ascorbic acid, and/or polyethylene glycol) to SPI and the changes in SPI protein structure induced by the ingredients were investigated. The reduction of disulfide bonds, increased surface hydrophobicity and increased unordered structure in SPI containing ascorbic acid alone or with polyethylene glycol, along with increased G C peak areas of indicator peaks in those SPI-SBF mixtures, were found to be associated with an increase in the perceived beef characteristic attributes in descriptive analysis. These results provide the basis for further research to elucidate strategies maximizing perception of beefy aroma in soy based products.  ii  Table of Contents Abstract Table of Contents List of Tables List of Figures List of Abbreviations Acknowledgements Dedication Co-Authorship Statements  ii iii vii x xiii xv xvi xvii  CHAPTER 1. OVERVIEW AND LITERATURE REVIEW  1  1.1. General Introduction  1  1.2. Soy Protein Isolate (SPI)  2  1.2.1. Specification and production 1.2.2. Major proteins in SPI 1.2.2.1. Glycinin 1.2.2.2. P-Conglycinin 1.2.3. Soy protein isolates used for functional ingredients 1.2.4. Structure changes in soy proteins 1.2.5. Modification  2 3 3 4 5 6 8  1.3. Flavour of soy products  8  1.3.1. Volatile compounds in soybean and soy products 1.3.2. Interaction o f soy protein with flavour compounds 1.3.2.1. Typical flavour binding model 1.3.2.2. Binding of flavour model compounds to soy protein 1.3.2.3. Soy protein isolate on binding of flavours  8 11 11 12 15  1.4. Beef flavour  16  1.4.1. Addition of flavours to soy based products to simulate beef flavour 1.4.2. Natural beef flavours 1.4.3. Simulated meat flavours  16 16 17  1.5. Analysis of aroma compounds  18  1.5.1. Adsorption of aroma compounds - Headspace solid phase microextraction 19 1.5.2. Identification of aroma compounds 19 1.5.3. Detection of aroma compounds - gas chromatography-olfactometry (GC-O) 19 1.5.3.1. Dilution methods 20 1.5.3.1.1. Aroma extract dilution analysis ( A E D A ) 20 1.5.3.1.2. Combined hedonic aroma response measurements ( C H A R M ) analysis..21 1.5.3.2. Intensity methods: Osme 21 1.5.3.3. Detection frequency method 22  1.6. Structure analysis of SPI 1.6.1. Sulfnydryl and disulfide groups 1.6.2. Surface hydrophobicity 1.6.3. Raman spectroscopy , 1.6.3.1. Characterization and application of Raman spectroscopy 1.6.3.2. Protein structure analysis using Raman spectroscopy  23 23 24 25 25 26  1.7. Hypotheses  29  1.8. Objectives  30  1.9. References  45  CHAPTER 2. DEVELOPMENT OF SOLID-PHASE MICROEXTRACTION METHODOLOGY FOR ANALYSIS OF HEADSPACE VOLATILE COMPOUNDS IN SIMULATED BEEF FLAVOUR  55  2.1. Introduction  55  2.2. Experimental methods 2.2.1. Materials 2.2.2. Fractional factorial design 2.2.3. Solid-phase microextraction procedure 2.2.4. Optimization of adsorption time 2.2.5. Gas chromatography 2.2.6. Statistical analysis  57 57 58 58 59 59 60  2.3. Results and Discussion 2.3.1. Screening o f significant factors on the headspace analysis o f the beef flavour 2.3.2. Effects o f S P M E fibres 2.3.3. Effects o f salt concentration 2.3.4. Effects of adsorption temperature 2.3.5. Effects o f adsorption time  60 60 61 61 62 62  2.4. Conclusion  63  2.5. References  71  CHAPTER 3. ODOUR-ACTIVE COMPONENTS OF SIMULATED BEEF FLAVOUR ANALYZED BY SOLID PHASE MICROEXTRACTION AND GAS CHROMATOGRAPHY-MASS SPECTROMETRY AND OLFACTOMETRY 74 3.1. Introduction  74  3.2. Experimental methods 3.2.1. Materials 3.2.2. Headspace solid phase microextraction (HS-SPME) 3.2.3. G C - F I D analysis 3.2.4. G C - 0 analysis 3.2.5. G C - M S analysis  77 77 78 78 79 79  3.3. Results and Discussion 3.3.1. Identification of volatile compounds 3.3.1.1. Furans and sulfur containing compounds 3.3.1.2. Nitrogen containing compounds 3.3.1.3. Hydrocarbons and carbonyl compounds 3.3.2. Determination of odour-active compounds  80 80 80 83 84 86  3.4. Conclusion  88  3.5. References  99  CHAPTER 4. CHANGES IN AROMA CHARACTERISTICS OF SIMULATED BEEF FLAVOUR BY ADDING SOY PROTEIN ISOLATE AS ASSESSED BY DESCRIPTIVE SENSORY ANALYSIS AND GAS CHROMATOGRAPHY -OLFACTOMETRY 105 4.1. Introduction  105  4.2. Experimental methods 4.2.1. Materials 4.2.2. Sensory analysis 4.2.2.1. Panelist training 4.2.2.2. Descriptive sensory analysis 4.2.3. Gas chromatography 4.2.4. Statistical analysis  107 107 107 107 107 110 Ill  4.3. Results and Discussion 4.3.1. Sensory analysis 4.3.1.1. Panel performance 4.3.1.2. Sensory characteristics of the samples 4.3.2. Selection of indicator peaks (IP) for the beef flavour  Ill 111 112 113 115  4.4. Conclusion  116  4.5. References  132  CHAPTER 5. ASSESSMENT OF ADDED INGREDIENT EFFECT ON T H E INTERACTION OF SIMULATED BEEF FLAVOUR AND SOY PROTEIN ISOLATE BY GAS CHROMATOGRAPHY AND SPECTROSCOPIC TECHNIQUES  134  5.1. Introduction  134  5.2. Experimental methods 5.2.1. Materials 5.2.2. Differential scanning calorimetry (DSC) 5.2.3. Gas chromatography 5.2.4. Sulfhydryl and disulfide content 5.2.5. Surface hydrophobicity 5.2.6. FT-Raman spectroscopy 5.2.7. Sensory evaluation 5.2.8. Statistical analysis  138 138 138 138 139 141 142 142 143  5.3. Results 5.3.1. Differential scanning calorimetry 5.3.2. Gas chromatography 5.3.3. Sulfhydryl and disulfide groups content 5.3.4. Surface hydrophobicity 5.3.5. FT-Raman spectroscopy 5.3.6. Sensory evaluation  144 144 144 145 145 146 147  5.4. Discussion 5.4.1. Changes in peak area by the ingredients in G C analysis 5.4.2. Changes in sulfhydryl and disulfide content by the ingredients 5.4.3. Changes in surface hydrophobicity by the ingredients 5.4.4. Changes in FT-Raman spectra by the ingredients 5.4.5. Descriptive sensory evaluation  148 148 149 150 152 154  5.5. Conclusion  155  5.6. References  172  CHAPTER 6. CONCLUSION  176  6.1. Overall conclusions as related to the proposed hypotheses  176  6.2. Significance of this thesis research to the field of study  178  6.3. Suggestions for future research  179  6.4. References  181  Appendix  182  List of Tables  Table 1.1. Average chemical compositions (% by weight, dry basis) o f defatted soy flour, soy protein concentrate, and soy protein isolate (Adapted from MacLeod and Ames, 1988)  39  Table 1.2. Five major soybean glycinin subunits (G1-G5) and their paired acidic (A) and basic (B) polypeptides (Adapted from L i u , 1997)  40  Table 1.3. Amino acid composition o f P-conglycinin and glycinin o f soybean (Adapted from Krishnan, 2005)  41  1  Table 1.4. Functional properties o f soy protein isolates in food systems (Adapted from Hettiarachchy and Kalapathy, 1997)  42  Table 1.5. Thermodynamic constants for the binding of model flavour compounds to soy protein  43  Table 1.6. Number o f binding sites on glycinin and P-conglycinin for various volatile flavour compounds at three different temperatures (Adapted from O'Keefe et al., 1991a and 1991b)  44  Table 2.1. Column assignment for the 4 factors and 3 interactions in this study based on Taguchi's L27(3 ) orthogonal array  68  Table 2.2. Experimental design based on Taguchi's L 7 ( 3 ) orthogonal array and the measured responses of total area count and the number of peaks in the gas chromatogram for each experimental run  69  13  13  2  Table 2.3. Analysis o f variance o f the main factors and the selected interactions between the factors on the total peak area and the number o f peaks of headspace volatile compounds in simulated beef flavour 70 Table 3.1. Volatile compounds in simulated beef flavour tentatively identified by mass spectral-search ( M S ) and linear retention index (LRI)  91  Table 3.2. Comparison o f volatile compounds in simulated beef flavour, boiled beef and roasted beef tentatively identified by G C - M S coupled with H S - S P M E  95  Table 4.1. Description and references for aroma terms used in the sensory training session for descriptive analysis o f simulated beef flavour and soy protein isolate 121 Table 4.2. Definitions o f the selected 5 attributes agreed upon by panelists through the training sessions in the descriptive analysis for simulated beef flavour and soy protein isolate  122  vii  Table 4.3. Samples used in the sensory evaluation and G C analysis  123  Table 4.4. Summary o f analyses o f variance with F values, mean squares o f error (MSE), and degree of freedom (df) for main effects and their interactions for each o f the five attributes 124 Table 4.5. Adjusted F values o f the sample effects using mean squares o f sample by replication instead of mean squares o f error for each of the five attributes  125  Table 4.6. Results of Duncan's multiple comparison test on mean sensory scores o f each panelist for 5 attributes  126  Table 4.7. Pearson correlation coefficients and probabilities o f sample means for each sensory attribute between individual panelist and the other 7 panelists  127  Table 4.8. F ratio and probability of replication effect on the 5 attributes for individual panelist  128  Table 4.9. Pearson correlation coefficients and probabilities between mean sensory score for each sensory attribute and the concentration of SBF and SPI  129  1  1  1  Table 4.10. The mean intensity values of the 5 attributes for the 12 mixtures o f S B F and SPI in descriptive sensory evaluation  1  130  Table 4.11. The 15 selected indicator peaks for the simulated beef flavour  131  Table 5.1. Effect o f added ingredient in SBF-SPI mixture incubated at R T or 60 °C on the peak areas o f volatile compounds captured by H S - S P M E under the adsorption condition at R T '  162  Table 5.2. Effect o f added ingredient in SBF-SPI mixture incubated at R T or 60 °C on the peak areas o f volatile compounds captured by H S - S P M E under the adsorption condition at 60 ° C ' '  163  Table 5.3. Sulfhydryl groups and disulfide bonds in SPI and SPI treated with various ingredients'  165  Table 5.4. Surface hydrophobicity (So) o f SPI and SPI treated with various ingredients  166  Table 5.5. Tentative assignment of major bands in the FT-Raman spectrum o f SPI and SPI treated with various ingredients (Adapted from Li-Chan, 1996)  167  1  2  2  Table 5.6. Normalized intensity values at selected regions of the FT-Raman spectra o f SPI and SPI treated with various ingredients 168 Table 5.7. Composition o f secondary structure in SPI and SPI treated with various ingredients ' .  169  1 2  viii  Table 5.8. Results of Fisher's least significant difference test on mean sensory scores of each sample for 5 attributes  170  Table 5.9. Location of the mean intensity values of the 5 attributes for the 3 samples in comparison with the samples in Table 4.10  171  1  ix  List of Figures  Figure 1.1. Preparation o f soy protein isolates from defatted soy meal (Adapted from Hettiarachchy and Kalapathy, 1997)  31  Figure 1.2. Mechanisms for dissociation of soybean glycinin from the hexameric form possessing one disulfide bond (SS) into its subunits and further into the acidic (A) and basic (B) polypeptides (Adapted from Nielson, 1985)  32  Figure 1.3. Schematic subunit interaction between P-conglycinin (7S globulin) and glycinin (1 IS globulin) on heating (Adapted from Yamauchi et al., 1991). , interaction by secondary force; S-S, disulfide bond; O , acidic polypeptide; • , basic polypeptide; A , a, a'-polypeptide; • , P-polypeptide  33  Figure 1.4. Example of flavour dilution chromatogram from aroma extract dilution analysis (Source: Blank, 1997)  34  Figure 1.5. Example o f Charm chromatogram from CharmAnalysis (Source: Mistry et al., 1997)  35  Figure 1.6. Example of osmegram from intensity methods (Source: Blank, 1997)  36  Figure 1.7. Example o f aromagram from detection frequency method (Source: Le Guen et al., 2000)  37  Figure 1.8. The relationships between infrared absorption, Rayleigh scattering and Raman scattering; Dotted line indicates a "virtual state" (Adapted from Li-Chan, 1996)  38  Figure 2.1. The standard linear graph of orthogonal array L 7 ( 3 ) used in this study  64  Figure 2.2. Total ion chromatograms (TICs) of headspace volatile compounds in simulated beef flavour using the S P M E adsorption conditions described in Table 2.2 for experiment number 25, 26, and 27. 'Experiment numbers correspond to those of Table 2.2  65  13  2  Figure 2.3. Effects o f adsorption temperature, adsorption time, salt concentration, and S P M E phase on (a) means of total area counts and (b) means of the number o f peaks obtained in the analysis of headspace volatile compounds in simulated beef flavour. Different letters (a and b) within each plot indicate significant difference (p < 0.05). 'Levels 0, 1, and 2 o f the S P M E phase refer to 50/30 pm D V B / C A R / P D M S , 65 pm P D M S / D V B , and 65 pm C W / D V B , respectively. different at 10 % significance level (p = 0.0571) 66  Figure 2.4. Effect of adsorption time on the headspace volatile analysis o f simulated beef flavour by 50/30 pm D V B / C A R / P D M S S P M E at 60 °C. Points are averages from triplicate analyses and error bars are ± standard deviation. 'Different letters indicate significant difference (p < 0.05)  67  Figure 3.1. (a) G C - F I D chromatogram of simulated beef flavour and (b) aromagram of volatile compounds o f simulated beef flavour expressed as retention time and detection frequency by G C - 0 (number of panelists = 8). Peak numbers correspond to the peak numbers in Table 3.1  90  Figure 4.1. Cobweb diagram o f the sensory scores from the descriptive analysis data for (a) 6 unheated mixtures o f beef flavour and soy protein isolate (b) 6 heated (98 °C, 30 minutes) mixtures (n=12; 6 selected panelists with 2 replications). L - S B F and H - S B F represent low (150 mg) and high (500 mg) dose o f simulated beef flavour while N - S P I , M - S P I , and H-SPI symbolize no (0 mg), medium (250 mg), and high (500 mg) amount of soy protein isolate in the sample. Refer to Table 4.3 for the sample codes  118  Figure 4.2(a). P C loadings and scores of the sensory attributes and the sample mixtures by principal component analysis; P C I versus P C 2 . L - S B F and H - S B F represent low (150 mg) and high (500 mg) dose o f simulated beef flavour while N - S P I , M - S P I , and H-SPI symbolize no (0 mg), medium (250 mg), and high (500 mg) amount o f soy protein isolate in the sample. Refer to Table 4.3 for the sample codes  119  Figure 4.2(b). P C loadings and scores of the sensory attributes and the sample mixtures by principal component analysis; P C I versus P C 3 . L - S B F and H - S B F represent low (150 mg) and high (500 mg) dose o f simulated beef flavour while N-SPI, M-SPI, and H-SPI symbolize no (0 mg), medium (250 mg), and high (500 mg) amount of soy protein isolate in the sample. Refer to Table 4.3 for the sample codes  120  Figure 5.1. Differential scanning calorimetric (DSC) thermograms o f SPI (0.2 %, 5 % , and 10 % in p H 7.5 phosphate buffer)  157  Figure 5.2. FT-Raman spectrum (400-1700 cm' ) o f SPI powder obtained by freeze-drying SPI solution (5 % w/w in 0.05 M Tris-HCl buffer p H 7.4) 158 1  Figure 5.3. FT-Raman spectra o f SPI and SPI treated with various ingredients (baseline corrected, ingredient spectrum subtracted, and normalized to the intensity of phenylalanine peak at 1003 cm"'). G : glucosamine; S : sucrose; P : polyethylene glycol; A : ascorbic acid; A P : ascorbic acid with polyethylene glycol 159  Figure 5.4. Cobweb diagram o f the sensory scores from the descriptive analysis of simulated beef flavour in the presence of soy protein isolate (SPIF), SPI containing ascorbic acid ( A F ) and SPI containing ascorbic acid with polyethylene glycol (APF) (n=14; 7 panelists with 2 replications)  160  Figure 5.5. P C loadings and scores of the sensory attributes and the sample mixtures by principal component analysis; P C I versus P C 2 . L - S B F and H - S B F represent low (150 mg) and high (500 mg) dose o f simulated beef flavour while N-SPI, M - S P I , and H-SPI symbolize no (0 mg), medium (250 mg), and high (500 mg) amount of soy protein isolate in the sample. Refer to Table 4.3 and Figure 5.4 for the sample codes  161  List of Abbreviations  A  SPI containing ascorbic acid  AEDA  Aroma Extract Dilution Analysis  AF  SBF with SPI containing ascorbic acid  ANOVA  Analysis of Variance  ANS  l-Anilinonaphthalene-8-sulfonic acid  AP  SPI containing ascorbic acid and polyethylene glycol  APF  SBF with SPI containing ascorbic acid and polyethylene glycol  CAR/PDMS  Carboxen/polydimethylsiloxane  CHARM  Combined Hedonic Aroma Response Measurements  CPA  cis-Parinaric acid  CW/DVB  Carbowax/divinylbenzene  DA  Descriptive Analysis  DF  Detection Frequency  df  Degree of Freedom  DH  Degree of Hydrolysis  DSC  Differential Scanning Calorimetry  DTNB  5,5'-Dithio-bis-2-nitrobenzoic acid  DVB/CAR/PDMS  Divinylbenzene/carboxen/polydimethylsiloxane  FD  Flavour Dilution  FID  Flame Ionization Detector  FT-Raman  Fourier-Transform Raman  G  SPI containing glucosamine  g  Grams  g  Gravitational Force  GC  Gas Chromatography  GC-MS  Gas Chromatography-Mass Spectrometry  GC-O  Gas Chromatography-Olfactometry  GF  SBF with SPI containing glucosamine  HS-SPME  Headspace-Solid Phase Microextraction  IP  Indicator Peak  kDa  K i l o Dalton  LSD  Least Significant Difference  MSD  Mass Selective Detector  MSE  Mean Squares o f Error  NIR  Near-Infrared  P  SPI containing polyethylene glycol  PC  Principal Component  PCA  Principal Component Analysis  PF  S B F with SPI containing polyethylene glycol  PRODAN  6-Propionyl-2(N,N-dimethyl-amino)naphthalene  RFI  Relative Fluorescence Intensity  RSAP  Raman Spectral Analysis Package  S  SPI containing sucrose  SBF  Simulated Beef Flavour  SF  S B F with SPI containing sucrose  SPI  Soy Protein Isolate  SPIF  S B F with SPI  SPME  Solid Phase Microextraction  TIC  Total Ion Chromatogram  xiv  Acknowledgements M y deepest and most sincere gratitude goes to my supervisor, D r . Eunice Li-Chan, for her outstanding guidance and thoughtful encouragement to complete this thesis. She is not only a knowledgeable academic supervisor but also a mentor in my mind who is devoted and caring with warm heart. It is a blessed pleasure to be her student. I also sincerely appreciate my supervisory committee members, Dr. M . Cliff, Dr. C . Seaman and M s . W. Willis, for their invaluable advice and suggestions to my studies. Their wisdom offered great help to proceed to each step. I appreciate their continuous support throughout the studies. I would also like to express my thanks to Dr. T. Durance, Dr. D . Kitts, Dr. S. Nakai and Dr. B . Skura for their academic advice and comments during classes and work. M y special thanks are extended to M s . T. Nguyen, Dr. A . Pedro, M s . V . Skura, Dr. P. Yaghmaee, and M r . S. Yee in Food Science at U B C , Dr. T. Cottrell at the Pacific Agri-Food Research Station of Agriculture and Agri-Food Canada, M s . L . Madilao in the Wine Research Centre at U B C as well as M r . K . Cummings, M s . A . McCannel, and Dr. G . Sandberg for their technical support and considerate help. I would also like to thank all the graduate students and fellows in Food Science at U B C especially members in Dr. Li-Chan's group, Andrea, Anusha, B o , Crystal, Imelda, Judy, Tarn, Tom, and Wendy as well as Dr. Alizadeh-Pasdar, Dr. K i m and Dr. Meng for showing faithful friendship and sharing their experience with me. I will also remember friendship o f Sung-Yon, Soon-Hye, Kyu-Won, Ji-Eun, Ro-Mee and other friends in Korea. This research was supported by research funding and postgraduate scholarships from the Natural Sciences  and  Engineering Research  Council  of Canada  (NSERC),  Graduate  Research  Engineering and Technology ( G R E A T ) scholarship granted by Science Council of British Columbia, and the University Graduate Fellowship provided by the University o f the British Columbia. Last but not least, my parents, parents-in-law, brother and sister in Korea, who have shown endless support, love, and prayers to me, receive my deepest gratitude. They always have been protective background surrounding me. XV  Dedication  This thesis is dedicated to:  M y husband, Kwangsoo Daughter, Jennifer (Yewon) Son, Marie (Jongwon)  Your understanding and encouragement based on sacrifice made this dissertation possible.  xvi  Co-Authorship Statements  The work presented in Chapter 2 of this thesis was published in "Food Chemistry" (2004), 88, 141-149, entitled "Development o f solid-phase microextraction methodology for analysis of headspace volatile compounds in simulated beef flavour". Soo Yeun Moon, the thesis author, was the principal author and Eunice C . Y . Li-Chan, Soo Yeun Moon's supervisor, was the co-author.  The work presented in Chapter 3 o f this thesis was published in "Food Research InternationaF (2006), 39, 294-308, entitled "Odour-active components o f simulated beef flavour analysed by solid phase microextraction and gas chromatography-mass spectrometry and -olfactometry". Soo Yeun Moon, the thesis author, was the principal author in the publication. Margaret A . Cliff, member of Soo Yeun Moon's supervisory committee, and Eunice C . Y . Li-Chan, Soo Yeun Moon's supervisor, were the co-authors.  The work presented in Chapter 4 of this thesis w i l l be submitted for publication in the "The Journal of the American Oil Chemists' Society", entitled "Changes in aroma characteristics of simulated beef flavour by adding soy protein isolate as assessed by descriptive sensory analysis and gas chromatography-olfactometry". Soo Yeun Moon, the thesis author, was the principal author and Eunice C . Y . Li-Chan, Soo Yeun Moon's supervisor, was the co-author.  The work presented in Chapter 5 of this thesis will be submitted for publication in the "The Journal of the American Oil Chemists' Society", entitled "Assessment o f added ingredient effect on the interaction of simulated beef flavour and soy protein isolate by gas chromatography and spectroscopic techniques". Soo Yeun Moon, the thesis author, was the principal author and Eunice C . Y . Li-Chan, Soo Yeun Moon's supervisor, was the co-author.  Soo Yeun Moon  Eunice C. Y . Li-Chan  xvii  CHAPTER 1. OVERVIEW AND LITERATURE REVIEW  1.1 General Introduction  For centuries, soybean has been one o f the major sources of edible plant oil, while the remaining defatted soy flakes provide nutritional and economical advantages to food with its high protein content and relatively low cost. The little round bean has a long history with the first reference dating back to 2838 B C in a book "Pen-Ts'ae-Kung-Mu" written by Shen Nung (Hoogenkamp, 2006). It has been used as the starting material for soy flour, soymilk and tofu and various fermented foods such as soy sauce, miso, natto, and tempeh, which are part of the staple diet in oriental countries (Smith and Circle, 1978).  Apart from being an important source of protein in  food, soy proteins have the capacity to be functional ingredients in foods with the practical functionalities such as gelation, emulsification, stabilization, water adsorption, fibre formation, film formation, elasticity, and foaming properties (Ohren, 1981). Although soybeans have been part of the staple diet in many Asian countries for a long period, it has only been commercially grown in the North America since 1922 as an inexpensive source o f edible o i l . Significant production started since the 1960s but still a limited quantity o f soybeans is consumed directly in North America and Europe (Hoogenkamp, 2006). In recent years, hypocholesterolemic effects of soy protein were reported and much evidence has been gathered demonstrating that consumption of soy protein helps to reduce risks of cardiovascular disease (Anderson et al., 1995). In October 1999, the Food and Drug Administration approved health claims for soy protein and coronary heart disease ( U S F D A , 1999), followed by a similar health claim by the Joint Health Claim Initiative in 2002 indicating "the inclusion of at least 20 grams of soy protein per day, as part of a diet low in saturated fat, can help reduce blood cholesterol levels" (Hoogenkamp, 2006). In addition, soy protein has become popular as meat substitutes with increasing number of vegetarians and health-conscious consumers wishing to reduce their intake of meat.  However, in spite o f the high nutritional quality and great functionalities o f soy proteins, flavourassociated problems have been a major practical hindrance to expand the usage o f soy proteins in food products and consumer acceptance of soy products (MacLeod and Ames, 1988; Maheshwari et al., 1995; Rah et al., 2004; Schutte and Van den Ouweland, 1979). In addition to the 1  indigenous undesirable soy aroma components that are problematic to eradicate, interaction of flavour compounds with soy proteins has been reported (Damodaran and Kinsella, 1981a and 1981b; Gremli, 1974; Malcolmson and McDaniel, 1987). Considerable research has been performed to understand the flavour-binding nature of soy protein (Aspelund and Wilson, 1983; Beyeler and Solms, 1974; Damodaran and Kinsella, 1981a and 1981b; O'Keefe et al., 1991a and 1991b; L i et al., 2000; Zhou and Cadwallader, 2004). However, most o f these studies used model systems o f single ingredient or selective volatile model compounds associated with flavour or off-flavour, such as a series o f aldehydes, ketones, alcohols, or alkanes. Even though valuable thermodynamic information was obtained from these studies, the knowledge may not be directly applied to the real food system, in which the flavour ingredients usually include combinations o f a broad array o f subclasses of compounds. Soy proteins bind with certain desirable flavour compounds, which could have an impact on flavour suppression or alteration of flavour profiles in the mixture or final food products. Therefore, it is imperative to elucidate the nature o f the interactions of soy proteins with flavour compounds for the development o f soy products with acceptable flavour quality.  In this thesis, interaction between soy protein isolate and simulated beef flavour used as commercially available ingredients in the food industry was investigated to elucidate integral flavour holding properties o f the soy protein isolate and changes in sensory characteristics o f the mixture.  1.2. Soybean Protein Isolate (SPI)  1.2.1. Specification and production Soy protein isolate (SPI), also known as soybean protein isolate or isolated soybean protein, is the most concentrated form o f commercially available soybean protein products. According to the specification defined by the Association of American Feed Control Officials, Inc. ( A A F C O ) , "soy protein isolate is the major proteinaceous fraction of soybeans prepared from dehulled soybeans by removing the majority o f non-protein components and must contain not less than 90 % protein on a moisture-free basis". Soy protein concentrate, another commercially available soy protein product, "is prepared from high quality sound, clean, dehulled soybean seeds by 2  removing most of the oil and water soluble non-protein constituents and must contain not less than 70 % protein on a moisture free basis" (Berk, 1992). The average chemical composition of SPI, compared to defatted soy flour and soy protein concentrate, is presented in Table 1.1.  To produce soy-related products, whole beans are soaked in water until they contain about 11 % moisture. After removal of the hulls, dehulled beans are milled into flakes or more fine particles to produce full-fat soy flour, which consists of 43 % protein, 23 % lipids, 29 % carbohydrate, and 5 % ash (MacLeod and Ames, 1988). To produce defatted soy flour, coarse flakes are extracted with a solvent such as hexane, and then deodourized by treatment at about 85 °C before milling to produce a fine powder. SPI is prepared by extracting defatted flakes or flours with dilute alkali solution (pH 7-10) at 50 - 55 °C, followed by centrifugation to eliminate the insoluble polysaccharide residue, as shown in Figure 1.1. After the alkali extract is clarified, it is acidified to p H 4.5, which is the isoelectric point of the major soy proteins, resulting in precipitation of the soy protein. The precipitate is neutralized to a p H about 6.8 and then spray-dried, leading to a highly soluble proteinate form of SPI.  Alternatively, the precipitate may be centrifuged, washed,  and made into a slurry with water and then spray-dried to produce the isoelectric form of SPI (Hettiarachchy and Kalapathy, 1997).  1.2.2. M a j o r proteins in S P I Most of the proteins in soybean are categorized as globulins, which are soluble in salt solution. According to their sedimentation properties, they can be classified into four fractions namely 2S, 7S, U S , and 15S. The two major globulin components in SPI are glycinin ( U S globulin) and Pconglycinin (7S globulin), which account for about 80 % o f the total storage proteins in soybean (Moriyama et al., 2005). The average calculated molecular weight o f SPI is 237 k D a based on 15 %, 34 %, 42 %, and 9 % of 2S (18-33 kDa), 7S (180-210 kDa), U S (300-350 kDa), and 15S (600 kDa) in SPI, respectively (Li et al., 2000).  1.2.2.1. G l y c i n i n Glycinin is the purified form o f the 11S fraction, which accounts for over 40 % of the total seed globulin in soybeans ( L i u , 1997). A t ambient temperature and p H 7.6, glycinin is believed to exist in the form o f hexameric complexes (1 IS) with a M W of about 360 kDa; each monomeric 3  subunit is composed o f an acidic polypeptide (ca. 34-44 kDa) and a basic polypeptide (ca. 20 kDa) with one single disulfide bond between the two polypeptides, which can be represented as A - S - S - B or A - B (Liu, 1997). Six monomeric subunits form a glycinin molecule ( A - S - S - B ) , with 6  two hexagonal rings stacked one on top of the other (Yamauchi et al., 1991). However, a trimeric complex (7S) form was also found at p H 3.8 and a 15S fraction is thought to be a polymer form of the P-conglycinin (Renkema et al., 2002). The secondary structure o f glycinin was reported by Abbott et al. (1996) to be 24 % a-helix, 30 % |3-sheet, 42 % turns and 12 % unordered. Glycinin has a complex quaternary structure, which depends on p H and ionic strength (Renkema et al., 2002).  The bonds responsible for the quaternary structure can be disrupted by urea, strong acid,  strong base, heat, or sodium dodecylsulfate in combination with a reducing agent, leading to dissociation into the subunits and further into the acidic and basic polypeptides, and resulting in altered structure o f glycinin as shown in Figure 1.2.  Five major subunits G1-G5 o f glycinin that have been purified from the soybean cultivar C X 6 3 5 1-1-1 are presented in Table 1.2. Those subunits can be divided into two groups (I and II) according to their physical and chemical properties. Compared to group II subunits, group I subunits have more uniform M W (58 kDa), contain more methionine, and exhibit about 90 % sequence homology among members, while the group II subunits show 60-70 % sequence homology between members.  Among the five major subunits, the G5 subunit ( A A B ) is most 5  4  3  unique in its general structure, consisting of acidic components composed of two different polypeptides ( A A ) . Except for A 4 in G subunit, all acidic and basic polypeptide are linked via 5  4  disulfide bonds, while polypeptide A  5  4  is not covalently linked with a basic polypeptide.  Interestingly, the G5 subunit was not found in Raiden cultivar; instead a subunit with a different acidic polypeptide A6 was discovered (Liu, 1997).  1.2.2.2. p-Conglycinin |3-Conglycinin, the other major constituent  in SPI, is a glycoprotein containing 4-5 %  carbohydrate and is generally called 7S globulin. It is a trimer with a M W o f about 180 k D a comprised of three subunits, a', a, and p, with M W of 57-72 kDa, 57-68 kDa, and 42-52 kDa, respectively (Yamauchi et al., 1991). However, another subunit, called P', was also reported in some soybean varieties (Morita et al., 1996). Unlike the polypeptides in glycinin, the subunits in 4  P-conglycinin are non-covalently associated through hydrophobic and hydrogen bonding without any disulfide bonds. A s shown in Table 1.3, all three major subunits are abundant in aspartate, asparagine, glutamate, glutamine, leucine and arginine, and the amino acid composition is especially similar between the a and a' subunits (Liu, 1997). Results from recent research showed important  physiological functions  of  P-conglycinin such  as  hypocholesterolemic  and  hypotriglycemic activities in rats or suppression of serum triglyceride, insulin and glucose levels in normal and genetically obese mice (Moriyama et al. 2005).  1.2.3. Soy protein isolates used for functional ingredients SPIs have been applied extensively in meat products as extenders and meat analogs, bakery products, and dairy products. The use of soy proteins as a food ingredient arises from their functional properties  such as solubility, viscosity, gelation, emulsification, and foaming  properties, as shown in Table 1.4, which are based on structural and conformational attributes o f the proteins. Since non-covalent forces such as ionic or electrostatic interaction, hydrogen bonding and hydrophobic interaction of amino acid side chains as well as covalent disulfide bonds are responsible for formation and stability of the protein conformation, any factor that changes these interactions could affect the secondary and tertiary structure along with the exposure of amino acid side chains to the surface. In turn, these structural changes could lead to alteration in the functional properties of the protein.  Differences in amino acid composition between P-conglycinin and glycinin are shown in Table 1.3, including the higher contents of cysteine and methionine in glycinin than p-conglycinin. Aside from the compositional difference, the structural difference between the two proteins also results in considerable variation in functional properties o f the proteins (Yamauchi et al., 1991). While glycinin was reported to have superior gel formation property, P-conglycinin had greater emulsifying power and emulsion stability.  Both P-conglycinin and glycinin formed gels by heat  and/or a coagulant during tofu manufacture but glycinin needed higher heating temperature to form a gel than p-conglycinin due to different heat denaturation characteristics (Hettiarachchy and Kalapathy, 1997). Utsumi and Kinsella (1985) stated that hydrogen bonding and disulfide bonds were important in maintaining gel network structures. It was also reported that disulfide cleavage weakened the gel strength at low protein concentration while it helped to reinforce 5  gelation at high protein concentration (Hettiarachchy and Kalapathy, 1997).  1.2.4. Structural changes in soy proteins Protein denaturation is "any modification in conformation (secondary, tertiary, or quaternary structure) not accompanied by the rupture o f peptide bonds involved in primary structure" (Cheftel et al., 1985). The conformation o f a protein involving its secondary and tertiary structure is susceptible to be changed by physical causes such as heat, pressure, irradiation, or mechanical stress, and by chemical agents such as acids, alkalis, salts, organic solvents, surfactants, or reducing agents. Among those factors, heat-induced structural changes in soy proteins have been extensively researched as heating is one o f the most frequently applied methods during processing.  Heating glycinin with p-conglycinin resulted in complex formation between dissociated Pconglycinin subunits and glycinin subunits (German et al., 1982). Thermal aggregation of glycinin in the presence o f a reducing agent such as 10 m M 2-mercaptoethanol at 80 °C due to aggregation of the basic subunits was observed by Damodaran and Kinsella (1982). However, the thermal aggregation o f glycinin was prevented by addition o f isolated conglycinin, and formation of a soluble complex between the subunits of p-conglycinin and the basic subunits o f glycinin was suggested (Damodaran and Kinsella, 1982). Involvement o f breakage and formation of disulfide bonds between basic subunits o f glycinin as a result o f heating treatment was also reported by Utsumi et al. (1984). In addition, dissociation o f both P-conglycinin and glycinin by heat treatment at 80 °C was detected, followed by subsequent interaction between the Pconglycinin and glycinin globulins; in particular, the basic subunits o f glycinin and the P subunit of P-conglycinin were predominantly found in the aggregates.  The heat-induced subunit interaction between P-conglycinin and glycinin was schematically represented by Yamauchi and coworker (1991) as shown in Figure 1.3. Relatively mild heating triggered dissociation o f both glycinin and P-conglycinin and further heating caused subsequent interaction with each other, resulting in the formation o f soluble polymerized polypeptides from dissociated glycinin and P-conglycinin subunits. Although the occurrence o f precipitation depends on the concentration o f the proteins, the basic and P subunits tend to be located in the 6  precipitate while the acidic and a , a ' subunits are likely to be found in the supernatant through disulfide bonding among the acidic, a , and a ' subunits.  Plant proteins including soy proteins contain high contents o f asparagine and glutamine residues, and the side chain amide groups are known to be important to stabilize the protein structure through hydrogen bonding, which are easily modified by relatively mild treatment (Matsudomi et al., 1985b). Changes in conformation and functional properties of soy protein by mild acid treatment (2 % soy protein solution in 0.05 N HC1) were reported (Matsudomi et al., 1985b). In the dilute acid condition, preferential deamidation without significant cleavage of the peptide bonds resulted, leading to an increase in functional properties such as solubility, emulsifying properties and foaming properties as well as increased surface hydrophobicity which was considered to be associated with conformational changes due to deamidation and acid -induced denaturation.  Although most soy protein studies have been conducted at neutral p H , it was reported that Pconglycinin exhibited association-dissociation behavior depending on the p H and ionic strength of the solution, p-conglycinin (7S) was found as a trimer structure at p H 7.6 at high ionic strength (I > 0.5) or acidic p H (pH < 4.8), while it existed as a hexamer (10S) at low ionic strength (I < 0.2) in the p H region 4.8-11.0 (Thanh and Shibasaki, 1979). Dissociation of glycinin hexamers to their constituent polypeptides has also been reported. Glycinin is mainly present in a hexameric form (1 IS) but with lowering of the ionic strength to 0.01 at p H 7.6, it was dissociated from the U S form mainly into the 7S form, which was believed to be the trimeric form and less structured conformation (Utsumi et al., 1987). Lakemond et al. (2000) also confirmed that glycinin formed hexameric complexes (1 IS) at p H 7.6 and an ionic strength o f 0.5 while it existed as trimers (7S) at p H 3.8 and at an ionic strength of 0.03. They also examined tryptophan fluorescence spectra using fluorescence spectroscopy to determine differences in the tertiary interactions within glycinin and reported that no significant structural changes were found among different ionic strengths in the range of 0.03-0.5 at p H 7.6, but a more tightly packed fluorophore environment was observed when the p H is lowered from 7.6 to 3.8  Simultaneous heat and reducing treatment may also affect structural characteristic of SPI. 7  Remondetto et al. (2002) reported that sulfite at low temperature promoted denaturation by reducing disulfide bonds. Cleavage of disulfide bonds by sulfite at 67-75 °C caused increase in molecular flexibility o f SPI leading to formation o f new disulfide bonds. Enhanced functional properties such as solubility, gelation, foaming and emulsification by increasing molecular flexibility due to partial reduction o f disulfide bonds were also reported (Petruccelli and Anon, 1995).  1.2.5. Modification Food proteins including SPI generally need some modification o f their composition and structure to be used as food ingredients with suitable functional properties. Although various chemical modifications such as acylation, alkylation, esterification, phosphorylation, glycosylation, deamidation and reduction o f disulfide bonds have been used to improve functionalities o f food proteins, acylation seemed to be most widely researched. For instance, acylation increased the solubility of soy protein in acid solutions for use in coffee whiteners (Meyer and Williams, 1977). Acylation with either acetic anhydride or succinic anhydride significantly decreased the texture of SPI extrudate but increased its solubility (Simonsky and Stanley, 1982). SPI treated with increasing concentrations o f Na S03, which could cause cleavage o f disulfide bond, N a C l , or 2  Na2SC>4 showed decreased viscosity and adhesive strength (Kalapathy et al., 1996).  1.3. Flavour of soy products  1.3.1. Volatile compounds in soybean and soy products Soybean foods such as soy flour, soymilk, and textured soy protein have been reported to have a typical odour, which is often described as green (raw/fresh), grassy, and beany (Cowan et al., 1973). In many cases, these beany odours detected in soy products are regarded as undesirable and even offensive, resulting in unpleasant soy products. The attributes could be detected by 80 % o f the sensory panel even when soy flour was diluted with non-odorous wheat flour at 1:750 in the study o f Moser et al. (1967). The off-flavour associated with soybean is a major barrier in the increased usage of soy proteins in human foods.  There have been several reports regarding the effect of heat treatment on the sensory properties of 8  soybean odour. The grassy/beany notes were decreased during the heating o f soy flours such as texturization while cereal /grain-like and toasted attributes were retained or developed (Warner et al., 1983). Kato et al. (1981) also reported that the odour note o f soybeans changed remarkably when dry heat was applied to soy flour at 200 °C with considerable weight loss due to decomposition of nonvolatile precursors and water evaporation. The subsequently roasted soy flour showed a pleasant odour with toasted note, which could mask other off-odours.  However,  when defatted soy proteins were retorted or sterilized, an unpleasant cooked odour was produced (Greuell, 1974).  The volatile compounds identified from soybeans, flours, soy protein concentrates, SPIs, or textured soy proteins were described in the comprehensive review by MacLeod et al. (1988) where a total number o f 334 individual components were listed comprising 18 aliphatic hydrocarbons, 3 alicyclic hydrocarbons, 14 terpenoids, 31 aliphatic alcohols, 31 aliphatic aldehydes, 32 aliphatic ketones, 4 alicyclic ketones, 10 aliphatic carboxylic acids, 10 lactones, 13 aliphatic esters, 1 alicyclic ester, 6 aliphatic ethers, 7 aliphatic amines, 1 aliphatic nitrile, 5 chlorine-containing compounds, 67 benzenoids, 12 aliphatic sulfur compounds, 24 furanoids, 12 thiophenoids, 5 pyrroles, 1 pyridine, 19 pyrazines, 3 thiazoles, and 5 other sulfur heterocyclic compounds.  Recently, there have been several reports that were conducted using gas chromatographyolfactometry (GC-O) and gas chromatography-mass spectrometry ( G C - M S ) to determine the major volatile compounds that contribute to the beany odour.  The most potent odourants in dry  commercial SPIs determined by G C - 0 were butyric acid, 2-methyl butyric acid methyl ester, 2pentyl pyridine and hexanal (Boatright et al., 1997), while dimethyl trisulfide, trans, trans-2,4decadienal, 2-pentyl pyridine, trans,trans-2,4-non&diena\, hexanal, acetophenone and l-octen-3one were identified in aqueous solution of SPIs (Boatright et al., 1999).  L e i et al. (2001)  repotted that acetaldehyde, methanethiol, hexanal, dimethyl trisulfide, and 2-pentyl furan were detected by G C - 0 and G C - M S in the headspace o f the aqueous slurries o f soy protein concentrate, and methionine was reported as the methyl group donor for formation o f sulfite-associated methanethiol in aqueous slurries of SPI (Lei and Boatright, 2006).  9  The beany odour compounds in soy products have been reported to include aliphatic carbonyls, volatile fatty acids, alcohols, furans and amines.  Hsieh et al. (1981) reported that the major  volatile compounds of soy products were 1-hexanol, 1-pentanol, hexanal, l-octen-3-ol, and 2pentylfuran.  Medium-chain aldehydes such as pentanal, hexanal, and heptanal were also  reported to be the key class of compounds contributing to beany odours of soy proteins (Maheshwari et al., 1995). These volatile components could arise from the beans themselves by the action of soybean lipoxygenase and subsequent formation of lipid oxidation products, which are of critical importance in soybean off-flavour (Sessa et al., 1977).  Oleic (18:1), linoleic  (18:1), and linolenic (18:2) acids are the most important precursors since soy lipids are characterized by a relatively high content of unsaturated fatty acids, especially linoleic and oleic acids, which comprise 53.2 % and 23.4 %, respectively of the fatty acids in soybean oil. Other possible reactions include the result of heat on sugars and/or amino acids, thermal decomposition of phenolic acids and thiamine, and the oxidative and thermal degradation of carotenoids (MacLeod et al., 1988).  Even though it is not easily recognized due to the complexity and  variety of the reactions, secondary reactions can be induced from the products of primary reactions.  Moreover, the grassy and objectionable odour could also be developed as a result of  processing and during storage of soybeans and flour (Warner et al., 1983).  MacLeod et al. (1988) drew several conclusions in terms of the volatile compounds of commercially produced materials such as full-fat flakes/flours, defatted flakes/flours, toasted defatted flakes/flours, concentrates, isolates, and textured soy protein (TSP).  Firstly, the  commercially produced full-fat flakes/flours or soy concentrate did not show additional volatile components compared to those reported from ground beans and heated ground beans.  Secondly,  significantly fewer alcohols, aldehydes, and benzenoids were found in defatted flakes/flours. Thirdly, toasted defatted flakes/flours contained a great number of pyrazines. However, those had already been identified in either heated ground beans or defatted flours.  Lastly, considerably  fewer benzenoids and aliphatic alcohols were found in soy isolate, while a large number of aliphatic sulfur compounds, furans, and thiophens were recognized.  10  1.3.2. Interaction of soy protein with flavour compounds 1.3.2.1. Typical flavour binding model The composition and concentration o f volatile compounds in the headspace above food rather than in the food itself more directly affects the aroma profile of the food. Research to investigate binding behavior o f proteins to flavour compounds has been characterized using gel filtration, equilibrium dialysis, and headspace method, and interpreted most widely by the Scatchard equation (Scatchard, 1949) and Klotz plot (Klotz, 1946) as described by O ' N e i l l (1996).  Interaction of a protein (P) with a ligand (L; flavour molecule) may be represented as: P + L = PL  [equation 1]  As P(total) = P L + P, the equation 1 can be expressed as below with the association constant, K (PL) = K(P)(L)  [equation 2]  (PL) = K(L)[P(total)-(PL)]  [equation 3]  The equation 3 can be converted using v, the number o f moles o f ligand bound per mole o f protein, as below. v = (PL)/P(total)  [equation 4]  v = K(L)/[1+K(L)]  [equation 5]  The equation 5 can be extended with a protein having indistinguishable and independent binding sites (ri) v = «K(L)/[ 1 +K(L)]  [equation 6]  y/L = «K - Kw  [equation 7]  Therefore, plotting v/L versus v gives the Scatchard plot with the slope, - K , and y-intercept, nK. The binding equation also can be transformed into the double reciprocal equation (Klotz plot) to linearize the binding data: 1  1  oK(L)  +  1  n  [equation 8]  Experimental determination o f the value o f the K as a function o f temperature allows determination of thermodynamic parameters such as the Gibb's free energy o f binding (AG ), the 0  enthalpy of binding (AH°), and the entropy of binding (AS°). A G = -RTlnK 0  [equation 9] 11  AH° = - R d l n K / J ( l / T )  [equation 10]  A S = (AH° - AG°)/T  [equation 11]  0  where T is the absolute temperature in °Kelvin (°K), and R is the gas constant.  1.3.2.2. B i n d i n g of flavour model compounds to soy protein A number o f studies have examined the interaction between proteins and flavour compounds in aqueous model systems. Most o f the research was conducted using simple model compounds or a series of alcohols, aldehydes, ketones or carboxylic acid, which are recognized as off-flavour related volatile compounds in soybean.  Binding of n-hexanal and n-hexanol to native, partially denatured, denatured, and enzymatically hydrolyzed soy protein was compared using vacuum distillation and gel filtration techniques to investigate the effectiveness of enzymatic proteolysis in removing off-flavours from soy protein concentrate (Arai et al., 1970). In the study, binding constants of n-hexanal and n-hexanol for native soy protein were determined to be 173.4 and 80.3 M " , respectively (Table 1.5), and the 1  interactions increased with the degree of denaturation and decreased with protein hydrolysis.  Gremli (1974) studied the effects of adding flavour compounds to SPI by headspace analysis and a high vacuum transfer system. In contrast to the results of Arai et al. (1970), aldehydes, especially unsaturated compounds, reacted more strongly with the proteins than ketones while alcohols did not interact with soy protein. The interactions showed both reversible and, of less importance, irreversible features and the retention of aldehydes by 5 % soy protein solution was positively correlated to the chain length of the aldehydes ranging from hexanal (C6) to dodecanal (C12) (Gremli, 1974).  The binding constant o f SPI investigated with an equilibrium dialysis method was reported by Beyeler and Solms (1974) to decrease in the order of aldehydes, ketones, and alcohols (Table 1.5). N o binding affinity was found with carboxylic acids, dimethylpyrazine, aniline or phenylalanine. The binding effects o f SPI were found to be rather independent of p H and temperature, and characterized by weak and unspecific binding forces.  12  Damodaran and Kinsella (1981a) investigated the interaction of carbonyl compounds with soy protein using an equilibrium dialysis method. Binding affinities of ketones (C7 to C9) to soy protein increased with increasing chain length. The binding constant increased by nearly 3 times with the change of about -600 cal/CFb residue for each additional methylene group on the ligand (Table 1.5).  The binding was suggested to be spontaneous and thermodynamically favorable  due to the negative value of AG.  In terms of the position of the keto group among nonanal, 2-  nonanone, and 5-nonanone, shift of keto group toward the middle of the chain was observed to be associated with a decrease in binding affinity (Table 1.5). Steric hindrance of the relatively polar keto group to the hydrophobic interaction of ligand to the binding site of the protein was suggested to explain the result.  Partial denaturation of soy protein induced by heat treatment  increased the binding affinity for 2-nonanone compared to native soy protein (Table 1.5). The change in quaternary structure of soy protein was proposed to occur through certain reorganization of the subunits, which may affect the hydrophobicity of the sites resulting in increasing of binding affinity for the ligand (Damodaran and Kinsella, 1981a)  Different binding affinities were observed in glycinin and p-conglycinin in isolated systems. Damodaran and Kinsella (1981b) reported that the binding constant and the total number of binding sites of P-conglycinin for 2-nonanone were about the same as those of whole soy protein whereas almost no affinity was found in glycinin indicating P-conglycinin might be the responsible component for off-flavour binding in soy protein. The difference was assumed to be due to the different spatial arrangement of the subunits in the two proteins i.e. possible hydrophobic regions accessible for ligand binding in P-conglycinin versus not available or buried hydrophobic regions inside the glycinin. The authors also found an increase in the binding affinity of glycinin when ionic strength was increased from 0.03 M to 0.5 M , and suggested that the lack of affinity of glycinin for 2-nonanone was due to the dissociated form of glycinin isolated in low ionic strength. Both the binding affinity and the binding capacity of soy protein for 2-nonanone were considerably affected by urea or chemical modification with succinic anhydride (Damodaran and Kinsella, 1981b). Decreased fluorescence intensity of tryptophan residues with increasing urea concentration or succinylation treatment was explained by destabilization of the hydrophobic regions in soy protein. However, O'Neill and Kinsella (1987) reported binding constants (K) of whole soy protein, P-conglycinin, and glycinin for 2-nonanone 13  of 570, 3050, and 540 M " , respectively, and approximately 5, 2, and 3 primary binding sites per 1  100,000 daltons o f whole soy protein, P-conglycinin, and glycinin, respectively, demonstrating that P-conglycinin had greater affinity for 2-nonanone than soy protein as well as glycinin. The discrepancy was partly explained by differences in the composition of the soy proteins used and different absorption coefficient applied in estimation of the P-conglycinin concentration.  The thermodynamics of binding for soybean glycinin and P-conglycinin with flavour ligands such as butanal, pentanal, hexanal, octanal, 2- and 3-hexanone, 2- and 5-nonanone, hexanol, and hexane was studied using a headspace technique by O'Keefe et al. (1991a).  The number of  binding sites and binding constants were greater for glycinin than P-conglycinin for all flavour ligands (including 2-nonanone) at all three temperatures, namely 5 °C, 20 °C, and 30 °C, in disagreement with the results from Damodaran and Kinsella (1981b) and O ' N e i l l and Kinsella (1987).  O'Keefe et al. (1991a) also reported that affinity for aldehydes increased with  increasing chain length for glycinin, whereas it remained constant for P-conglycinin.  Binding affinities of hexanal to soy glycinin and p-conglycinin under different conditions were evaluated by O'Keefe et al. (1991b). In this study, in contrast to the results reported by O'Keefe et al. (1991a), similar binding constants of 270 and 303 M " were observed for glycinin and P1  conglycinin, respectively, and the number of binding sites o f hexanal to soy glycinin and Pconglycinin were 108 and 26, respectively, in 0.3 M Tris buffer (pH 8.0). The binding parameters changed by addition of 0.5 M N a C l , 0.02 % N a N  3  or 10 m M P-mercaptoethanol (Table 1.6).  However, the reported numbers o f binding sites for glycinin and P-conglycinin were much higher than those reported in previous studies, as shown in Table 1.6, and the researchers explained this gap by the differences in the model systems used and more careful approaches to saturation of the system with ligand used in the study (O'Keefe et al., 1991a).  Comparison of the reported literature on binding of flavour compounds to soy proteins reveals some apparently conflicting data, which may be attributed to the differences in methods used (gel filtration technique, equilibrium dialysis method, or headspace analysis), origin and degree of denaturation of soy protein material during preparation and the experimental conditions applied. Most o f the research was performed in aqueous systems, where p H and other soluble component 14  may alter the binding affinity o f flavour compounds to soy protein. For example, salts from different buffers and presence of sodium azide may possibly affect the equilibria in the sample evaluated. Inclusion o f a reducing agent in some of the studies could significantly influence the conformational structure of soy protein resulting in possible changes in binding constant and the number of binding sites. The concentration of soy protein and the flavour compounds in each study were different, which may lead to different binding behavior o f the soy protein. In addition, experimental errors should be considered as binding constants and the numbers of binding sites in some studies were acquired by extrapolation of very small portions o f the binding curves, as indicated by O'Keefe (1991b).  In summary, even though the results from these previous studies have shown interaction of flavour compounds with soy protein data, the mode of the interactions has not been clearly elucidated as some conflicting results in terms of binding constant or the nature and magnitude of the protein flavour binding behavior were observed.  1.3.2.3. Soy protein isolate on binding of flavours The binding affinity o f SPI with vanillin in aqueous model system was compared with those of casein and whey protein isolates ( L i et al., 2000). It was concluded that SPI demonstrated lower affinity to vanillin than whey protein isolates and casein; binding o f vanillin to soy protein might be increased by conformational change, which could be induced by any process causing denaturation of soy protein.  Recently, Zhou et al. (2006) used inverse gas chromatography to examine the binding of selected butter flavour compounds such as diacetyl, hexanal, y-butyrolactone, and butyric acid in wheat based soda crackers, compared to the binding in soda crackers where 25 % of the wheat was replaced with soy protein isolate. Binding of diacetyl and hexanal was not affected by soy protein isolate in the wheat cracker but increased binding was observed with y-butyrolactone and butyric acid.  Comparison of the flavour binding properties o f SPIs from three different commercial sources produced by different methods i.e., membrane processing versus traditional extraction techniques 15  was evaluated at different relative humidity levels with hexane, hexanal, and 1-hexanol (Zhou and Cadwallader, 2006). Although the SPIs showed similar flavour binding patterns, differences in absolute flavour binding potential of individual volatile compounds across the SPIs were observed probably due to slightly different lipid content and degree of denaturation resulting from variation in the extent o f heat and chemical treatments during the processing.  1.4. Beef flavour 1.4.1. Addition of flavours to soy based products to simulate beef flavour Beef flavourings have been increasingly required for use in meat analogues as well as for the more traditional use in convenience or processed beef foods.  In recent years, soy protein has  become popular as a material for meat substitutes with increasing number o f vegetarians and the reported hypocholesterolemic effects of the protein (Anderson et al., 1995).  Due to the  difficulty of removing the indigenous undesirable soy aroma components, attention has turned to masking the residual off-flavour with simulated beef flavours in meat analogue products.  1.4.2. Natural beef flavours Depending on the temperature and method of cooking, over 1000 volatile compounds in meat have been isolated and identified since the 1960s including various hydrocarbons, aldehydes, ketones, alcohols, carboxylic acids, esters, lactones, furans, pyridines, pyrazines, alkylphenols, thiols, thiophenols, thiazoles, other nitrogen compounds, halogenated compounds, and sulfurcontaining compounds (Shahidi et al., 1986). However, the search for a specific character impact compound has not succeeded (Imafidon and Spanier, 1994).  Chang and Peterson (1977)  reported lactones, non-aromatic heterocyclic compounds, acyclic sulfur-containing compounds, and aromatic heterocyclic compounds as important contributors to beefy aroma notes.  In their  studies on the neutral fraction of roast beef, M i n et al. (1977 and 1979) suggested that lactones, substituted aromatics, furans, and sulfur-containing compounds contribute to roast beef flavour. The review by MacLeod et al. (1981) described more than 450 compounds identified from cooked beef but no single character compound was reported to be uniquely responsible for cooked beef aroma.  Even though some compounds have been thought to contribute more than  others, it seems to be very difficult to reconstitute the beef flavour by combination of a few 16  compounds.  Shahidi et al. (1986) concluded that, in contrast to fruits or chocolates, a particular  class of compounds did not in itself result in the meat flavour and that a number o f volatiles of different chemical classes existing in specific quantitative proportions were responsible for the meat flavours.  The researchers proposed several reasons for the difficulty in obtaining a  comprehensive understanding about the contribution of flavour compounds to cooked beef aroma such as the non-existence o f a true character impact compound, either a single compound or a particular class o f compounds, qualitative rather than quantitative nature o f the reported studies, either too small or not easily assessable threshold values for some flavour compounds, and possible involvement of additive, antagonistic, and synergistic effects.  Generally, beef flavours are derived from the complex interactions o f flavour precursors such as amino acids, peptides, sugars, thiamine, metabolites of nucleotides, lipids and products of lipid oxidation (Imafidon and Spanier, 1994).  In particular, the Maillard reaction has been known to  be important to cooked meat flavour and several research studies have been done to develop meat-like flavour by Maillard reaction with various amino acids and sugars (Ko et al., 1997a and 1997b).  1.4.3. Simulated meat flavours Although there have been various types of simulated meat flavourings such as simply blended spices (Hill, 1973), meat extracts from a byproduct from the corned beef industry (Pyke, 1975), hydrolyzed vegetable protein ( H V P ) by enzymatic, alkaline or acid hydrolysis (Prendergast, 1974) and hydrolyzed yeast by autolysis with proteolytic enzymes naturally present in the yeast (Cogman and Sarant, 1977), the most common types are thermally produced simulated meat flavours, the so called "reaction product" meat flavours (Wilson, 1975).  Generally, cooked beef  flavours are obtained by heating several amino acids with a reducing sugar i.e. by the Maillard reaction.  In particular, the reaction o f cysteine as a "sulfur-donor" compound with a reducing  sugar was reported to be important (May, 1974).  A s individual amino acids are relatively  expensive, protein hydrolysates which contain free amino acids, peptides, nucleotides, reducing sugars, carbonyl compounds, and sulfur compounds, have been used to produce beef flavours. The time and temperature of the reaction are critical for producing different flavours and there have been a great number o f "reaction product" patents (MacLeod et al., 1981). 17  1.5. Analysis of aroma compounds 1.5.1. Adsorption of aroma compounds - Headspace solid phase microextraction Solid phase microextraction (SPME) is a sample preparation technique that has been gaining popularity in recent years. Traditionally, liquid-liquid extraction, solid phase extraction, supercritical fluid extraction, static headspace sampling or dynamic headspace (purge-and-trap) methods have been used to extract and concentrate volatile compounds for analysis by G C , but one or more drawbacks o f each o f the sample preparation method have been reported such as high cost, multi-step preparation, low sensitivity, prolonged extraction time, artifact formation or solvent contamination (Braggins et al., 1999). Compared to the previous methods, the S P M E technique has been regarded as a simple, rapid, and economical method requiring no solvent (Yang and Peppard, 1994). It is a useful technique to isolate volatile components from a sample matrix, purify and concentrate the analytes, in which a fused silica fibre coated with a polymeric organic liquid is exposed into the sample headspace. The extracted volatiles in the coating are transferred to analytical equipment such as G C or H P L C for desorption and analysis.  Volatiles are captured by S P M E based on the theory o f equilibrium partitioning o f the analytes between extraction medium i.e. the solid phase of S P M E , and sample matrix (Zhang and Pawliszyn, 1993). A t equilibrium, the amount of analyte absorbed by a liquid coating is directly related to its concentrations in the sample  K VfC V  s  K V +V  s  f s  f s  0  f  where n is the mass o f an analyte absorbed by the coating; V f and V are the volumes of the s  coating in the fibre and the sample, respectively; Kf is the partition coefficient of the analyte s  between the coating in the fibre and the sample matrix; and Co is the initial concentration of the analyte in the sample (Zhang et al., 1994). This equation shows the linear relationship between the amount o f analytes absorbed by the fibre coating and the initial concentration of the analytes. In headspace S P M E , the driving force for analytes to transfer from sample matrix to fibre coating 18  involves three phases, which are matrix, headspace, and coating. In the case o f aqueous samples, the headspace/water partition coefficients (Kh ) for most compounds are directly related to s  Henry's constants, determined by volatility and hydrophobicity of the compounds. The sensitivity of headspace S P M E was reported to be almost the same as that o f direct S P M E (Zhang et al., 1994). When ion trap mass spectrometry was used, the detection limits o f the headspace S P M E and direct S P M E techniques were reported to be at the 1 ppt level (Zhang and Pawliszyn, 1993). Agitation o f the matrix, reduction o f headspace volume, or temperature increase of samples can be employed to shorten the equilibration time for less volatile compounds.  Originally, the S P M E technique was developed for analysis of pollutants in environmental water samples, by immersing the S P M E fibre in an aqueous sample (Arthur and Pawliszyn, 1990; Arthur et al., 1992).  It has also now been applied to flavour analysis, by employing headspace  S P M E sampling in foods such as cheese (Lecanu et al., 2002), edible oils (Steenson et al., 2002), coffee (Akiyama et al., 2003), ginger (Shao et al., 2003), kimchi (Lee et al., 2003), and sweet wines (Rodriguez-Bencomo et al., 2003).  1.5.2. Identification of aroma compounds A wide array o f techniques using a variety o f equipment has been developed to isolate and identify aroma components in food commodities and the usage and applications are well established. A great deal o f research on food flavour has been done with gas chromatography (GC) in conjunction with various detectors including flame ionization detector (FID) and mass selective detector ( M S D ) to identify volatile compounds in food matrix. With the increased sensitivity and precision o f G C instruments along with development o f techniques for adsorption of volatiles from food, a great number o f volatile compounds have been listed as aroma components in foods and beverages (Mistry et al., 1997).  1.5.3. Detection of aroma compounds - gas chromatography-olfactometry (GC-O) Even though the location o f peaks in G C chromatograms gives us a clue to identify volatiles in the foods, the areas o f the peaks do not necessarily reflect the aroma intensity o f the foods. It is difficult to decide the relative importance o f each aroma component in terms o f relative contribution to the overall flavour o f food as key aroma compound(s). It is also not a 19  straightforward process to select a few specific aroma chemicals responsible for the characteristic flavour notes including off-flavour, existing at very low concentration and/or with low odour thresholds in foods.  In that sense, G C analysis in conjunction with an olfactometric technique can be a useful tool to detect potent odour-active components from a complex mixture, which may contribute to the characteristic aroma o f a given food. Several GC-olfactometry (GC-O) methods have been reported, which can be divided into three categories, namely, dilution methods, intensity methods, and detection frequency methods.  1.5.3.1. Dilution methods The two most commonly used techniques of the dilution methods are aroma extract dilution analysis by Grosch (1993) and C H A R M analysis by Acree et al. (1984).  1.5.3.1.1. Aroma extract dilution analysis (AEDA) In A E D A , volatile compounds in the sample extract are isolated though an appropriate G C column and analyzed, more specifically, sniffed in the olfactometric detector (sniffing port of GC) generally in combination with a FID detector. The extract is subjected to stepwise serial dilution, typically in a 1:2 or 1:3 series, to be sniffed by panelists. Each panelist is asked to perceive odour components emerging in G C effluents and to provide a sensory descriptor for each perceived aroma. Subsequent analysis of each dilution in the series is performed until the odourants of interest cannot be perceived in the sniffing port.  The results can be expressed as  flavour dilution (FD) factor, which is the highest dilution value where the odourants can be still detected by G C - O . For instance, i f a two times dilution series was conducted and an odourant was detected at the sixth dilution but not detected at the seventh dilution, the F D factor for the compound would be 2 =64. The flavour profile o f a sample in G C - O can be presented in a F D 6  chromatogram consisting o f the logarithm o f the F D factor on the Y axis and the Kovats retention index (RI) on the X axis (Figure 1.4). The retention index o f each odourant can be calculated through a subsequent experiment where a series of ^-paraffins is analyzed under identical conditions as the odourant analysis, and retention times of paraffins are converted to retention indices according to their carbon number. For example, retention time o f paraffin with 6 carbons 20  is converted to 600 as retention index.  A E D A has been reported as a useful method to screen potent odourants in sample. However, this technique is very time consuming (Mistry et al., 1997) as several dilutions have to be analyzed to obtain F D factors o f odourants. Therefore, it can be a big obstacle to perform duplicate or triplicate analyses or to check reproducibility for a considerable number o f panelists. In addition, Abbott et al. (1993) observed the "gaps" in the coincident responses i.e. inconsistent responses for a series of dilutions o f the beer samples for 4 out of the 6 panelists in the study; a panelist did not detect an aroma compound at a particular dilution but detected it again at a higher dilution.  1.5.3.1.2. Combined Hedonic Aroma Response Measurements (CHARM) analysis Like A E D A , the C H A R M (Combined Hedonic Aroma Response Measurements) analysis is a dilution method based on odour-detection threshold, requiring several injections o f stepwise dilutions o f the original extract until odour-active compounds are no longer detected.  In  C H A R M analysis, panelists are asked to press a button when an odour is perceived and to hold it down until the characteristic o f the odour changes or disappears. The analysis can be conducted in conjunction with a commercially available system called CharmAnalysis™ along with Charmware™ software. The final results can be presented in a Charm chromatogram, which is created by accumulating the areas of square-shaped peaks generated from samples in all dilution series at a given retention index (Figure 1.5). Summed peak areas can be plotted with dilution value ( Y axis) and retention index ( X axis) and the areas under the curve are referred to as "charm values". Consequently, odour intensity is measured over the whole analysis time in CharmAnalysis and expressed as log "charm value", which refers to the area under the curve, vs. Kovats retention index while maximum odour intensity for each odour compound at a certain retention index is measured in A E D A and presented as log F D factors vs. Kovats retention index.  1.5.3.2. Intensity methods: Osme While A E D A and CharmAalysis have been used as screening tools for ranking odour-active compounds in foods based on detection thresholds o f odour components, the Osme method has been developed by McDaniel and co-workers (Miranda-Lopez et al., 1992) to directly measure odour intensity o f a compound perceived in sniffing port. Therefore, theoretically, Osme needs  21  only one injection per sample i f panels are well trained. In the Osme method, a group of panelists is used to rate aroma intensity o f odour components in the undiluted extract eluting from G C - O by using a computerized time-intensity device with 16 point scale, which would provide an aromagram called an osmegram, in which Y axis and X axis represent average intensity by panels and Kovats retention index, respectively (Figure 1.6). The panelists are also asked to describe the sensory characteristic of each odour compound, which is recorded. However, very high variability o f intensity evaluations within and between panelists (Guichafd et al., 1995) as well as day-to-day variation in sensitivity (Pollien et al., 1997) were reported, indicating the importance of panel training in the Osme method.  1.5.3.3. Detection frequency method Detection frequency methods were originally developed by Pollien et al. (1997), in which sniffing runs are replicated by a group of panelists (minimum 6-8, ideally 8-10 assessors). Panelists are asked to evaluate the effluents of the undiluted extract and record by a computerized device the beginning and the end of elution for each odour compound.  In this method, the  number of panelists detecting an odour compound at a certain retention index is employed to quantify the aroma intensity of the compound, rather than the dilution values measured in A D E A and CharmAnalysis or the perceived intensity measured in Osme. A n aromagram o f odour compounds is obtained by plotting the detection frequency (DF) or nasal impact frequency (NIF) on the Y axis and the Kovats retention index on the X axis (Figure 1.7). A value of 100 % D F or NIF means detection by all panelists. Satisfactory repeatability between independent panels was reported to be achieved without any panel training before analysis (Pollien et al., 1997).  Comparison o f the three G C - O methods, which were A E D A (dilution method), Osme (intensity method), and olfactometry global analysis (detection frequency method), was performed by Le Guen and co-workers (Le Guen et al., 2000) to evaluate the main impact odourants of cooked mussels. The results from the three olfactometric methods were very similar and significantly positively correlated (p values ~ 0.00001).  22  1.6. Analysis of structural properties of SPI  1.6.1. Sulfhydryl and disulfide groups The most widely used method to determine the content of S H and SS groups in proteins is Ellman's method (Ellman, 1959), in which S H groups in proteins react with 5,5'-dithio-bis-2nitrobenzoic acid ( D T N B ) to produce 5-thiobis(2-nitro)benzoic acid (TNB), which has an extinction coefficient o f 13,600 IVT'cm" at 412 nm. To analyze the total S H groups in proteins, 1  denaturants are used to expose or make the buried S H groups accessible (Beveridge et al., 1974). For total content o f SS+SH groups in proteins, the SS bonds are first cleaved by the excess sodium sulfite in the presence o f guanidine thiocyanate or sodium dodecyl sulphate (SDS) as a denaturant, and the liberated thiols can be measured by subsequent reaction with disodium-2nitro-5-thiobenzoate (Thannhauser et al., 1984)  The sulfhydryl (SH) group has been regarded as the most reactive and important functional group in various proteins as it affects many characteristics o f food proteins including emulsifying activity, foaming, whipping, gelation, solubility, swelling, water binding, antioxidant action, bread baking quality, and surface chemistry of food proteins (Owusu-Apenten, 2005). Cysteine residues with S H groups are found in almost all the storage and structural proteins while disulfide bond (SS) associated with cystine residues are also observed as essential structural units in most proteins (Friedman, 2001).  A broad range of SH+SS contents of soy proteins have been reported. Nakamura et al. (1984) reported the average values o f surface, internal, and total S H contents o f glycinin from five soybean cultivars as 0.6, 1.3, and 1.9 mole SH/mole glycinin, respectively. Total S H contents of SPI in buffer at p H 8 and p H 3 were reported to be 0.29 and 5.4 pmole/g protein, respectively, and these values were decreased with high pressure treatment between 400 and 600 M P a (Puppo et al., 2004). In addition, total S H content of SPI was reported as 8.3 pmole/g protein and it was increased to 10.1 pmole/g protein with addition of the antioxidant Tenox 22, while the total S H + SS content o f 52.9 pmole/g protein was decreased to 43.9 pmole/g protein (Boatright and Hettiarachchy, 1995).  23  1.6.2. Surface hydrophobicity Surface hydrophobicity (So), which can be measured using hydrophobic fluorescent probes as the initial  slope o f relative fluorescence  intensity versus  protein concentration,  should  be  distinguished from total hydrophobicity, which is the calculated average hydrophobicity o f a protein multiplied by the number of residues in the molecule. Several methods to assess hydrophobicities in proteins have been developed aside from using hydrophobicity scales of the constituent amino acids e.g. partition methods where partition coefficients o f the proteins were measured from the solubility ratio between a polar and an immiscible nonpolar solvent, binding methods with hydrophobic ligand, and spectroscopic methods  measuring either intrinsic  fluorescence intensity o f aromatic amino acids or fluorescent probes bound to proteins (Li-Chan, 1991). Many anionic fluorescent probes such as l-anilinonaphthalene-8-sulfonate (ANS) and cisparinaric acid ( C P A ) have been widely used, which have low quantum yield of fluorescence in aqueous solution but amplify upon binding to accessible hydrophobic regions of proteins. However, Alizadeh-Pasdar and Li-Chan (2000) reported that interpretation  of measured  hydrophobicity based on these anionic probes was not easy due to the influence o f electrostatic interactions between anionic probe and protein under various p H conditions. They suggested the use of uncharged fluorescent probe, 6-propionyl-2(N,N-dimethyl-amino)naphthalene ( P R O D A N ) based on the comparison of surface hydrophobicity of proteins at various p H using those different probes i.e. A N S , C P A , and P R O D A N .  So is regarded as one of the important parameters used to investigate the hydrophobic nature relating to functionality o f proteins including soy protein since hydrophobic regions at the protein surface have been suggested as the main structural factor governing the functional properties of food proteins by many researchers (Hayakawa and Nakai, 1985; Molina et al., 2001; Molina Ortiz et al., 2004; Rickert et al., 2004). For instance, emulsion stability correlated linearly with the surface hydrophobicity o f glycinin at each ionic strength (0.01, 0.1, and 0.5) while foaming power correlated curvilinearly with the surface hydrophobicity implying influences by other factors (Matsudomi et al., 1985a). The reported So values for SPI and SPI treated by mild acid treatment with 0.05 N HC1, enzymatic deamidation with chymotrypsin, or heat treatment at 100 °C for 10 minutes measured using A N S were 37, 143, 250, and 1025, respectively (Boatright and 24  Hettiarachchy, 1995; Li-Chan, 1991; Molina Ortiz et al., 2004; Rao et al., 2002). The S values of 0  soy protein, P-conglycinin, and glycinin by C P A were measured to be 105, 99, and 44, respectively (Kato et al., 1987; Matsudomi et al., 1985b). Generally, the variability o f the So values previously reported in the literature is probably due to differences o f SPI source, processing, purity or composition, and/or diverse analysis conditions in terms of p H , ionic strength, and temperature in addition to different fluorescent probes or methods for surface hydrophobicity.  1.6.3.  Raman spectroscopy  1.6.3.1.  Characterization and application of Raman spectroscopy  Raman spectroscopy has proved to be a practical tool to investigate the relationship between structure and function as the Raman effect depends on interactions between atomic positions, electron distribution and intermolecular forces (Carey, 1999). In Raman spectroscopy, molecules are irradiated by intense laser beams in visible, U V or near infrared (NIK) region (vo) and excited up to a "virtual state", which is between the ground state and the first excited electronic state (Figure 1.8). Although most of the excited molecules return to the original energy level, which is called Rayleigh scattering, a very small fraction of photons arrives at different vibration energy levels from the excitation laser, which is known as the Raman effect. Therefore, vibrational frequency (v ) is measured as a shift from the incident beam frequency in Raman spectroscopy, m  and depends on the chemical structure o f the molecules responsible for the scattering (Ferrato and Nakamoto, 1994).  Although both Raman and IR spectroscopies provide information on vibrational bending and stretching modes of molecules, there are unique features to each spectroscopic technique. Some particular vibrations are intrinsically strong in Raman spectra while weak in IR. In general, strong vibrations are observed in Raman i f the bond is non-polar (C=C, C = C , C - C , and S-S) whereas strong vibrations are shown in IR i f the bond is polar ( O - H and N - H ) and stretching vibrations are commonly stronger than bending vibrations in Raman spectra (Ferraro and Nakamoto, 1994). Due to a strong interfering band from water, generally dry or non-aqueous samples are applied in IR but Raman spectra of samples in aqueous solution can be analyzed without major interference as water is a weak contributor to Raman signal (Li-Chan, 1996). Another great advantage 25  compared to other spectroscopic methods is that Raman spectroscopy does not need optically clear samples, which can be aqueous or non-aqueous liquids, gels, films, powders, crystals or even gasous state, and also only a small sample area is required as the diameter of the laser beam is only 1-2 mm. In addition, Raman spectra can be obtained with samples such as hygroscopic or air-sensitive component, simply placed in sealed glass tubing, which is not possible in IR spectroscopy as IR radiation is absorbed by glass tubing (Ferraro and Nakamoto, 1994).  However, Raman spectroscopy has several disadvantages.  Compared to strong Rayleigh  scattering, Raman scattering is inherently weak as only one out o f each million photons experiences such energy shift, requiring relatively high concentration of analyte or considerable excitation laser power. Recent improvement in detectors and optical filters has minimized the limitation (Reipa, 2005). Another frequent problem is the possible interference by fluorescence in some samples, which sometimes overlap with the Raman scattering signal. Fluorescence can sometimes be alleviated with cautious sample purification or photo bleaching before data acquisition (Reipa, 2005). The limitation due to the background fluorescence can be also resolved by using a newer technique known as Fourier-transform (FT) Raman spectroscopy that uses nearinfrared (NIR) excitation with the N d : Y A G laser line at 1064 nm; this technique also offers improved resolution, shorter spectral acquisition time, and improved signal to noise ratios over conventional Raman spectroscopy (Li-Chan et al., 2002).  1.6.3.2. Protein structure analysis using Raman spectroscopy Raman spectroscopy has been reported as a valuable tool to understand the secondary structure of proteins and reveal the information of protein side chains of amino acid residues such as tryptophan, tyrosine, phenylalanine, and sulfhydryl groups. Structural changes o f various food proteins have been investigated using Raman technique such as SPI, spray-dried egg white and whey protein isolate (Zhao et al., 2004a and 2004b), whey protein (Nonaka et al., 1993), cod myosin (Careche and Li-Chan, 1997), oat globulin (Ma et al., 2000), cod collagen (Badii and Howell, 2003), red bean globulin (Meng et al., 2003), yam proteins (Liao et al., 2004), hake muscle protein (Herrero et al., 2004). In addition, interaction o f trehalose with egg white lysozyme (Belton and G i l , 1994), interactions of lysozyme with whey proteins (Howell and L i Chan, 1996), and interaction of lysozyme with corn oil (Howell et al., 2001) were also examined 26  using Raman spectroscopy.  The frequency and the relative intensity in Raman spectra provide information on vibrational motions of various amino acid side chains and - C O - N H - polypeptide back bone, which are sensitive to chemical changes and the surrounding microenvironments. The side chains of aromatic amino acids such as tryptophan, tyrosine, and phenylalanine demonstrate characteristic Raman bands at 760 cm" , 850/830 cm" , and 1003 cm" , respectively. The changes in band 1  1  1  intensity of tryptophan or the ratio from tyrosine doublet are useful to monitor the polarity of the microenvironment, or hydrogen bond involvement.  The peak intensity o f phenylalanine is not  dependent on microenvironment and hence is often used as an internal standard. The C - H stretching and bending mode o f aliphatic amino acid residues shown at 2800-3000 cm" and 14401  1465 cm" , respectively, can be applied to monitor hydrophobic interactions between aliphatic 1  residues. In addition, disulfide and sulfhydryl groups may be detected in bands near 500-550 cm"  1  and 2550-2580 cm" , respectively (Li-Chan, 1996). Among several distinct vibrational modes o f 1  C O - N H - amide or peptide bond, amide I (near 1660 cm" ) and amide III bands (near 1240 cm" ) 1  1  are the most powerful for the investigation of secondary structure and conformation of the protein molecule containing the propensity df a-helix, P- sheet, unordered and turn structures (Li-Chan et al., 2002). Williams and Dunker (1981) reported that Raman spectra in amide I region were useful to determine the detailed and reasonably accurate estimates of secondary structure of proteins by demonstrating less than 6 % absolute difference between X-ray and Raman estimates of the secondary structure of 17 protein samples.  Determination of the extent o f acetylation in 3 protein samples (SPI, spray-dried egg white, and whey protein isolate) by Raman spectroscopy was investigated (Zhao et al., 2004a). A new C = 0 stretching vibration at 1737 cm" appeared during acetylation, which was used to quantify the 1  extent of modification. Correlation coefficients obtained by linear regression analysis o f the calibration curves for the Raman intensity ratio (1737 cm"Vl003 cm" ) to the extent of O1  acetylation in SPI, egg white, and whey protein were 0.9984, 0.9979, and 0.9989, respectively. In addition, while random coils and p-sheet at 1246 cm" and 1667 cm" were reported to be major 1  1  secondary structures in unmodified SPI, protein denaturation and dissociation at the initial stages of modification and transformation from random coil to P-sheet structure at the later stages of 27  acetylation were speculated according to the Raman spectral data. A similar approach was used to examine the effect of chemical modification of proteins with succinic anhydride (Zhao et al., 2004b). In the study, conformational changes such as transition of ordered into disordered structures, relocation of tryptophan residues from interior hydrophobic microenvironment to less hydrophobic exterior region, and conformational shift of cystine residue were reported.  28  1.7. Hypotheses 1.  Changes in releasing o f individual volatile compounds in simulated beef flavour (SBF) from soy protein isolate (SPI) can be monitored effectively by using gas chromatographic techniques coupled with headspace-solid phase microextraction ( H S - S P M E ) .  2.  Reduction in the perceived aroma intensity of S B F when added to a soy-based matrix results from binding o f aroma impact compounds to SPI.  3.  Aroma intensities o f volatile compounds in S B F , which are results o f perception by the human nose, can be depicted using gas chromatography-olfactometry.  4.  Aroma intensities of SBF-SPI mixtures, as evaluated by gas chromatography and descriptive analysis, are affected by the structural properties of the proteins, which can be altered by addition o f ingredients through changes in disulfide bond, hydrogen bond, or hydrophobic interaction.  5.  Structural changes in SPI induced by added ingredients can be detected using spectroscopic techniques  such as  fluorescence  and FT-Raman  spectroscopy  or sulfhydryl  group  measurement.  29  1.8. Objectives  The overall objective o f this thesis is to elucidate the interaction between SPI and aroma components of S B F . The specific objectives are to:  1.  Develop a sensitive and reproducible analytical method for H S - S P M E that is able to monitor changes in the headspace composition of samples containing SPI and S B F .  2.  Identify the volatile compounds in the headspace of the S B F .  3.  Determine aroma impact volatile compounds in the headspace o f S B F .  4.  Characterize the sensory perception of samples with SPI and S B F in terms o f describing the attribute and intensity of the beefy note and explore the correlation between the instrumental chromatographic data and sensory evaluation information.  5.  Investigate the changes in flavour binding behavior of the SPI in terms o f beef attribute related aroma compounds as affected by added ingredients.  6.  Monitor changes in the protein structure in SPI induced by the ingredients to understand the effect of each ingredient on the reduction of SBF binding to SPI.  30  Soy meal (defatted)  Extract (aqueous, p H 9.0) Centrifuge  Residue  Extract Isoelectric (pH 4.5) Precipitation  Protein Curd  Wash Dry  Soy protein isolate Isoelectric form  Whey  Wash Neutralize Dry  Soy protein isolate Proteinate form  Figure 1.1. Preparation o f soy protein isolates from defatted soy meal (Adapted from Hettiarachchy and Kalapathy, 1997).  (A-S-S-B)  , _ „ ^ 6(A-S-S-B) 60 k D a r  L  N  6  B-mercaptoethanol -  •  6(A) + 6(B) 40 k D a 20 k D a  Figure 1.2. Mechanisms for dissociation of soybean glycinin from the hexameric form possessing one disulfide bond (SS) into its subunits and further into the acidic (A) and basic (B) polypeptides (Adapted from Nielson, 1985).  32  s  Glycinin  2m  O  6 ^7 Heating  A  O  <I  Further heating  1  Precipitate  •Is!: :;?!•  s 0  •i h  # 1  I;:s:-:  .Polymer  Polymer  Q  A  :  .: I :: ::S!:I  .:s>  i  AA A  Supernatant  P-Conglycinin  0  •s  s  5  0  A  Monomer  A Monomer, .  :• O l i g o m e r :: 01igomer :  Figure 1.3. Schematic subunit interaction between P-conglycinin (7S globulin) and glycinin (1 IS globulin) on heating (Adapted from Yamauchi et al., 1991). , interaction by secondary force; S-S, disulfide bond; O , acidic polypeptide; • , basic polypeptide; A , a, a'-polypeptide; A , P-polypeptide  33  20484  25 21 26  5121 •2  35  17  8  128  12  30  3334  19  28  a  36  U-.  3216-  1  i!38  2  700  900  1100  1300  1500  1700  (RIiOV-1701)  Figure 1.4. Example of flavour dilution chromatogram from aroma extract dilution analysis (Source: Blank, 1997).  34  Dilution. I  Sniff Port Responses T f 1 JTL  Dilution 2 Dilution 3  Coincident Response Chromatogram  n  _—J—\  ^ " k .  n-I  Charm Chromatogram F = Dilution Factor Charm = peak area  a o • 3  3  5, Retention Index  Figure 1.5. Example of Charm chromatogram from CharmAnalysis (Source: Mistry et al., 1997).  35  12-  :V>:  3  «l A  8  ja  o a.  :>: «  8i  a. « o  E ©  Imax  ca  i—  O V  o  2  . — —  1200  1300  1400  1500  1600  Kovats Indices  Figure 1.6. Example of osmegram from intensity methods (Source: Blank, 1997).  36  Retention Index  Figure 1.7. Example of aromagram from detection frequency method (Source: Le Guen et 2000).  Excited electronic states  t  vo  vo  OX  Vo  v  Vo  0  +Vi  -Vi  4)  Vo  C  w  Ground electronic states  Vi  Infrared absorption  Rayleigh scattering  Stokes'  Anti-Stokes'  1 Raman scattering  Stokes'  Resonance Raman scattering  Figure 1.8. The relationships between infrared absorption, Rayleigh scattering and Raman scattering; Dotted line indicates a "virtual state" (Adapted from Li-Chan, 1996).  38  Table 1.1. Average chemical compositions (% by weight, dry basis) o f defatted soy flour, soy protein concentrate, and soy protein isolate (Adapted from MacLeod and Ames, 1988).  Component  Defatted soy flour (%)  Soy concentrate (%)  Soy isolate (%)  Protein  56.0  72.0  96.0  Lipids  1.0  1.0  0.1  14.0 19.5  2.5 15.0  0 0.3  Fibre  3.5  4.5  0.1  Ash  6.0  5.0  3.5  Carbohydrate Soluble Insoluble  39  Table 1.2. Five major soybean glycinin subunits (G1-G5) and their paired acidic (A) and basic (B) polypeptides (Adapted from L i u , 1997).  Group  Subunit  I  GI  Ai B  I  G2  Ai Bl  I  G3  A B  l a  58  II  G4  A3B4  62  II  G5  A A B  Structure o f subunit  a  2  5  4  58  2  b  M w (kDa)  b  3  58  69  f  40  Table 1.3. A m i n o acid composition of P-conglycinin and glycinin of soybean (Adapted from Krishnan, 2005). 1  P-Conglycinin (%) Amino acid Alanine Arginine Asparagine Aspartic acid Cysteine Glutamine Glutamic acid Glycine Histidine Isoleucine Leucine Lysine Methionine Phenylalanine Proline Serine Threonine Tryptophan Tyrosine Valine  Glycinin (%)  a'  a  p  GyP  Gy2  Gy~3  Gyl  Gy5  4.2 6.7 6.5 4.5 0.8 8.9 13.6 4.7 3.5 4.7 7.2 7.2 0.3 4.5 5.7 7.4 2.0 0.5 2.5 4.5  4.5 7.9 7.1 4.8 0.9 8.3 13.6 4.1 1.4 5.3 8.3 6.2 0.2 4.6 6.7 7.2 1.9 0.3 2.6 4.1  5.3 7.0 8.0 5.1 0.0 8.0 8.9 4.3 1.9 6.3 9.9 4.8 0.0 6.8 5.1 7.5 2.4 0.0 2.9 5.8  5.7 5.7 7.8 3.6 1.7 10.1 8.6 7.4 1.7 5.5 6.9 5.0 1.3 4.2 6.1 6.7 4.2 0.8 2.3 4.8  6.6 6.2 8.6 3.9 1.7 10.9 7.9 7.3 0.9 4.9 7.1 3.9 1.5 4.1 5.6 6.4 3.9 0.9 2.4 5.6  5.8 5.6 7.5 3.4 1.7 10.5 8.6 7.3 1.3 5.2 6.9 3.9 1.1 5.2 6.2 7.3 3.9 0.9 2.4 5.4  4.3 7.3 6.1 6.0 1.1 9.5 9.9 6.3 2.8 3.9 6.9 4.8 0.4 3.0 6.9 7.3 3.5 0.9 2.6 6.5  3.7 6.3 6.7 4.7 1.2 9.5 8.7 7.9 3.2 3.7 6.9 3.7 0.6 3.2 7.3 7.7 3.9 0.8 3.2 7.1  Calculated amino acid composition of mature proteins from the sequences obtained from protein database. Gyl, Gy2, Gy3, Gy4, and Gy5 are genes encoding glycinin G I , G2, G3, G4, and G 5 , respectively.  2  Table 1.4. Functional properties of soy protein isolates in food systems (Adapted from Hettiarachchy and Kalapathy, 1997).  Functional Properties  Mode o f Action  Example of food applications  Solubility  Protein solvating  Beverages  Viscosity  Thickening, water binding  Soups, gravies  Gelation  Protein matrix forming and setting  Meats, curds, cheeses  Cohesion-adhesion  Protein acting as adhesive material  Elasticity  Disulfide linking in deformable gels  Emulsification  Forming and stabilizing of fat emulsions  Meats, sausages, baked goods, pasta products Meats, bakery items  Sausages, bologna, soups, cakes  Fat absorption  Binding o f free fat  Meats, sausages, doughnuts  Flavour-binding  Adsorbing, entrapping, releasing  Simulated meats, bakery items  Foaming  Forming film to entrap gas  Whipped toppings, chiffon desserts, angel cakes  42  Table 1.5. Thermodynamic constants for the binding of model flavour compounds to soy protein.  Ligand  Type o f soy  N o . of binding  preparation  sites (»)  Binding constant i (K<. , M ) q  AG, kcal/mol  Reference  1-Hexanal  Native  2  173.4  -3.007  1  1-Hexanol  Native  -  80.3  -2.558  1  Butanal  Native  -  10916  -  2  2-Butanone  Native  -  4975  -  2  1-Butanol  Native  -  2100  -  2  2-Heptanone  Native  4  110  -2.781  3  2-Octanone  Native  4  310  -3.395  3  2-Nonanone  Native  4  930  -4.045  3  2-Nonanone  Partly denatured  4  1240  -4.215  3  2-Nonanone  Succinylated  2  850  -3.992  3  5-Nonanone  Native  4  541  -3.725  3  Nonanal  Native  4  1094  -4.141  3  2-Nonanone  Native  5.5  570  _  4  n\ the total number o f binding sites per mole o f protein K : the intrinsic binding constant  'Reference from 1 - Arai et al. (1970); 2 - Beyeler and Solms (1974; 20 °C, p H 4.5); 3 Damodaran and Kinsella (1981a; 25 °C); 4 - O ' N e i l l and Kinsella (1987, 25 °C). 2  - : not reported  43  Table 1.6. Number of binding sites on glycinin and P-conglycinin for various volatile flavour compounds at three different temperatures (Adapted from O'Keefe et al., 1991a and 1991b). No. of binding sites Compound  *t 5 °C glycinin  Butanal  4  Pentanal Hexanal Octanal  4  4  4  1  at 20 °C  P-conglycinin  glycinin  at 30 °C  P-conglycinin  glycinin  P-conglycinin  163±32  30±13  205±73  22±8  171±46  21 ±46  242±82  28±9  196±24  37±6  193±36  49±36  149±16  23±4  96±6  32±7  108±10  26±10  101±5  18±3  76±6  38±5  66±3  59±3  2-Hexanone  4  165±49  23±3  58±6  24±2  67±7  38±7  3-Hexanone  4  165±14  35±18  54±5  36±3  60±6  22±5  2-Nonanone  4  189±33  18±2  71±4  46±42  96±14  30±14  5-Nonanone  4  245±48  19±2  67±8  44±14  65±31  20±31  Hexanol  79±13  12±3  42±12  8±5  42±10  10±10  Hexane  14±1  45±12  NA  NA  NA  NA  _3  _  108±10  26±10  _  _  2  Hexanal  4  Hexanal  5  -  -  38±3  68±8  -  -  Hexanal  6  -  -  76±3  32±2  -  -  Hexanal  7  112±10  32±2  'Mean number of binding sites ± standard deviation per mole of protein from triplicate analysis 2  N o affinity  3  - : not reported  4  0.03 M Tris buffer (pH 8.0)  5  0.03 M Tris buffer with 0.5 M N a C l  6  0.03 M Tris buffer with 0.02 % N a N  7  3  0.03 M Tris buffer with 10 m M P-mercaptoethanol 44  1.9. References Arai, S., Noguchi, M . , Yamashita, M . , Kato, H . , & Fujimaki, M . (1970). Studies on flavor components in soybean Part VI: Some evidence for occurrence of protein-flavor binding. Agricultural and Biological Chemistry, 34, 1569-1573.  Abbott, N . , Etievant, P., Issanchou, S., and Langlois, D. (1993). Critical evaluation of two commonly used techniques for the treatment of data from extract dilution sniffing analysis. Journal ofAgricultural and Food Chemistry, 41, 1698-1703.  Abbott, T. P., Nabetani, H., Sessa, D. J., and Wold, W. J. (1996). Effects of bound water on FTIR spectra of glycinin. Journal ofAgricultural and Food Chemistry, 44, 2220-2224.  Acree, T. E., Barnard, J., and Cunningham, D. G. (1984). A procedure for the sensory analysis of gas chromatographic effluents. Food Chemistry, 14(4), 273-18. Akiyama, M . , Murakami, K., Ohtani, N . , Iwatsuki, K., Sotoyama, K., Wada, A . , Tokuno, K., Iwabuchi, H., and Tanaka, K . (2003). Analysis of volatile compounds released during the grinding of roasted coffee beans using solid-phase microextraction. Journal of Agricultural and Food Chemistry, 51(7), 1961-1969. Anderson, J. W., Johnstone, B. M . , and Cook-Newell, M . L . (1995). Meta-analysis of the effects of soy protein intake on serum lipids. The New England Journal of Medicine, 333(5), 276282. Arthur, C. L., and Pawliszyn, J. (1990). Solid phase microextraction with thermal desorption using fused silica optical fibers. Analytical Chemistry, 62, 2145-2148. Arthur, C. L., Killam, L . M . , Motlagh, S., Lim, M . , Potter, D. W., and Pawliszyn, J. (1992). Analysis of substituted benzene compounds in groundwater using solid-phase microextraction. Environmental Science and Technology, 26, 979-983.  Aspelund, T., and Wilson, L . (1983). Adsorption of off-flavor compounds onto soy protein: A thermodynamic study. Journal ofAgricultural and Food Chemistry, 31, 539-545.  Badii, F., and Howell, N . K . (2003). Elucidation of the effect of formaldehyde and lipids on frozen stored cod collagen by FT-Raman spectroscopy and differential scanning calorimetry. Journal ofAgricultural and Food Chemistry, 51, 1440-1446.  Belton, P. S., and Gil, A . M . (1994). IR and Raman spectroscopic studies of the interaction of trehalose with hen egg white lysozyme. Biopolymers, 34, 957-961. Berk, Z. (1992). Isolated soy bean protein. In Technology of Production of Edible Fours and Protein Products from Soybeans, (pp. 83-105). Rome: Food and Agriculture Organization of the United Nations. 45  Beveridge, T., Toma, S. J., and Nakai, S. (1974). Determination o f S H - and SS-groups in some food proteins using Ellman's reagent. Journal of Food Science, 39, 49-51. Beyeler, M . and Solms, J. (1974). Interaction o f flavor model compounds with soy protein and bovine serum albumin. Lebensmittel Wissenschaft und Technologie, 7, 217-219. Blank, I. (1997). Gas chromatography-olfactometry in food aroma analysis. In R. Marsili, Techniques for Analyzing Food Aroma, (pp. 293-329). N e w York: Marcel Dekker. Boatright, W . L . , and Crum, A . D . (1997). Odor and flavor contribution o f 2-pentyl pyridine to soy protein isolates. The Journal of the American Oil Chemists' Society, 74(12), 1575-1581. Boatright, W . L . , and Hettiarachchy, N . S. (1995). Soy protein isolate solubility and surface hydrophobicity as affected by antioxidants. Journal of Food Science, 60, 798-800. Boatright, W . L . , and L e i , Q . (1999). Compounds contributing to the "beany" odor o f aqueous solutions o f soy protein isolates. Journal of Food Science, 64(4), 667-670. Braggins, T. J., Grimm, C . C , and Visser, F. R. (1999). Analysis o f food volatiles using S P M E . In J. Pawliszyn, Applications of Solid Phase Microextraction, (pp. 407-422). Cambridge: The Royal Society o f Chemistry. Careche, M . and Li-Chan, E . C . Y . (1997). Structural changes in cod myosin after modification with formaldehyde or frozen storage. Journal of Food Science, 62, 717-723. Carey, P. R. (1999). Raman spectroscopy, the sleeping giant in structural biology, awakes. The Journal of Biological Chemistry, 274, 26625-26628. Chang, S. S., and Peterson, R. J. (1977). Symposium: The basis o f quality in muscle food. Recent developments in the flavor o f meat. Journal of Food Science, 42, 298-305. Cheftel, J. C , Cuq, J., and Lorient, D . (1985). Amino acids, peptides, and proteins. In O. R. Fennema, Food Chemistry, (pp. 245-369). N e w York: Marcel Dekker Inc. Cogman, D . G . K . , and Sarant, R . (1977). N e w development in savory flavor and enhancement. Food Trade Review, 47(1), 15-16. Cowan, J. C , Rackis, J. J., and Wolf, W. J. (1973). Soybean protein flavor components: a review. The Journal of the American Oil Chemists' Society, 50 427A-435A, 444A. Damodaran, S., and Kinsella, J. E . (1981a). Interaction o f carbonyls with soy protein: Thermodynamic effects. Journal of Agricultural and Food Chemistry, 29, 1249-1253. Damodaran, S., and Kinsella, J. E . (1981b). Interaction o f carbonyls with soy protein: Conformational effects. Journal of Agricultural and Food Chemistry, 29, 1253-1257.  46  Damodaran, S., and Kinsella, J. E . (1982). Effect of conglycinin on the thermal aggregation of glycinin. Journal of Agricultural and Food Chemistry, 30, 812-817. Ellman, G . L . (1959). Tissue sulfhydryl groups. Archives of Biochemistry and Biophysics, 82, 7072. Ferraro, J. R., and Nakamoto, K . (1994). Basic theory. In Introductory Raman Spectroscopy, (pp. 1-94). San Diego: Academic Press, Inc. Friedman, M . (2001). Application o f the S-pyridylethylation reaction to the elucidation o f the structure and functions of proteins. Journal of Protein Chemistry, 20, 431-453. German, B . , Damodaran, S., and Kinsella, J. E . (1982). Thermal dissociation and association behavior of soy proteins. Journal of Agricultural and Food Chemistry, 30, 807-811. Gremli, H . A . (1974). Interaction o f flavor compounds with soy protein. The Journal of the American Oil Chemists' Society, 51, 95A-97A. Greuell, E . H . M . (1974). Some aspects of research in the application o f soy proteins in foods. The Journal of the American Oil Chemists' Society, 51, 98A-100A. Grosch, W . (1993). Detection of potent odorants in foods by aroma extract dilution analysis. Trends in Food Science and Technology, 4(3), 68-73. Guichard, H . , Guichard, E . , Langlois, D . , Issanchou, S., and Abbott, N . (1995). G C sniffing analysis: olfactive intensity measurement by two methods. European Food Research and Technology, 201, 344-350. Hayakawa, S., and Nakai, S. (1985). Relationships of hydrophobicity and net charge to the solubility o f milk and soy proteins. Journal of Food Science, 50, 486-491. Herrero, A . M . , Carmona, P., and Careche, M . (2004). Raman spectroscopic study of structural changes in hake (Meluccius merluccius L.) muscle proteins during frozen storage. Journal of Agricultural and Food Chemistry, 52, 2147-2153. Hettiarachchy, N . and Kalapathy, U . (1997). Soybean protein products. In K . L i u (Ed.), Soybeans: Chemistry, Technology and Utilization, (pp. 379-411). N e w York: Chapman & Hall. H i l l , M . H . (1973). Compounded seasonings for meat products. Flavour Industry, 4, 164-168. Hoogenkamp, H . W . (2006). Soy journey. In Soy Protein and Formulated Meat Products, (pp. 16). Cambridge: C A B I publishing. Howell, N . and Li-Chan, E . (1996). Elucidation of interactions o f lysozyme with whey proteins by Raman spectroscopy. International Journal of Food Science and Technology, 31, 439451. 47  Howell, N . K . , Herman, H . , and Li-Chan, E . C . Y . (2001). Elucidation of protein-lipid interactions in a lysozyme-corn oil system by Fourier transform Raman spectroscopy. Journal of Agricultural and Food Chemistry, 49, 1529-1533. Hsieh, O. A . L . , Huang, A . S., and Chang, S. S. (1981). Isolation and identification of objectionable volatile flavor compounds in defatted soybean flour. Journal of Food Science, 47, 16-18. Imafidon, G. I., and Spanier, A . M . (1994). Unraveling the secret of meat flavor. Trends in Food Science & Technology, 5(10), 315-321. Jung, S., Murphy, P. A . , Johnson, L . A . (2005). Physicochemical and functional properties of soy protein substrates modified by low levels of protease hydrolysis. Journal of Food Science, 70(2), C180-C187. Kalapathy, U . , Hettiarachchy, N . S., Myers, D . and Rhee, K . C . (1996). Alkali-modified soy proteins: Effect o f salts and disulfide bond cleavage on adhesion and viscosity. The Journal of the American Oil Chemists' Society, 73, 1063-1066. Kato, H . , D o i , Y . , Tsugita, T., Kosai, K , Kamiya, T., and Kurata, T. (1981). Changes in volatile flavour components of soybeans during roasting. Food Chemistry, 7, 87-94. Kato, A . , Tanaka, A . , Lee, Y . , Matsudomi, N . , and Kobayash, K . (1987). Effects of deamidation with chymotrypsin at p H 10 on the functional properties of proteins, Journal of Agricultural and Food Chemistry, 35, 285-288. Klotz, I. M . , Walker, F., and Pivan, R. B . (1946). Alkali-modified soy proteins: Effect of salts and disulfide bond cleavage on adhesion and viscosity. The Journal of the American Oil Chemists' Society, 73, 1063-1066. K o , S. N . , Nam, H . S., and K i m , W . J. (1997a). Development o f meat-like flavor by Maillard reaction with addition o f natural flavoring. Korean Journal of Food Science and Technology, 29(5), 839-846. K o , S. N . , Yoon, S. H . , Yoon, S. K . , and K i m , W. J. (1997b). Development o f meat-like flavor by Maillard reaction o f model system with amino acids and sugars. Korean Journal of Food Science and Technology, 29(5), 827-838. Krishnan, H . B . (2005). Engineering soybean for enhanced sulfur amino acid content. Crop Science, 45, 454-461. Lakemond, C . M . M . , de Jongh, H . H . J., Hessing, M . , Gruppen, H . , and Voragen, A . G . J. (2000). Soy glycinin: Influence of p H and ionic strength on solubility and molecular structure at ambient temperatures. Journal of Agricultural and Food Chemistry, 48, 1985-1990.  48  Le Guen, S., Prost, C . , and Demaimay, M . (2000). Critical comparison of three olfactometric methods for the identification of the most potent odorants in cooked mussels (Mytilus edulis). Journal of Agricultural and Food Chemistry, 48, 1307-1314. Lecanu, L . , Ducruet, D . , Jouquand, C , Gratadoux, J. J., and Feigenbaum, A . (2002). Optimization o f headspace solid-phase microextraction ( S P M E ) for the odor analysis of surface-ripened cheese. Journal of Agricultural and Food Chemistry, 50(13), 3810-3817. Lee, J. H . , Kang, J. FL, and M i n , D . B . (2003). Optimization o f solid-phase microextraction for the analysis o f headspace volatile compounds in kimchi, a traditional Korean fermented vegetable product. Journal of Food Science, 68(3), 844-848. Lei, Q., and Boatright, W. L . (2001). Compounds contributing to the odor of aqueous slurries of soy protein concentrate. Journal of Food Science, 66(9), 1306-1310. Lei, Q., and Boatright, W . L . (2006). Methionine is the methyl group donor for sulfite-associated methanethiol formation in isolated soy proteins. Journal of Food Science, 71, C527-C531. Li-Chan, E . C . Y . (1991). Hydrophobicity in food protein systems. In Encyclopedia of Food Science and Technology, (pp. 1429-1439). N e w York: John Wiley and Sons, Inc. Li-Chan, E . C . Y . (1996). The applications of Raman spectroscopy in food science. Trends in Food Science & Technology, 7, 361-370. Li-Chan, E . C . Y . , Ismail, A . A . , Sedman, J., and van de Voort, F. R. (2002). Vibrational spectroscopy o f food and food products. In J. M . Chalmers and P. R. Griffiths (Eds.), V o l . 5. Handbook of Vibrational Spectroscopy, (pp.3644-3646). Chichester: John Wiley & Sons Ltd. L i , Z . , Griin, I.U., and Fernando, L . N . (2000). Interaction of vanillin with soy and dairy proteins in aqueous model systems: A thermodynamic study. Journal of Food Science, 65, 997-1001. Liao, Y . - H . , Wang, C . - H . , Tseng, C . - Y . , Chen, H . - L . , L i n , L . - L . , and Chen, W. (2004). Compositional and conformational analysis o f yam proteins by near infrared Fourier transform Raman spectroscopy. Journal of Agricultural and Food Chemistry, 52, 8190-8196. L i u , K . (1997). Chemistry and nutritional value o f soybean components. In Soybeans: Chemistry, Technology and Utilization, (pp. 25-113). N e w York: Chapman & H a l l . Ma, C . - Y . , Rout, M , K . , Chan, W . - M . , and Phillips, D . L . (2000). Raman spectroscopic study of oat globulin conformation. Journal of Agricultural and Food Chemistry, 48, 1542-1547. MacLeod, G , and Ames, J. (1988). Soy flavor and its improvement. CRC Critical Reviews in Food Science and Nutrition, 27(4) 219-400. MacLeod, G , and Seyyedain-Ardebili, M . (1981). Natural and simulated meat flavors. CRC Critical Reviews in Food Science and Nutrition, 14(4), 309-437. 49  Maheshwari, P., O o i , E . T., and Nikolov, Z . L . (1995). Off-flavor removal from soy-protein isolate by using liquid and supercritical carbon dioxide. The Journal of the American Oil Chemists' Society, 72(10), 1107-1115. Malcolmson, L . J., and McDaniel, M . R. (1987). Flavor protein interactions in a formulated soup containing flavored soy protein. Canadian Institute of Food Science and Technology, 20(4), 229-235. Matsudomi, N . , M o r i , H . , Kato, A . , and Kobayashi, K . (1985a). Emulsifying and foaming properties of heat-denatured soybean 11S globulins in relation to their surface hydrophobicity. Agricultural and Biological Chemistry, 49, 915-919. Matsudomi, N . , Sasaki, T., Kato, A . , and Kobayashi, K . (1985b). Conformational changes and functional properties o f acid-modified soy protein. Agricultural and Biological Chemistry, 49, 1251-1256. May, C . G . (1974). A n introduction to synthetic meat flavors. Food Trade Review, 44(1), 7, 9-10, 12-14. Meng, G . , M a , C . - Y . , and Phillips, D . L . (2003). Raman spectroscopic study o f globulin from Phaseolus angularis (red bean). Food Chemistry, 81, 411-420. Meyer, E . W., and Williams, L . D . (1977). Chemical modification o f soy proteins. Advances in Chemistry Series, 160, 52-66. M i n , D . B . S., Ina, K . , Peterson, R. J., and Chang, S. S. (1977). The alkyl benzenes in roast beef. Journal of Food Science, 42, 503-505. M i n , D . B . S., Ina, K . , Peterson, R. J., and Chang, S. S. (1979). Preliminary identification of volatile flavor compounds in the neutral fraction of roast beef. Journal of Food Science, 44, 639-642. Miranda-Lopez, R., Libbey, L . M . , Watson, B.T., and McDaniel, M . R . (1992). Odor analysis of Pinot noir wines from grapes o f different maturities by a gas chromatography-olfactometry technique (Osme). Journal of Food Science, 57(4), 985-993, 1019. Mistry, B.S., Reineccius, T., and Olson, L . K . (1997). Gas chromatography-olfactometry for the determination o f key odorants in foods. In R. Marsili, Techniques for Analyzing Food Aroma, (pp. 265-292). N e w York: Marcel Dekker. Molina, E., Papadopoulou, A . , and Ledward, D . A . (2001). Emulsifying properties of high pressure treated soy protein isolate and 7S and 1 IS globulins. Food Hydrocolloids, 15, 263269. Molina Ortiz, S. E . , Puppo, M . C , and Wagner, J. R. (2004). Relationship between structural changes and functional properties of soy protein isolates-carrageenan systems. Food Hydrocolloids, 18, 1045-1053. 50  Morita, S., Fukase, M . , Yamaguchi, M . , Fukuda, Y . , and Morita, Y . (1996). Purification, characterization, and crystallization of single molecular species of P-conglycinin from soybean seeds. Bioscence Biotechnology and Biochemistry, 60, 866-873. Moriyama, T., Machidori, M . , Ozasa, S., Maebuchi, M . , Urade, R., Takahashi, K . , Ogawa, T., and Naruyama, N . (2005). A novel enzyme-linked immunosorbent assay for quantification of soybean P-conglycinin, a major soybean storage protein, in soybean and soybean food products. Journal of Nutritional Science and Vitaminology, 51, 34-39. Moser, H . A . , Evans, C . D . , Campbell, R. E . , Smith, A . K . , and Cowan, J. C . (1967). Sensory evaluation o f soy flour. Cereal Science Today, 12, 296-299, 314. Nakamura, T., Utsumi, S., Kitamura, K . , Harada, K . , and M o r i , T. (1984). Cultivar differences in gelling characteristics of soy bean glycinin. Journal of Agricultural and Food Chemistry, 32, 647-651. Nielson, N . C . (1985). Structure and complexity o f the 1 IS polypeptides in soybean. The Journal of the American Oil Chemists' Society, 62(12), 1680-1685. Nonaka, M . , Li-Chan, E . , and Nakai, S. (1993). Raman spectroscopic study o f thermally induced gelation o f whey proteins. Journal of Agricultural and Food Chemistry, 41,1176-1181. Ohren, J. A . (1981). Process and product characteristics for soya concentrates and protein isolates, The Journal of the American Oil Chemists' Society, 58, 333-335. O'Keefe, S. F., Resurreccion, A . P., Wilson, L . A . , and Murphy, P. A . (1991a). Temperature effect on binding o f volatile flavor compounds to soy protein in aqueous model systems. Journal of Food Science, 56, 802-806. O'Keefe, S. F., Wilson, L . A . , Resurreccion, A . P., and Murphy, P. A . (1991b). Determination of the binding of hexanal to soy glycinin and P-conglycinin in an aqueous model system using a headspace technique. Journal of Agricultural and Food Chemistry, 39, 1022-1028. Olsen, H . S., and Alder-Nissen, J. (1981). Application o f ultra- and hyperfiltration during production o f enzymatically modified proteins (including application to soy proteins). ACS Symposium Series, 154, 133-169. O ' N e i l l (1996). Flavor binding by food proteins: A n overview. In R. J. McGorrin and J. V . Leland (Eds.), Chapter 6. Flavor-Food Interactions, A C S Symposium Series 633, (pp.5974). Washington, D C : American Chemical Society. O ' N e i l l , T. E . , and Kinsella, J. E . (1987). Flavor protein interactions: Characteristics o f 2nonanone binding to isolated soy protein fractions. Journal of Food Science, 52, 98-101. Owusu-Apenten, R. (2005). Colorimetric analysis of protein sulfhydryl groups in milk: Applications and processing effects. Critical Reviews in Food Science and Nutrition, 45, 123. 51  Pelton, J. T., and McLean, L . R. (2000). Spectroscopic methods for analysis o f protein secondary structure. Analytical Biochemistry, 277, 167-176. Petruccelli, S., and Anon, M . C . (1995). Partial reduction o f soy protein isolate disulfide bonds. Journal of Agricultural and Food Chemistry, 43, 2001-2006. Pollien, P., Ott, A . , Montigon, F., Baumgartner, M . , Munoz-Box, R., and Chaintreau, A . (1997). Hyphenated headspace-gas chromatography-sniffing technique: screening o f impact odorants and quantitative aromagram comparisons. Journal of Agricultural and Food Chemistry, 45(7), 2630-2637. Prendergast, K . (1974). Protein hydrolysate, a review. Food Trade Review, 44(1), 14, 16-21. Puppo, E . , Chapleau, N . , Speroni, F, de Lamballerie-Anton, M . , Michel, F., Anon, C , and Anton, M . (2004). Physicochemical modifications o f high-pressure-treated soybean protein isolates. Journal of Agricultural and Food Chemistry, 52, 1564-1571. Pyke, M . (1975). Processing and flavoring of proteins: a taste for meat. International Flavors and Food Additives, 6, 275-277. Rah, J. H . , Hasler, C . M . , Painter, J. E . , and Chapman-Novakofski, K . M . (2004). Applying the theory o f planned behavior to women's behavioral attitudes on and consumption of soy products. Journal of Nutrition Education Behavior, 36, 238-244. Rao, A . , Shallo, H . E . , Ericson, A . P., and Thomas, R. L . (2002). Characterization o f soy protein concentrate produced by membrane ultrafiltration. Journal of Food Science, 67, 1412-1418. Reipa, V . (2005). Infrared and Raman spectroscopy characterization o f biological products. In A . R. Mire-Sluis, State of the Art Analytical Methods for the Characterization of Biological Products and Assessment of Comparability, (pp. 85-91). Basel, Switzerland: S. Karger A G . Remondetto, G . E . , Gonzalez, R . J., and Anon, M . C . (2002). Effects o f simultaneous heat and reduction treatments on some structural characteristics o f soy isolates obtained at pilot plant scale. Food Science and Technology International, 8, 223-228. Renkema, J. M . S. and van Vliet, T. (2002). Heat-induced gel formation by soy proteins at neutral p H . Journal of Agricultural and Food Chemistry, 50, 1569-1573. Rickert, D . A . , Johnson, L . A . , and Murphy, P. A . (2004). Functional properties o f improved glycinin and P-conglycinin fractions. Journal of Food Science, 69, 303-311. Rodriguez-Bencomo, J. J., Conde, J. E . , Garcia-Montelongo, F., and Perez-Trujillo, J. P. (2003). Determination o f major compounds in sweet wines by headspace solid-phase microextraction and gas chromatography. Journal of Chromatography A, 991, 13-22.  52  Scatchard, G. (1949). The attractions of proteins for small molecules and ions. Annals of the New York Academy Sciences, 51, 660-672. Schutte, L . and Van den Ouweland, G. (1979). Flavor problems in the application of soy protein materials. The Journal of the American Oil Chemists' Society, 56, 289-290. Sessa, D. J., and Rackis, J. J. (1977). Lipid-derived flavors of legume protein products. The Journal of the American Oil Chemists 'Society, 54, 468-473. Shahidi, F., Rubin, L. J., and D'Souza, L. A . (1986). Meat flavor volatiles: A review of the composition, techniques of analysis, and sensory evaluation. CRC Critical Reviews in Food Science and Nutrition, 24(2), 141-243. Shao, Y . , Marriott, P., Shellie, R., and Hugel, H . (2003). Solid-phase micro-extraction comprehensive two-dimensional gas chromatography of ginger (Zingiber officinale) volatiles. Flavour and Fragrance Journal, 18, 5-12. Simonsky, R. W., and Stanley, D. W. (1982). Texture-structure relationships in textured soy protein. V . Influence of pH and protein acylation on extrusion texturization. Canadian Institute of Food Science and Technology Journal, 15(4), 294-301. Smith, A . K., and Circle, S. J. (1978). Protein products as food ingredients. In A . K. Smith, Soybeans: Chemistry and Technology, (pp. 339-388). Connecticut: The A V I Publishing Company, Inc. Steenson, D. F., Lee, J. H., and Min, D. B. (2002). Solid phase microextraction of volatile soybean oil and corn oil compounds. Journal of Food Science, 67(1), 71-76. Sung, H. Y., Chen, H. J., Liu, T. Y., and Su, J. C. (1983). Improvement of the functionality of soy protein by introduction of new thiol groups through a papain-catalyzed acylation. Journal of Food Science, 48(3), 708-711. Thanh, V . H . and Shibasaki, K . (1979). Major proteins of soybean seeds. Reversible and irreversible dissociation of P-conglycinin. Journal of Agricultural and Food Chemistry, 27, 805-809. Thannhauser, T. W., Konishi, Y., and Scheraga, H. A. (1984). Sensitive quantitative analysis of disulfide bonds in polypeptides and proteins. Analytical Biochemistry, 138, 181-188. Tsumura, K., Saito, T., Kugimiya, W., and Inouye, K . (2004). Selective proteolysis of the glycinin and P-conglycinin fractions in a soy protein isolate by pepsin and papain with controlled pH and temperature. Journal of Food Science, 69(5), C363-C367. U.S. Food and Drug Administration. (1999). Food labeling, health claims, soy protein and coronary heart disease. Federal Regulation, 57, 699-733.  53  Utsumi, S., Damodaran, S., and Kinsella, J. E . (1984). Heat-induced interactions between soybean proteins: Preferential association of 11S basic subunits and p subunits o f 7S. Journal of Agricultural and Food Chemistry, 32, 1406-1412. Utsumi, S., Matsumura, Y , and M o r i , T. (1997). Structure-function relationships o f soy proteins. In S. Damodaran and A . Paraf, Food Proteins and Their Applications, (pp. 257-291). New York: Marcel Dekker Inc. Warner, K . , Mount, S. T. L . , Rackis, J. J., and Wolf, W. J. (1983). Relationship o f sensory characteristics and gas chromatographic profiles o f soybean protein products. Cereal Chemistry, 60, 102-106. Williams, R. W., and Dunker, A . K . (1981). Determination o f the secondary structure o f proteins from the amide I band o f the laser Raman spectrum. Journal of Molecular Biology, 152, 783-813. Willson, R. A . (1975). A review o f thermally produced imitation meat flavours. Journal of Agricultural and Food Chemistry, 23, 1032-1037. Yamauchi, F., Yamagishi, T., and Iwabuchi, S. (1991). Molecular understanding o f soybean protein. Food Reviews International, 7, 283-322. Yang, X . , and Peppard, T. (1994). Solid-phase microextraction for flavor analysis. Journal of Agricultural and Food Chemistry, 42(9), 1925-1930. Zhang, Z . , and Pawliszyn, J. (1993). Headspace Chemistry, 65(14), 1843-1852.  solid-phase microextraction. Analytical  Zhang, Z . , Yang, M . J., and Pawliszyn, J. (1994). Solid-phase microextraction. Analytical Chemistry, 66, 844A-853A. Zhao, Y . , M a , C . - Y . , Yuen, S. - N . , and Phillips, D . L . (2004a). Study o f acetylated food proteins by Raman spectroscopy. Journal of Food Science, 69, FCT206-FCT213. Zhao, Y . , M a , C . - Y . , Yuen, S. - N . , and Phillips, D . L . (2004b). Study o f succinylated food proteins by Raman spectroscopy. Journal of Agricultural and Food Chemistry, 52, 18151823. Zhou, Q., and Cadwallader, K . R. (2004). Inverse gas chromatographic method for measurement of interactions between soy protein isolate and selected flavor compounds under controlled relative humidity. Journal of Agricultural and Food Chemistry, 52, 6271-6277'. Zhou, Q., Lee, S.-Y., and Cadwallader, K . R. (2006). Inverse gas chromatographic evaluation o f the influence o f soy protein on the binding o f selected butter flavor compounds in a wheat soda cracker system. Journal of Agricultural and Food Chemistry, 54, 5516-5520.  54  CHAPTER  2.  DEVELOPMENT  OF  SOLID-PHASE  MICROEXTRACTION  METHODOLOGY FOR ANALYSIS OF HEADSPACE VOLATILE COMPOUNDS IN SIMULATED BEEF FLAVOUR  1  2.1. Introduction  A characteristic beef odour is o f prime importance to the quality o f processed beef products as well as beef analogue products such as vegetable protein based meat substitutes for vegetarian consumers.  Among the vegetable protein materials, soy protein has become especially popular  as meat substitutes over the past few years due to the reported hypocholesterolemic effects of soy protein resulting in reducing risks o f cardiovascular disease (Anderson et al., 1995).  However,  flavour problems have been a major technical impediment to the increased usage of soy proteins in human foods (Maheshwari et al., 1995).  Many studies have been reported on flavour of soy  protein products with respect to the "beany" odour (Boatright and L e i , 1999; L e i and Boatright, 2001; W o l f 1975), and on reducing the off-flavour (Inouye et al., 2002; Maheshwari et al., 1995; McDaniel and Chan, 1988).  In addition to indigenous undesirable soy aroma components that  are difficult to remove, soy proteins may interact with desirable components of added flavour formulations such as simulated beef flavour. The presence o f soy protein in aqueous systems has been reported to increase the retention of volatile components in samples (Gremli 1974), and suppression of chicken flavour in a formulated soup at high levels o f soy protein has also been reported (Malcolmson and McDaniel, 1987).  Moreover, a significant reduction in perceived  flavour intensity was observed when simulated beef flavours were added at the supplier's recommended dosage to formulate soy protein based beef substitutes (Yves Veggie Cuisine, 2002, personal communication).  However, there is little information on the specific underlying  mechanism for these observations, which could provide potential solutions for the problem.  To  elucidate the interaction between soy protein and simulated beef flavours, it is crucial to use a sensitive and reproducible but simple method for the beef flavour analysis.  A version of this chapter has been published. Moon, S. - Y . , and Li-Chan, E . C . Y . (2004). Development of solid-phase microextraction methodology for analysis o f headspace volatile compounds in simulated beef flavour. Food Chemistry, 88:141-149. 1  55  Headspace solid phase microextraction (HS-SPME) is a sample preparation technique that has been gaining popularity in recent years.  Volatiles are captured by S P M E based on the theory of  equilibrium partitioning of the analytes between extraction medium i.e. the solid phase of S P M E and liquid or solid sample matrix (Zhang and Pawliszyn, 1993), and can be analyzed by gas chromatography ( G C ) .  Traditional methods for the extraction and concentration of volatile  compounds for analysis by G C such as liquid-liquid extraction, solid phase extraction, supercritical fluid extraction, static headspace sampling or dynamic headspace (purge-and-trap) methods have one or more drawbacks such as high cost, multi-step preparation, low sensitivity, prolonged extraction time, artifact formation or solvent contamination (Braggins et al., 1999). Compared to the previous methods, the S P M E technique has been regarded as a simple, rapid, and economical method requiring no solvent (Yang and Peppard, 1994).  Although the S P M E  technique was originally developed for analysis of pollutants in environmental water samples (Arthur et al., 1992), it has also now been applied to flavour analysis, by employing headspace S P M E sampling in various food commodities such as cheese (Lecanuet al., 2002), edible oils (Steenson et al., 2002), coffee (Akiyama et al., 2003), ginger (Shao et al., 2003), kimchi (Lee et al., 2003b), and sweet wines (Rodriguez-Bencomo et al., 2003).  However, it has also been reported that the S P M E analysis is drastically affected by several factors such as the nature o f the S P M E solid phase, adsorption time, adsorption temperature, salt addition, stirring condition, and sample size (Lee et al., 2003b; 2002; Steenson et al., 2002).  Rodriguez-Bencomo et al.,  Generally 'one factor at a time' experiments have been  conducted in most o f the previous studies to determine the analysis conditions with S P M E , but "one factor at a time' designs often overlook interactions among the factors.  In contrast,  L i u and Yang (2002) used response surface methodology ( R S M ) coupled with a two-factor central composite rotatable design to study the interaction between adsorption time and adsorption temperature.  However, R S M is not the best choice when dealing with multiple  factors due to the large number o f experiments required.  In order to investigate the effects of  multiple factors as well as potential interactions between these factors in a time and cost effective manner, fractional factorial design based on Taguchi's orthogonal array can be considered (Arteaga et al., 1994).  To date, there has been no report on important factors for optimum  adsorption condition to analyze headspace volatile compounds in simulated beef flavour. 56  A s a result, Taguchi's fractional factorial experimental design was proposed in this study to screen significant factors that would have a great impact on adsorption condition for headspace analysis o f simulated beef flavour.  Therefore, the objectives o f this study were to investigate the conditions that may influence H S S P M E o f volatile compounds from simulated beef flavour by means o f the Taguchi's method, and to investigate the effects o f the factors along with potential interactions between these factors. The achievement o f a sensitive and reproducible analytical method that is able to monitor changes in the headspace composition of the flavour sample would be the basis o f future research to elucidate the mechanism o f soy protein - flavour interactions that may be responsible for suppression of perceived intensity o f beef flavour in soy protein products.  2.2. Experimental methods 2.2.1. Materials The simulated beef flavour (SBF) used in this study was a commercially produced blended flavour ("vegetarian  beef type flavour F96x49" from Mastertaste (Arlington Heights, IL),  containing maltodextrin, autolyzed yeast extract, natural flavours, onion powder and silicon dioxide). Its beef character is primarily derived from Maillard reaction during roasting o f the materials and it has been used in the industry to provide beefy flavour in vegetarian products. The solid  phase  assembly  holder  polydimethylsiloxane/divinylbenzene  and  four  commercially  (PDMS/DVB),  65  available pm  fibres,  65  pm  carbowax/divinylbenzene  ( C W / D V B ) , 75 pm carboxen/polydimethylsiloxane ( C A R / P D M S ) , and 50/30 pm stablefiex divinylbenzene/carboxen/polydimethylsiloxane  (DVB/CAR/PDMS)  Supelco (Sigma-Aldrich Canada, Oakville, ON).  were  purchased  from  Before use, each fibre was exposed to a  splitless/split injection port under helium flow and conditioned for the recommended time at recommended temperatures according to the manufacturer's instruction (i.e. at 250 °C 30 minutes for P D M S / D V B , at 220 °C 30 minutes for C W / D V B at 300 °C 120 minutes for C A R / P D M S , and at 270 °C 180 minutes for D V B / C A R / P D M S ) . G C sample vials with 15 m L capacity and polypropylene hole cap with PTFE/silicone septa were purchased from Supelco.  57  2.2.2. Fractional factorial design A Taguchi's L 7 ( 3 ) orthogonal array was used to evaluate potentially significant factors 13  2  affecting the adsorption o f S B F compounds onto S P M E fibres (Table 2.1).  The main effects  of 4 factors were investigated: adsorption temperature ( A ) , adsorption time (B), salt concentration ( C ) , and S P M E phase (D).  The 3 levels selected for each factor were 30 °C,  45 ° C , and 60 °C for adsorption temperature, 20, 40, and 60 minutes for adsorption time, 0, 3, and 6 % o f added sodium chloride concentration, and 65 pm P D M S / D V B , 65 pm C W / D V B , and 50/30 pm D V B / C A R / P D M S for S P M E phase.  In  addition to the  4 main factors,  3 interactions were also investigated: adsorption  temperature x adsorption time ( A x B ) , adsorption temperature x salt concentration ( A x Q , and adsorption temperature x S P M E phase ( A x D ) .  The linear graph used for this experimental  design is illustrated in Figure 2.1 and the column assignment and the levels o f each factor are shown in Table 2.2.  In this design, columns 1,2, 5, and 8 were assigned to the main factors  ( A , B , C , and D ) .  Columns 3+4, 6+7, and 9+10 were used to investigate three o f the  possible interactions between these factors, while columns 11, 12, and 13 were employed for the estimation o f the error term. The levels o f each factor were selected within the applicable range o f the processing for commercial meat substitutes (Ross, 1996) based on the feedback by a development staff at a food company in Vancouver.  2.2.3. Solid-phase microextraction procedure Flavour stock solution consisting o f 10 g of SBF and 40 g o f 30 m M T r i s - H C l buffer (pH 6.0) was prepared fresh daily and 2.5 g of flavour stock solution and 3 g o f buffer solution were mixed in a 15 m L capacity G C sampling vial with a magnetic stirring bar to provide a flavour concentration similar to that o f commercial beef flavour products.  To examine the salt effect, 0  g, 0.165 g, or 0.330 g N a C l was added to give 0 % , 3.0 % , and 6 % (w/w) added salt concentration, respectively.  Since the S B F originally had 20 % salt, this resulted in 1.8 % ,  4.8 % and 7.8 % final salt concentration in the samples. The range o f salt levels o f beef substitute products  (500-800 mg/100 g o f finished product) was considered to select these salt  concentrations. The vial was tightly capped with a polypropylene hole cap with a PTFE/silicone septum. The S P M E  fibre (either 65 pm P D M S / D V B , 65 pm C W / D V B , or 50/30 pm 58  D V B / C A R / P D M S ) was exposed to the headspace above the sample solution for 20, 40, or 60 minutes at 30 °C, 45 ° C , or 60 °C according to the experimental design shown in Table 2.2. Stirring with a magnetic stirring bar was consistently applied for all samples.  2.2.4. Optimization of adsorption time To study the effect o f adsorption time at 60 °C on S P M E by 50/30 pm D V B / C A R / P D M S fibre, samples, as prepared as in section 2.2.3, were maintained for 20, 40, 60, 80, 100 or 120 minutes with stirring.  A l l treatments were made in triplicate, and results are expressed as the average  and standard deviation values.  2.2.5. Gas chromatography A Hewlett-Packard 5890 gas chromatograph with flame ionization detector (FID) and a D B - 5 analytical fused silica capillary column (30 m x 0.32 mm x 0.25 um film thickness from J & W Scientific, Folsom, C A ) were used for analysis o f the volatile compounds. conducted in a splitless mode for 3 min at 250 °C.  The injection was  The oven temperature was held at 40 °C for  3 min, ramped to 180 °C at the rate of 3 °C/min and then to 260 °C at 10 °C/min, and maintained at 260 °C for 2 min.  Helium was used as a carrier gas at a column-head pressure o f 12 p.s.i. (1  p.s.i.=6894.76 Pa). The temperature o f the FID detector was 280 °C, and 35 mL/min of hydrogen, 350 mL/min o f air and 30 mL/min o f helium as a make-up gas were used.  T o increase  reliability in terms o f number o f peaks and peak area, several parameters in the chromatogram were set.  Peak width was set to 0.04 and initial threshold was set to 1.  The peaks with peak  area under 10,000 were not regarded as reliable peaks.  A 0.75 mm I.D. inlet liner was employed to minimize broadening effect, and resulted in decreased peak width compared to a 2.0 mm injection glass liner.  The type o f injection mode  (splitless/split), desorption time (0.5 - 3 min) and desorption temperature (220 °C - 260 °C) along with oven temperature programs were also tested to improve peak shape and sensitivity while reducing carry-over from the previous analysis.  The detection o f the analytes was improved  using splitless mode, 3 minutes as desorption time, and 250 °C as desorption temperature.  59  2.2.6. Statistical analysis General linear model of analysis of variance ( A N O V A ) was performed by using Minitab (version 13.30, Minitab inc. P A U S A ) to analyze the significance o f the 4 main factors, namely, adsorption temperature ( A ) , adsorption time (B), salt concentration (C), and S P M E phase (D), and 3 interactions between the factors ( A * B , A * C , and A x D ) .  Tukey's multiple-range test was  conducted for comparison o f mean values o f the data at the 95 % confidence level.  2.3. Results and Discussion  2.3.1. Screening of significant factors on the headspace analysis of the beef flavour The F I D responses in terms o f total area and number of peaks for the chromatograms of each of the 27 experimental runs are shown in Table 2.2. Typical total ion chromatograms (TICs) are illustrated in Figure 2.2, which demonstrates the variation in both total peak area and number of peaks in the TICs o f the headspace volatile compounds in S B F under three different adsorption conditions corresponding to experiment numbers 25, 26, and 27 (Table 2.2).  Table 2.3 shows the results o f A N O V A for significance of the main factors and the selected interactions between the factors, on the total peak area and the number o f peaks of headspace volatile compounds in S B F .  For both total peak area and number of peaks, adsorption  temperature and adsorption time were significant factors (p < 0.05) whereas S P M E phase was highly significant (p < 0.001).  Salt concentration did not have a significant effect (p > 0.05)  within the studied range.  The number o f interaction effects that can be examined is limited by the number of columns in a Taguchi's fractional factorial design, and interactions that are higher than second order are assumed not to be important.  In this experiment, in addition to the 4 main factors, 3 selected  interactions were investigated, which were adsorption temperature x adsorption time, adsorption temperature x salt concentration, and adsorption temperature x S P M E phase.  These interactions  were selected based on previous research results demonstrating that adsorption temperature was one of the most important factors affecting adsorption efficiency by S P M E (del Castillo and Dobson, 2002; Diaz et al., 2002; L i u et al., 2002; Steenson et al., 2002).  None o f the selected 60  interactions was statistically significant at the 5 % significance level.  2.3.2. Effects of SPME fibres Based on the response o f the total peak area obtained in preliminary experiments (data not shown), 3 types o f S P M E fibres (i.e. 65 pm P D M S / D V B , 65 pm C W / D V B , and 50/30 pm D V B / C A R / P D M S ) were chosen from the 4 fibres mentioned above to conduct the fractional factorial experiment.  Among the 3 S P M E fibres tested, 50/30 pm D V B / C A R / P D M S fibre  coating showed remarkably higher signal response (p < 0.001) than the other fibres for both total area and number o f peaks (Figure 2.3).  Figure 2.2 also illustrates the larger total peak area and  number of peaks obtained using the 50/30 pm D V B / C A R / P D M S fibre.  This fibre has improved  stability o f coating materials compared to other commercially available fibres (Supelco catalog 2003/2004). A coefficient o f variation o f 2.9 % was observed for five replicate analyses o f the total area counts of the headspace volatiles in beef flavour, using the 50/30 pm D V B / C A R / P D M S fibre for 60 min at 60 °C.  In addition, no considerable carry-over from the previous extraction  was observed after 3 min desorption in the injector.  Therefore, 50/30 pm D V B / C A R / P D M S  fibre was selected as the most suitable fibre due to the high sensitivity with good reproducibility as well as durable property of the fibre.  2.3.3. Effects of salt concentration The addition o f salt at concentrations up to 6 % (which is equivalent to 7.8 % o f total salt concentration considering the indigenous salt content o f the flavour) did not significantly affect either total peak area or the number of peaks of the beef flavour samples as shown in Table 2.3. Generally, the presence o f salt has been reported to stimulate adsorption o f the volatile components from samples by changing the phase border properties and decreasing the solubility of hydrophobic components in the aqueous phase, the so called "salting out" effect (Yang et al., 1994).  However, a salt concentration of 20-30 % (w/v) was suggested to be required to yield an  adsorption effect for most flavour compounds (Lee et al., 2003a).  L i u et al. (2002) reported that  low concentrations o f salt in the range o f 0 % to 10 % did not affect adsorption efficiency o f volatiles such as ethyl isovalerate and isoamyl acetate while adsorption o f both compounds significantly increased at 20 % salt.  In our study, an upper level o f up to 6 % o f sodium  chloride was added since higher salt concentration would not be relevant in the context of the 61  type o f food systems under consideration, i.e. beef flavoured vegetable products.  High  concentration of salt is also known to stimulate denaturation o f proteins (Cheftel et al., 1985).  2.3.4. Effects of adsorption temperature A s adsorption temperature increased from 45 °C to 60 °C, both total peak area (p < 0.1) as well as the number of total peaks (p < 0.05) in the G C chromatograms o f the samples increased while there was no significant difference between 30 °C to 45 °C (Figure 2.3). The increases o f volatile compounds can be explained by the fact that higher temperature tends to drive the volatiles from the liquid phase to the gas phase.  Diaz et al. (2002) also demonstrated that the  adsorption efficiency o f brominated analogs in water increased by increasing the temperature up to 50 °C and Steenson et al. (2002) found an increase o f volatile compounds in vegetable oils with higher extraction temperature.  In this study, 60 °C was proposed as the adsorption  temperature for the beef flavour due to the high sensitivity o f both total area counts and number of peaks.  Furthermore, beef analogue products such as vegetarian hot dogs are frequently  served at temperatures close to 60 °C (Yves Veggie Cuisine, 2002, personal communication).  2.3.5. Effects of adsorption time Table 2.3 shows adsorption time was also one o f the significant factors in the analysis o f headspace volatile compounds o f the beef flavour. The effect o f adsorption time on the analysis of headspace volatile compounds is illustrated in Figure 2.3. Even though there were no statistically significant differences between the responses at 20 min and 40 min or at 40 min and 60 min, significant difference was found between 20 min and 60 min (p < 0.05) in total peak area in addition to the number o f total peaks.  The increase o f volatile compounds with adsorption  time was also reported by other research groups (Steenson et al., 2002; Tombesi and Freije, 2002). In contrast, Lee et al. (2003b) reported that the isolation o f volatile compounds in kimchi adsorbed onto S P M E increased with longer time up to 30 min, while it tended to decrease at times longer than 30 min depending on the adsorption temperature.  Since there was no interaction between factors, subsequent experiments were conducted using 50/30 pm D V B / C A R / P D M S fibre at 60 °C to investigate the adsorption time required to reach maximum total area responses o f the samples. Figure 2.4 shows the effect o f adsorption time at 62  60 °C by 50/30 um D V B / C A R / P D M S fibre on the headspace volatile compounds of the beef flavour.  The total area o f volatile compounds was increased with longer adsorption times up to  80 min (p < 0.05) when equilibrium between the headspace o f the beef flavour sample and the S P M E fibre was reached.  However, it was suggested that the S P M E sampling time should be  no longer than the total G C run time for maximum productivity, because good precision can be accomplished without achieving equilibrium as long as the adsorption time is controlled accurately (Penton, 1999).  Therefore, the adsorption time of 60 min was selected, considering a  good reflection of transferring o f volatile compounds from sample solution to headspace, a suitable analysis time for the routine analysis, and a low coefficient o f variation (2.9 %).  2.4. Conclusion  This study has demonstrated the application of a fractional factorial experimental design based on Taguchi's orthogonal array for screening the significant factors in S P M E for analysis o f the headspace volatile compounds in S B F . Out of 4 main factors and 3 potential interactions researched, adsorption temperature, adsorption time, and S P M E phase were the factors that significantly affect the headspace analysis of the S B F in terms o f the total peak area and the number of peaks. The selected adsorption conditions for S P M E were adsorption by 50/30 pm D V B / C A R / P D M S fibre for 60 minutes at 60 °C.  The establishment o f this reproducible and  representative analytical method by H S - S P M E paves the way for ongoing research to monitor changes in the headspace composition o f the S B F , to elucidate the basis for the interactions between flavour compounds and soy protein in vegetarian food products.  63  Figure 2.1. The standard linear graph of orthogonal array L ( 3 ) used in this study. 1 3  2 7  64  ,3 . O  «2 Exp N o . 25  S3 . O «  1  -1-  f ^..J  1.  . . . . . , A _ . . J . - I .. ,.L  i  l  l  ^JaJlAjjU-J,  c 3  O o  a  o o  ID  Exp N o . 27  . l : ; ; O e 4 j  A . I  ISO  : 3 0  4 0  I.  .  t  li-  o o  Time (min)  Figure 2.2. Total ion chromatograms (TICs) of headspace volatile compounds in simulated beef flavour using the S P M E adsorption conditions described in Table 2.2 for experiment number 25, 26, and 27. 'Experiment numbers correspond to those of Table 2.2.  500000  Temp  Time  Salt  SPME  Figure 2.3. Effects of adsorption temperature, adsorption time, salt concentration, and SPME phase on (a) means of total area counts and (b) means of the number of peaks obtained in the analysis of headspace volatile compounds in simulated beef flavour. Different letters (a and b) within each plot indicate significant difference (p < 0.05). 'Levels 0, 1, and 2 of the SPME phase refer to 50/30pm DVB/CAR/PDMS, 65pm PDMS/DVB, and 65pm CW/DVB, respectively. different at 10 % significance level (p = 0.0571). 66  140 Time  (min)  Figure 2.4. Effect o f adsorption time on the headspace volatile analysis of simulated beef flavour by 50/30pm D V B / C A R / P D M S S P M E at 60 °C. Points are averages from triplicate analyses and error bars are ± standard deviation. 'Different letters indicate significant difference (p < 0.05).  67  Table 2.1. Column assignment for the 4 factors and 3 interactions in this study based on Taguchi's L27(3 ) orthogonal array. 13  Column number Fyn  JNo. L-/A.L/. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27  Exp Order  17 19 24 23 2 27 18 8 7 1 20 22 25 5 6 13 12 26 9 3 14 15 21 11 16 4 10  1  A  0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 2 2 2 2 2 2 2 2 2  2  B  0 0 0 1 1 1 2 2 2 0 0 0 1 1 1 2 2 2 0 0 0 1 1 1 2 2 2  3  4  A  A  X  X  B  B  0 0 0 1 1 1 2 2 2 1 1 1 2 2 2 0 0 0 2 2 2 0 0 0 1 1 1  0 0 0 1 1 1 2 2 2 2 2 2 0 0 0 1 1 1 1 1 1 2 2 2 0 0 0  5  C  0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2  6  7  A  A  X  X  C  C  0 1 2 0 1 2 0 1 2 1 2 0 1 2 0 1 2 0 2 0 1 2 0 1 2 0 1  0 1 2 0 1 2 0 1 2 2 0 1 2 0 1 2 0 1 1 2 0 1 2 0 1 2 0  8  D  0 1 2 1 2 0 2 0 1 0 1 2 1 2 0 2 0 1 0 1 2 1 2 0 2 0 1  9  10  A  A  X  X  D  D  0 1 2 1 2 0 2 0 1 1 2 0 2 0 1 0 1 2 2 0 1 0 1 2 1 2 0  0 1 2 1 2 0 2 0 1 2 0 1 0 1 2 1 2 0 1 2 0 2 0 1 0 1 2  11  12  13  e  e  e  0 1 2 2 0 1 1 2 0 0 1 2 2 0 1 1 2 0 0 1 2 2 0 1 1 2 0  0 1 2 2 0 1 1 2 0 1 2 0 0 1 2 2 0 1 2 0 1 1 2 0 0 1 2  0 1 2 2 0 1 1 2 0 2 0 1 1 2 0 0 1 2 1 2 0 0 1 2 2 0 1  A: adsorption temperature (30 °C, 45 °C, and 60 °C as levels 0, 1, and 2) B: adsorption time (20 min, 40 min, and 60 min as levels 0, 1, and 2) C: salt concentration (0 %, 3 %, and 6 % as levels 0, 1, and 2) D: SPME phase (50/30pm DVB/CAR/PDMS, 65pm PDMS/DVB, and 65pm CW/DVB as levels 0, 1, and 2) e : error term 68  Table 2.2. Experimental design based on Taguchi's L 7 ( 3 ) orthogonal array and the measured 13  2  responses of total area count and the number of peaks in the gas chromatogram for each experimental run. Factors  Exp.  Responses  1  No.  A  B  C  D  Total Area  N o . o f Peaks  1  30  20  0  DVB/CAR/PDMS  228801  9  2  30  20  3  PDMS/DVB  36325  1  30013  1  3  30  20  6  CW/DVB  4  30  40  0  PDMS/DVB  68431  1  5  30  40  3  CW/DVB  63933  2  6  30  40  6  DVB/CAR/PDMS  369733  15  7  30  60  0  CW/DVB  43944  1  8  30  60  3  DVB/CAR/PDMS  383206  14  9  30  60  6  PDMS/DVB  143972  5  10  45  20  0  DVB/CAR/PDMS  236792  11  11  45  20  3  PDMS/DVB  47968  1  12  45  20  6  CW/DVB  37258  2  87752  2  27000  1  13  45  40  0  PDMS/DVB  14  45  40  3  CW/DVB  15  45  40  6  DVB/CAR/PDMS  432501  19  24796  1  16  45  60  0  CW/DVB  17  45  60  3  DVB/CAR/PDMS  524510  22  154745  6  287959  13  18  45  60  6  PDMS/DVB  19  60  20  0  DVB/CAR/PDMS  20  60  20  3  PDMS/DVB  78910  3  21  60  20  6  CW/DVB  33728  2  22  60  40  0  PDMS/DVB  156345  8  23  60  40  3  CW/DVB  60726  3  24  60  40  6  DVB/CAR/PDMS  777677  35  25  60  60  0  CW/DVB  79574  5  26  60  60  3  DVB/CAR/PDMS  1089718  45  27  60  60  6  PDMS/DVB  271640  13  'Factors were assigned to columns as shown in Table 2.1, where A : adsorption temperature (°C), B : adsorption time (min), C : salt concentration (%), and D : S P M E phase 2  Area reject: 10,000, initial threshold: 1 and peak width: 0.04 69  Table 2.3. Analysis o f variance o f the main factors and the selected interactions between the factors on the total peak area and the number of peaks of headspace volatile compounds in simulated beef flavour.  Source  Total peak area  Degree o f  N o . of Peaks  Freedom  F-value  Probability  F-value  Probability  A  2  5.14  0.050*  9.13  0.015*  B  2  5.95  0.038*  6.53  0.031*  C  2  3.10  0.119  3.52  0.098  D  2  36.13  0.000***  43.15  0.000***  AxB  4  1.05  0.455  1.53  0.304  AxC  4  0.56  0.702  0.43  0.780  AxD  4  2.46  0.156  2.82  0.124  error  6  -  -  -  -  *, **, and *** significant at p < 0.05, p < 0.01, and p < 0.001 respectively. A : adsorption temperature, B : adsorption time, C : salt concentration, and D : S P M E phase.  70  2.5. References  Akiyama, M . , Murakami, K . , Ohtani, N . , Iwatsuki, K . , Sotoyama, K . , Wada, A . , Tokuno, K . , Iwabuchi, H . , and Tanaka, K . (2003). Analysis of volatile compounds released during the grinding o f roasted coffee beans using solid-phase microextraction. Journal of Agricultural and Food Chemistry, 51(7), 1961-1969. Anderson, J. W., Johnstone, B . M . , and Cook-Newell, M . L . (1995). Meta-analysis of the effects o f soy protein intake on serum lipids. The New England Journal of Medicine, 333(5), 276-282. Arteaga, G . E . , Li-Chan, E . , Vazquez-Arteaga, M . C , and Nakai, S. (1994). Systematic experimental designs for product formula optimization. Trends in Food Science and Technology, 5, 243-254. Arthur, C . L . , K i l l a m , L . M . , Motlagh, S., L i m , M . , Potter, D . W., and Pawliszyn, J. (1992). Analysis of substituted benzene compounds in groundwater using solid-phase microextraction. Environmental Science and Technology, 26, 979-983. Boatright, W . L . , and L e i , Q. (1999). Compounds contributing to the "beany" odor of aqueous solutions o f soy protein isolates. Journal of Food Science, 64(4), 667-670. Braggins, T. J., Grimm, C . C , and Visser, F. R. (1999). Analysis o f food volatiles using S P M E . In J. Pawliszyn, Applications of Solid Phase Microextraction, (pp. 407-422). Cambridge: The Royal Society o f Chemistry. Cheftel, J. C , Cuq, J., and Lorient, D . (1985). Amino acids, peptides, and proteins. In O. R. Fennema, Food Chemistry, (pp. 245-369). N e w York: Marcel Dekker Inc. del Castillo, M . L . R., and Dobson, G . (2002). Varietal differences in terpene composition of blackcurrant (Ribes nigrum L) berries by solid phase microextraction/gas chromatography. Journal of the Science of Food and Agriculture, 82, 1510-1515. Diaz, A . , Ventura, F., and Galceran, M . T. (2002). Development o f a solid-phase microextraction method for the determination of short-ethoxy-chain nonylphenols and their brominated analogs in raw and treated water. Journal of Chromatography A, 963, 159-167. Gremli, H . A . (1974). Interaction of flavor compounds with soy protein. The Journal of the American Oil Chemists' Society, 51, 95A-97A. Inouye, K . , Shiihara, M . , Uno, T., and Takita, T. (2002). Deodorization o f soybean proteins by enzymatic and physicochemical treatments. Journal of Agricultural and Food Chemistry, 50(6), 1652-1658.  71  Lecanu, L . , Ducruet, D . , Jouquand, C , Gratadoux, J. J., and Feigenbaum, A . (2002). Optimization o f headspace solid-phase microextraction ( S P M E ) for the odor analysis of surface-ripened cheese. Journal of Agricultural and Food Chemistry, 50(13), 38103817. Lee, J. FL, Diono, R., K i m , G . Y . , and M i n , D . B . (2003a). Optimization of solid phase microextraction analysis for the headspace volatile compounds o f parmesan cheese. Journal of Agricultural and Food Chemistry, 51(5), 1136-1140. Lee, J. H . , Kang, J. H . , and M i n , D . B . (2003b). Optimization of solid-phase microextraction for the analysis o f headspace volatile compounds in kimchi, a traditional Korean fermented vegetable product. Journal of Food Science, 68(3), 844-848. Lei, Q., and Boatright, W . L . (2001). Compounds contributing to the odor of aqueous slurries of soy protein concentrate. Journal of Food Science, 66(9), 1306-1310. L i u , T., and Yang, T. (2002). Optimization of solid-phase microextraction analysis for studying change o f headspace flavor compounds o f banana during ripening. Journal of Agricultural and Food Chemistry, 50, 653-657. Maheshwari, P., O o i , E . T., and Nikolov, Z . L . (1995). Off-flavor removal from soy-protein isolate by using liquid and supercritical carbon dioxide. The Journal of the American Oil Chemists' Society, 72(10), 1107-1115. Malcolmson, L . J., and McDaniel, M . R. (1987). Flavor protein interactions in a formulated soup containing flavored soy protein. Canadian Institute of Food Science and Technology, 20(4), 229-235. McDaniel, M . R., and Chan, N . (1988). Masking o f soy protein flavor by tomato sauce. Journal of Food Science, 53(1), 93-101. Penton, Z . (1999). Method development with solid phase microextraction. In S. A . S. Wercinski, Solid Phase Microextraction - A Practical Guide, (pp. 27-57). N e w York: Marcel Dekker, Inc. Rodriguez-Bencomo, J. J., Conde, J. E . , Garcia-Montelongo, F., and Perez-Trujillo, J. P. (2003). Determination o f major compounds in sweet wines by headspace solid-phase microextraction and gas chromatography. Journal of Chromatography A, 991, 13-22. Rodriguez-Bencomo, J. J., Conde, J. E . , Rodriguez-Delgado, M . A . , Garcia-Montelongo, F., and Perez-Trujillo, J. P. (2002). Determination of esters in dry and sweet white wines by headspace solid-phase microextraction and gas chromatography. Journal of Chromatography A, 963, 213-223. Ross, P. J. (1996). The design o f experiments process. In Taguchi Techniques for Quality Engineering 2 edition, (pp. 23-41). New York: M c G r a w - H i l l . n d  72  Shao, Y . , Marriott, P., Shellie, R., and Hugel, H . (2003). Solid-phase micro-extraction comprehensive two-dimensional gas chromatography of ginger (Zingiber officinale) volatiles. Flavour and Fragrance Journal, 18, 5-12. Steenson, D . F., Lee, J. H . , and M i n , D . B . (2002). Solid phase microextraction of volatile soybean oil and corn oil compounds. Journal of Food Science, 67(1), 71-76. Tombesi, N . B . , and Freije, H . (2002). Application of solid-phase microextraction combined with gas chromatography-mass spectrometry to the determination of butylated hydroxytoluene in bottled drinking water. Journal of Chromatography A, 963, 179183. Wolf, W. J. (1975). Lipoxygenase and flavor of soybean protein products. Journal of Agricultural and Food Chemistry, 23(2), 136-141. Yang, X . , and Peppard, T. (1994). Solid-phase microextraction for flavor analysis. Journal of Agricultural and Food Chemistry, 42(9), 1925-1930. Zhang, Z . , and Pawliszyn, J. (1993). Headspace solid-phase microextraction. Analytical Chemistry, 65(14), 1843-1852.  73  CHAPTER 3. ODOUR-ACTIVE COMPONENTS OF SIMULATED BEEF FLAVOUR ANALYZED BY SOLID PHASE MICROEXTRACTION AND GAS CHROMATOGRAPHY-MASS SPECTROMETRY AND -OLFACTOMETRY 2  3.1. Introduction  Generally, fresh raw meat gives off very little odour. However, over a thousand flavour compounds have been reported to be generated during cooking (Warriss, 2000). Among these, about 880 different volatile compounds have been identified in cooked beef (Mottram, 1994), including hydrocarbons, alcohols, phenols, aldehydes, ketones, carboxylic acids, esters, lactones, furans, pyrans, pyrroles, pyridines, pyrazines, oxazoles, oxazolines, thiophenes,  thiazoles,  thiazolines and other nitrogen or sulfur containing compounds. So far, the postulated primary reactions for beef flavour development on heating include pyrolysis o f amino acids and peptides, degradation of carbohydrates, ribonucleotides, thiamin and lipids, and interaction of sugars with amino acids or peptides, as extensively reviewed in the publications by Mottram (1991) and MacLeod (1994). It also has been reported that heating o f the lean portions o f beef, pork, chicken and lamb resulted in non-species-specific meaty flavour, while heating o f the fat in meats, especially the phospholipids and o f less importance the triglycerides, led to species-specific flavour of meat (Warriss, 2000).  Beef flavour seems to have been researched more widely than any other meat flavour (Mottram, 1994). Even though a great number o f volatile compounds have been reported in beef, only some of them are important in terms o f determining the beef flavour characteristics. Therefore, more and more, great effort has been made to find odour-active compounds and furthermore to identify key aroma compounds in beef. 2-Acetyl-2-thiazoline, furaneol, guaiacol, 2-ethyl-3,5-dimethyl pyrazine and 2,3-diethyl-5-methylpyrazine were identified as the character impact compounds for the roasty, caramel-like, burnt and earthy notes in roasted beef (Cerny and Grosch, 1992) and 4hydroxy-2,5-dimethyl-3(2/f)-furanone, 12-methyltridecanal, methional, 3-hydroxy-4,5-dimethyl-  A version o f this chapter has been published. Moon, S. - Y . , Cliff, M . A . , and Li-Chan, E . C. Y . (2006). Odour-active components of simulated beef flavour analysed by solid phase microextraction and gas chromatography-mass spectrometry and -olfactometry. Food Research International, 39:294-308. 2  74  2(5//)-furanone, and 2-furfurylthiol were demonstrated as odour impact volatile compounds in stewed beef (Guth and Grosch, 1993). 2-Methyl-3-furanthiol, 2-acetyl-l-pyrroline, methional, 1octen-3-one, phenylacetaldehyde, (£)-2-nonenal, (.E,ir)-2,4,decadienal, beta-ionone, and bis{2methyl-3-furyl) disulfide were identified as potent odourants with high aroma values from boiled beef (Gasser and Grosch, 1988) and Kerscher and Grosch (1997) reported that 2-furfurylthiol, 4hydroxy-2,5-dimethyl-3(2//)-furanone, and 2-methyl-3-furanthiol were recognized to be the most potent odourants o f boiled beef by aroma extract concentration analysis. In addition, 2-butanone, ethyl acetate, 2- and 3-methylbutanal, 2-octanone, and benzothiazole were identified with high detection frequency in cooked Irish beef (Machiels et al., 2003). Aside from the effects o f cooking conditions, method o f isolating volatile compounds and technique o f analyzing gas chromatography olfactometric data, significant differences were also observed in the odour-active compounds of cooked beef due to breed and diet (Machiels et al., 2004).  In the meantime, there has been an ongoing interest to develop a simulated beef flavour to meet the consumer demand for non-meat based or vegetarian products. Great effort has been applied to imitate beef flavour using blended spices, by-products from the corned beef industry, flavour enhancers such as monosodium-L-glutamate ( M S G ) , and protein hydrolysates (MacLeod and Seyyedain-Ardebili, 1981). In particular, there has been an increase in protein hydrolysates used as materials to produce meat-like savoury flavourings, including hydrolysed vegetable proteins produced from soy (Aaslyng et al., 1998), thermally treated yeast extracts (Munch and Schieberle, 1998), extruded enzyme-hydrolysed soybean protein (Baek et al., 2001) and soybean-based enzyme-hydrolysed vegetable protein (Wu and Cadwallader, 2002). In addition, processed meat flavour or so-called "reaction flavour" obtained by thermal treatment o f a mixture of food components, was employed to more closely simulate beef-like flavour (MacLeod et al., 1981). However, the products o f thermal processing have varying degrees o f meaty aroma, depending not only on heating temperature but also reactant ratio and heating time (Wasserman, 1979).  May (1974) proposed the main reactions occurring in the various model systems to synthesize meat flavours were reactions  of reducing sugars with amino compounds  consisting of  condensations, rearrangements, dehydrations and degradation, in addition to the reaction between reducing sugars and a sulfur donor such as cysteine. Due to the complexity o f the reactions, much  75  work on reaction flavours has been done with various model systems (Mussinan and Katz, 1973; Giintert et al., 1990; Hofmann and Schieberle, 1995; Hofmann and Schieberle, 1997; Mottram and Nobrega, 2002). However, in comparison to flavour compounds produced in model systems consisting of reducing sugars and amino compounds, not enough information is available in terms of odour-active components in simulated meat flavourings. Specifically, there is a lack of published literature describing odour activities of the volatile compounds and character impact compounds in commercially-available simulated meat flavour, compared to the reported flavour components in actual meat systems.  A  great number  of research  studies  on food  flavour  have  been  conducted with  gas  chromatography ( G C ) in conjunction with flame ionization detector (FID) and mass selective detector ( M S D ) to identify volatile compounds in food commodities. However, the analysis of the chemical components is not sufficient to determine the aroma profile because not all of the compounds are odourants and their contributions to flavour are not usually directly related to their abundance. In other words, although the relative retentions o f peaks in G C chromatograms give us clues to identify volatiles in the foods, the areas o f those peaks do not necessarily reflect the aroma intensity o f the foods. In that sense, G C analysis in combination with an olfactometric technique can be a useful tool to detect potent odour-active components from a complex mixture (Blank, 1997; Mistry et al., 1997), which may considerably contribute to the characteristic aroma of a given food. Dilution methods such as A E D A (aroma extract dilution analysis; Grosch, 1993) and CharmAnalysis™ (Acree et al., 1984), intensity method such as Osme (Miranda-Lopez et al., 1992), and the more recently developed detection frequency method (Linssen et al., 1993; Pollien et al., 1997), are the most commonly used GC-olfactometry (GC-O) techniques.  In the detection frequency method, the number of panelists detecting an odour at a certain retention index is utilized to quantify the aroma intensity of the effluent, rather than dilution values or perceived intensity as measured in the other methods. It was shown that detection frequency was directly related to the odour intensity (van Ruth and O'Connor, 2001). This method has been applied to determine odour-active compounds in foods such as bell peppers, dried leeks, dried French beans (van Ruth et al., 1995), cucumber (Marsili and Miller, 2000) and cooked beef (Machiels et al., 2003). Recently, Le Guen et al. (2000) compared three G C - O  76  methods, which were A E D A (dilution method), Osme (intensity method), and olfactometry global analysis (detection frequency method), to evaluate the main impact odourants o f cooked mussels. The results from the three olfactometric methods were very comparable and well correlated positively (with p values o f ~ 0.00001), indicating that all o f the methods were significantly interrelated.  The objectives o f this research were to identify headspace volatile compounds in simulated beef flavour by G C - M S analysis and to compare the profile o f the simulated beef flavour to the flavour compounds found in cooked beef in this study and in previous reports. Flavour compounds were analyzed by G C - M S coupled with a headspace S P M E method, which has been considered as a simple, rapid, solvent-free and economical method for extracting flavour components (Moon and Li-Chan, 2004). In addition, G C - O analysis based on detection frequency method was performed to evaluate the relative importance of the odour-active components and to determine the character impact compounds that contribute to the overall aroma o f the simulated beef flavour.  3.2. Experimental methods  3.2.1. Materials The simulated beef flavour ( S B F ; Mastertaste, Arlington Heights, IL) used in this study was a commercially-available blended flavour, as described in Section 2.2.1 o f this thesis. The solid phase assembly holder, 50/30 pm stableflex divinylbenzene/carboxen/polydimethylsiloxane (DVB/CAR/PDMS),  15 m L capacity G C sample vials and polypropylene hole cap with  PTFE/silicone septa were purchased from Supelco (Sigma-Aldrich Canada, Oakville, ON). A l l the fibres were conditioned by inserting them into a splitless/split injection port under helium flow for the recommended time at recommended temperatures according to the manufacturer's instruction before use and cleaned between analyses. Three «-paraffin mixes from Supelco (Cat. No. 47100, 47102 and 502243), specifically mixtures o f C5 to C 8 (C5, C6, C 7 and C8), C10 to C16 (C10, C12, C14, and C16), and C12 to C60 (C12, C14, C16, C18, C20, C22, C24, C26, C28, C30, C32, C36, C40, C44, C50, and C60), were used to determine the Kovats indices o f the volatiles for this study. 77  3.2.2. Headspace solid phase microextraction (HS-SPME) The sensitive and reproducible HS-SPME method that was previously established for analyzing SBF, as described in Chapter 2 of this thesis (Moon and Li-Chan, 2004), was applied to investigate the volatile components of SBF, boiled beef and roasted beef samples. Two grams of SBF and 5 g of distilled water were mixed in a 15 mL capacity GC sampling vial with a magnetic stirring bar. The vial was tightly capped with a polypropylene hole cap with a PTFE/silicone septum. To prepare roasted beef, 100 g of raw beef (Canada AAA steak, well marbled) was chopped into 0.8 cm and roasted in an oven at 190 °C for 15 minutes. Five grams of roasted beef was put in a 15 mL GC vial as the "roasted beef sample. To prepare boiled beef, 300 g of raw beef from the same package as used for roasted beef was boiled with 600 g water for 15 minutes then chopped. Five grams of chopped beef was put in a 15 mL GC vial as the "boiled beef sample. The 50/30 pm DVB/CAR/PDMS SPMEfibrewas exposed to the headspace above the sample for 60 minutes at 60 °C in a thermostat controlled water bath ( ± 2°C) to extract headspace volatile compounds. Stirring with a magnetic stirring bar was consistently applied. The selection of 60 °C as the adsorption temperature for HS-SPME was based on high sensitivity in terms of total area counts and number of peaks in the GC profile, as well as on the basis that beef analogues such as vegetarian hot dogs are usually served at temperatures close to 60 °C (Moon and Li-Chan, 2004).  3.2.3. GC-FID analysis A Hewlett-Packard (HP) 5890 gas chromatograph with flame ionization detector (FID) and a DB5 analytical fused silica capillary column (30 m x 0.32 mm x 0.25 pmfilmthickness from J&W Scientific, Folsom, CA) were used for analysis of the volatile compounds. conducted in a splitless mode for 3 min at 250 °C.  The injection was  The oven temperature was held at 40 °C for  3 min, ramped to 180 °C at the rate of 3 °C/min and then to 260 °C at 10 °C/min, and maintained at 260 °C for 2 min. Helium was used as a carrier gas at a column-head pressure of 12 psi (1 psi=6894.76 Pa). The temperature of the FID detector was 280 °C, and 35 mL/min of hydrogen, 350 mL/min of air and 30 mL/min of helium as a make-up gas were used.  Several parameters in  the chromatogram were adjusted to increase reliability in terms of number of peaks and peak area.  78  Peak width was set to 0.04 and initial threshold was set to 1.  The peaks with peak area under  10,000 were not regarded as reliable peaks.  3.2.4. GC-O analysis The G C - O was comprised of an H P 5890 G C including a sniffing port ( S G E , Texas, U S A ) with glass detection cone in addition to FID detector.  A t the end o f the capillary column the effluent  was split 1:3 for F I D and sniffing port, respectively, using deactivated and uncoated fused silica capillaries as transfer lines, and the sniffing cone was purged with humidified air to help in maintaining olfactory sensitivity by reducing dehydration of mucous membranes in the nasal cavity. The ratio o f 1:3 was used in order to allow enough flow to the FID detector while maximizing the effluent available to the sniffing port for detection by the panelist. The same column (DB-5) was used for G C - F I D and G C - O analyses, and the equipment conditions such as injection port temperature, oven temperature programming, carrier gas (helium) flow, detection temperature and gases flow (hydrogen and air) for G C - O were identical with the G C - F I D conditions as described above. ,  Detection frequency method using a panel of eight panelists was applied to obtain an odour profile for S B F . There was no specific training session for the S B F sample, but 5 of the 8 panelists had extensive experience with G C - O from other research. Each of the eight judges participated in perceiving the aroma compounds separated from S B F at the sniffing port, and the number of panelists detecting odour components during G C - O was summed to acquire an aromagram for S B F . The judges were asked to state odour characteristics, i f possible, whenever they detected an odour. A separate G C run was conducted for each panelist and the odour perception was recorded during the first 45 minutes, and any odour recognition at the sniffing port that was reported by fewer than four of the eight judges was considered as noise.  3.2.5. GC-MS analysis The G C - M S consisted o f an Agilent 6890N G C , equipped with the D B - 5 capillary column as above, in conjunction with Agilent H P 5973N mass selective detector ( M S D ) system. Separation was performed under the same conditions described above, using the same column as in the G C FID and G C - O analyses. The M S D was operated in the electron impact ionization mode with 70 79  eV and mass range was from 35 to 600 amu. The temperature of capillary direct interface, source, and quadrupole were 280 °C, 230 °C, and 150 °C, respectively.  The aroma compounds were identified by matching mass spectra with spectra of reference compounds in Wiley 275 Mass Spectral libraries (version D.01.00, Wiley, New York, 2000; from Agilent Technologies) and verified on the basis of mass spectra obtained and Kovats retention indices on DB-5 stationary phase from analysis of the aroma standards and FlavorNet by Acree and Arn (2004).  3.3. Results and Discussion  3.3.1. Identification of volatile compounds A typical GC-FID chromatogram obtained from the electrical responses of FID signals for the volatile compounds in the SBF is shown in Figure 3.1(a). The GC-FID chromatogram is significantly different from the aromagram obtained from the human response, as demonstrated in Figure 3.1(b). Among 88 peaks of volatile compounds in SBF detected by GC-MS, a total of 70 volatile compounds were either positively or tentatively identified by the corresponding mass spectra and Kovats indices, as shown in Table 3.1. A total of 40 peaks in roasted beef and 33 peaks in boiled beef were positively or tentatively identified and comparison with SBF in terms of the composition of the volatile compounds was demonstrated in Table 3.2. Overall, volatile compounds in SBF included more sulfur and nitrogen containing compounds and various alicyclic hydrocarbons while those in cooked beef in this study contained more aldehydes and ketones compared to SBF.  3.3.1.1. Furans and sulfur containing compounds 3-Methylfuran, 2-ethylfuran and 2-acetylfuran were identified in SBF (peaks 1, 2, and 7, respectively, in Table 3.1) and they were also reported to be found in cooked beef i.e. 3methylfiiran in canned beef (Persson and von Sydow, 1973), 2-ethylfuran in roast beef (Min et al., 1979), and 2-acetylfuran in roast beef and shallow fried beef (Min et al., 1979; Watanabe and Sato, 1972). 2-Pentylfuran, which seemed to be commonly found in beef under various conditions such as canned beef (Persson and von Sydow, 1973), boiled beef (Hirai et al., 1973), 80  roast beef (Min et al., 1979), canned and frozen stew (Peterson and Chang, 1982), roast beef, canned beef and canned beef stew (Shibamoto, 1980) and cooked ground beef (MacLeod and Ames, 1986) was also identified in roasted and boiled beef in this study but it was not found in SBF (Table 3.2). Furans can be produced by sugar caramelization and carbohydrate degradation (Shibamoto, 1980). Even though none o f the various furan compounds has been attributed as being crucial to meaty flavour, they have been regarded to contribute to the overall odour of broiled or roasted meat (Shahidi et al., 1986).  In addition, sulfur containing compounds such as 2-methylthiophene, 2-methyl-3-furanthiol, 2,5dimethylthiophene,  4,5-dimethylthiazole,  3-methyl-2-thiophenecarboxaldehyde,  and  dihydrothienothiophene were identified (peaks 4, 5, 6, 9, 28, and 41, respectively, Table 3.1) while methanethiol was found in boiled beef (Table 3.2). Methanethiol, which smells like cooked cabbage, was also reported to be found in cooked ground beef (MacLeod and Ames, 1986) and it can be perceived at a very low concentration due to its low threshold (0.2 ug/kg). Guth and Grosch (1994) concluded that methanethiol is one o f the character impact odour compounds o f stewed beef juice. Sulfur containing compounds have been reported to be important in cooked beef flavour since some heterocyclic sulfur compounds were described as possessing meat-like aromas (Mottram, 1994). It has been postulated that these sulfur containing heterocyclic compounds could be generated either from thermal degradation o f cystine or cysteine (Shu et al., 1985), through the interaction between carbonyl compounds and sulfur containing amino acids (Zhang and Ho, 1989) or as a result of thermal decomposition o f thiamin (Giintert et al., 1990). The former two pathways might be more important in flavour development in the S B F .  Most of the thiophenes found in meats were identified to be substituted at the 2-position (Werkhoff et al., 1993), similar to 2-methylthiophene and 2,5-dimethylthiophene found in this study. 2-Methylthiophene was isolated from pressure-cooked beef (Wilson et al., 1973; Madruga and Mottram, 1995) as well as roasted beef (Min et al., 1979), and reported to have sulfur (Acree et al., 2004) or green/sweet odour (Shahidi et al., 1986). 2,5-Dimethylthiophene was also identified in the neutral fraction of roasted beef (Min et al., 1979).  However, it has been stated  that 3-thiol-substituted thiophenes are the only thiophenes with meaty aroma (Werkhoff et al., 1993).  Meanwhile, van den Ouweland and Peer (1975) reported that a complex mixture of  81  compounds including mercapto-substituted furan and thiophene derivatives possessing roasted meat odours were generated from the reaction of 4-hydroxy-5-methyl-3(2//)-furanone or its thio analog with hydrogen sulfide such as 4-hydroxy-5-methyl-3(2r7) thiophenone. 4-Hydroxy-5methyl-3(2/f)-furanone could be a degradation product of ribonucleotides such as ribose-5phosphate, indicating that beefy aromas could possibly be generated from sugar degradation in addition to Maillard reaction.  Thiol, sulfide or disulfide group substituted furans at the 3-position have been regarded to be associated with meat-like aroma even though not many of the compounds in fact have been detected in cooked or roasted meat (Werkhoff et al., 1993).  2-Methyl-3-furanthiol was  identified in SBF and confirmed to possess meaty flavour in the sniffing port of GC-0 in this study. MacLeod (1994) reported through a search of the literature that only 25 chemical compounds of the 880 cooked beef aroma components had been described as meaty. Most of the compounds (20 out of 25) in the list were sulfur-containing compounds including 2-methyl-3furanthiol, which was present in the SBF in this study. Gasser and Grosch (1988) also demonstrated that 2-methyl-3-furanthiol, along with 6w(2-rnethyl-3-furyl) disulfide, was one of the odour compounds with high aroma values in cooked lean beef and which possessed meat-like odour quality based on aroma extract dilution analysis (AEDA). The researchers compared odour thresholds between 2-methyl-3-furanthiol (0.005-0.01 pg/kg) and 2-methyl-3-(methylthio)furan (25-30 pg/kg), which has been suggested to be a character impact compound in cooked beef (MacLeod and Ames, 1986) and suggested that due to the 2500 times lower odour threshold than 2-methyl-3-(methylthio)furan, 2-methyl-3-furanthiol had far greater importance to the cooked beef flavour.  Werkhoff et al. (1993) also reported that 2-methyl-3-furanthiol, 6w(2-methyl-3-  furyl)disulfide, and 2-methyl-3-(methylthio)furan were found in relatively large amounts in beef extracts while these compounds existed in pork and chicken in only trace amounts.  4,5-Dimethylthiazole, which was found in the SBF (peak 9), is known to possess roasted or grilled notes (Mottram, 1994) while 2,4-dimethylthiazole was reported to have meaty flavour (MacLeod, 1994). 4,5-Dimethylthiazole was reported in beef cooked at 140 °C (Madruga and Mottram, 1995), but 3-methyl-2-thiophenecarboxaldehyde and dihydrothienothiophene (peaks 28 and  41)  have  not  been  previously  reported  in  beef.  However,  3-methyl-282  thiophenecarboxaldehyde was identified in chicken (Werkhoff et al., 1993) and 5-methyl-2thiophenecarboxaldehyde was reported to be found in cooked beef (Wilson et al., 1973).  3.3.1.2. Nitrogen containing compounds Among the 10 nitrogen containing compounds tentatively identified in the SBF, most components, except acetylpyrrole (peak 23), belonged to the pyrazine class of compounds. These include 2-ethyl-6-methylpyrazine, trimethylpyrazine, 2-ethyl-3,6-dimethylpyrazine, 2,3-diethyl5-methylpyrazine,  2,5-diethyl-3,6-dimethylpyrazine,  and  3-isopentyl-2,5-dimethylpyrazine  (peaks 14, 15, 24, 30, 37, and 43, respectively, in Table 3.1), which were also reported to be found in cooked beef (MacLeod and Ames, 1986; Cerny and Grosch, 1994). In addition, the SBF contained 2,5-dimethyl-3-isobutylpyrazine, 2,6-diethyl-3,5-dimethylpyrazine, and 2-isoamyl-6methylpyrazine (peaks 34, 36, and 38, respectively), which have not been reported as volatile components in cooked beef (Table 3.1). 2-Acetylpyrrole (peak 23) was reported to be found in fresh, frozen beef stew and canned beef stew (Peterson et al., 1982), cooked ground beef (MacLeod and Ames, 1986) and shallow fried beef (Watanabe and Sato, 1972). It was also found in both roasted and boiled beef in this study (Table 3.2).  Pyrazines have been reported to be one of the main components in the volatiles of cooked meats; almost 41 % of the volatile constituents of pressure-cooked pork liver were pyrazines (Mussinan and Walradt, 1974). A variety of pyrazines can be produced from the Strecker degradation, which involves interaction of nitrogen containing molecules (e.g. a-amino acids) with dicarbonyls resulting from carbohydrate decomposition in a classic Maillard reaction.  MacLeod and  Seyyedain-Ardebili (1981) listed 46 pyrazines identified as volatile components of natural beef aroma in their review article and Mussinan et al. (1973) isolated 33 pyrazines from pressurecooked beef. Substantial similarities in aroma profile and roasted notes were found in the volatile compounds from beef heated with temperatures over 100 °C such as roasted beef, shallow fried beef, pressure-cooked beef and canned beef, and the roasted notes have been frequently derived from alkylpyrazines (MacLeod and Seyyedain-Ardebili, 1981). The odour of pyrazines has been traditionally regarded as nutty, roasted, or burnt; therefore Mussinan et al. (1973) proposed that the pyrazines contributed to characteristics of cooked foods due to the fundamental roasted aroma.  83  In this study, roasted beef included 2,6-dimethyl pyrazine and 2-ethyl-3,6-dimethylpyrazine whereas boiled beef did not have any pyrazine identified (Table 3.2).  3.3.1.3. Hydrocarbons and carbonyl compounds Both saturated and unsaturated aliphatic hydrocarbons were identified in SBF, such as 7-methyl3-methylene-l,6-octadiene, 1-tetradecene, tetradecane, and pentadecane (peaks 13, 51, 53, and 68, respectively, in Table 3.1), in addition to the aromatic hydrocarbons such as toluene and 4isopropyltoluene (peaks 3 and 19). With the exception of 7-methyl-3-methylene-1,6-octadiene, all of these compounds have also been reported to be found in cooked beef  (MacLeod and  Seyyedain-Ardebili, 1981). Meanwhile, tridecane, tetradecane and toluene were identified in both roasted and boiled beef in this study while dodecane and 4-isopropyl toluene was found in roasted beef but not in boiled beef (Table 3.2).  Although one terpene, delta-3-carene, was found in roasted beef (Table 3.2), a great number of alicyclic hydrocarbon compounds were tentatively identified in SBF, as shown in Table 3.1. Some of them have been reported as volatile aroma components in cooked beef, such as alphaphellandrene, gamma-terpinene, and beta-caryophyllene in canned beef stew (Chang and Peterson, 1977), limonene in roasted beef (Min et al., 1979), 2-pinene in heated ground beef (Ramarathnam et al., 1993). Peterson and Chang (1982) identified allo-ocimene, limonene, alphaphellandrene, gamma-terpinene, and beta-caryophyllene in fresh, frozen beef stew, and delta-3carene was found in roasted beef of this study (Table 3.2). Although a great number of saturated and unsaturated hydrocarbons in the range of C4 to C15 have been reported in beef, they are not believed to play an important role in roasted beef flavour (Min et al., 1979) due to their relatively weak, non beef-like odours. Compared to volatile compounds identified in cooked beef, the SBF contained a great number of alicyclic hydrocarbons including 25 terpenoids as shown in Table 3.1, which may contribute to differences in aroma characteristics between natural cooked beef and the SBF. Generally, terpenoids originate from herbs or plants. For example, the unique intricate terpenic odour of blackcurrant berries was given off from the terpene-containing glands in the plant (Ruiz del Castillo and Dobson, 2002).  Mitiku et al. (2001) investigated the  enantiomeric distribution in forty cold-pressed Citrus oil samples of alpha-pinene, beta-pinene, sabinene and limonene, which are some of the same terpenoids found in our study, and reported 84  that  the  species  terpenoids.  or  varieties  It c a n b e  showed  postulated  significant  that the  differences  in  the  enantiomeric  existence o f various terpenoids  in the  purity  of  S B F in this  the  study  are m o s t likely d e r i v e d f r o m plant o r i g i n ingredients.  F o r m a t i o n o f c a r b o n y l c o m p o u n d s m a y result f r o m o x i d a t i o n o f the u n s a t u r a t e d fatty a c i d s thermal  oxidative  decomposition  phenylacetaldehyde, (peaks have  11,  16, 2 1 ,  been  o f lipids in food  material. In this  study,  nonanal, decanal, and 5-methyl-2-phenyl-2-hexenal  27,  reported to  35,  and 64  be  in Table  identified  3.1).  in cooked  T h e first f i v e  beef, w h i l e  5-methyl-2-phenyl (1974).  Aldehydes  of  1983),  but  believed  also  to  react  reactions.  Various  aldehydes  and  ketones  aldehydes  and  5  found and  beef  aldehydes while  S B F  found  in the of  hexanal,  to  compounds have  beef,  reported  in roasted  Although most  and  roasted  4-dodecen-l-al  beef  were  heptanal,  beef  beef. were  was  It i s not  the  octanal,  interesting  identified  most  and  with  of  and  17  butanal,  through  of  this  and  compounds  et a l . ( 1 9 8 6 ) , r e l a t i v e l y l a r g e a m o u n t s r e p o r t e d as  l-octen-3-ol  listed  as  components were  found  only  to  octanal, and  note that m o s t  second  most  of  in  roasted  o f the  beef  while  has  been  (Moody,  unsaturated in w h i c h  4  ketones in  roasted  butanal,  beef. and  The  and  (E,E)-  were  was  as  major ketones  absence  in boiled  in Table  in  the  and  the  compound  such  major  benzaldehyde,  aldehydes  volatiles  in and  3.2.  a  review  l-penten-3-ol  study,  18  were  2,4-nonadienal,  boiled  flavour. In this beef  and  benzaldehyde  o f unsaturated alcohols  of boiled-beef  study  nonanal,  dominant  cooked  and  3-methyl  b y m a s s detector, as s h o w n  components  identified  in  S B F  amino-carbonyl  and ketones found  2-  in  f o u n d in S B F  2-hexenal  aldehydes  aldehydes  heptanal,  identified  saturated  findings  octanal,  including beef  flavour  range  the  beef  foods  in S B F . O f particular interest  the  Shahidi  and  wide  nonanal, nonenal,  55  1-heptanol,  a  were  hexanal,  A m o n g  were  produce  o f the  acetaldehyde,  roasted beef, respectively, b a s e d o n the response  l-octen-3-ol  to  odour  a n d 2- p r o p a n o n e appeared o n l y in roasted b e e f w h i l e  which  alcohol  the  consistent  identified  common,  in boiled  cooked  groups  ( T a b l e 3.2).  pentanal, hexanal,  aldehydes found  in  contribute  other  were  undecenal,  found  to  cooked  were  phenylacetaldehyde 2,4-decadienal,  in  ketones  only  with  research  in boiled beef boiled  not  were  o f these 6 aldehydes  reported to be f o u n d in c o o k e d p o r k b y M u s s i n a n a n d W a l r a d t  are  benzaldehyde,  and/or  by and  2-3,-butanediol,  l-penten-3-ol,  pentanol,  1-  85  hexanol, 1-heptanol, l-octen-3-ol, and 1-octanol were found in boiled beef. In addition, linalool and nerolidol (peaks 26 and 76) were found as straight-chain primary alcohols and an aromatic alcohol, 4-methyl-2,6-di-tert-butylphenol (peak 70), which is also known as BHT, was identified in SBF. Generally esters have been associated with fruity flavours, but a considerable number of esters were found in beef stew (Peterson and Chang, 1982). Two esters, bornyl acetate and cis11-hexadecen-l-yl acetate (peaks 40 and 79), were identified in the SBF.  From a quantitative point of view, based on the FID response in SBF, aldehydes (peaks 11 and 27 in Figure 3.1(a)) and terpenoids (peaks 44, 45, 48, 54, 58, 71 and 72 in Figure 3.1(a)) were abundant among a total of 70 volatile compounds tentatively identified in the SBF. However, this does not necessarily mean that they play an important role in the characteristic of the SBF, which is dependent on the odour characteristics and threshold of the volatile compounds.  3.3.2. Determination of odour-active compounds The volatile compounds extracted by SPME from SBF were isolated and detected by panelists at the sniffing port of GC-O. The results, presented as retention time and detection frequency by the 8 panelists, are shown in Figure 3.1(b). Of a total of 49 odour-active compounds in SBF, 21 compounds were identified, mainly consisting of heterocyclic sulfur or nitrogen compounds, aldehydes and terpenoids: 3-methyl furan, 2-methyl-3-furanthiol, 2,5-dimethylthiophene, 2acetylfuran, 4,5-dimethylthiazole,  benzaldehyde, trimethylpyrazine, delta-3-carene, alpha-  terpinene, 2-ethyl-3,6-dimethylpyrazine, nonanal, 2,3-diethyl-5-methyl pyrazine, beta-fenchyl alcohol, 2,5-dimethyl-3-isobutylpyrazine, decanal, 2-isoamyl-6-methylparazine, 3-isopentyl-2,5dimethylparazine, delta-elemene, beta-cubebene, calamenene, and caryophyllene oxide (peaks 1, 5, 6, 7, 9, 11, 15, 17, 18, 24, 27, 30, 33, 34, 35, 38, 43, 44, 49, 71, and 78, respectively, in Figure 3.1(b)).  Among the 49 odour-active compounds, 17 components (10 identified and 7 unidentified) were revealed to have high aroma values (detected by 6 or more out of 8 panelists). The most powerful odour impact compounds identified in the SBF were 2-methyl-3-furanthiol, delta-3-carene, alphaterpinene, and 2-ethyl-3,6-dimethylpyrazine (peaks 5, 17, 18, and 24, respectively). In addition, two unidentified peaks were noted with high detection frequency; one was detected before peak 1 86  by 7 panelists, and the other, between peaks 24 and 27, was detected by 8 panelists. It could be hypothesized that these two peaks are sulfur-containing aroma compounds which did not appear as peaks in GC-FID chromatogram due to their relatively low concentration in SBF along with low sulfur sensitivity of the FID detector, but could be detected by most of the panelists owing to the nature of low thresholds of sulfur-containing compounds.  2-Methyl-3-furanthiol is the compound that was found by Gasser and Grosch (1988) to have high aroma values in cooked lean beef, along with methional, 2(£T)-nonenal, 2(£),4(£)-decadienal and 6/s(2-methyl-3-furyl) disulfide. Kerscher and Grosch (1997) also proposed 2-methyl-3-furanthiol as one of the most potent odourants in boiled beef. Among the pyrazines identified in this study (including trimethylpyrazine, 2-ethyl-3,6-dimethylpyrazine, 2,3-diethyl-5-methyl pyrazine, 2,5dimethyl-3-isobutylpyrazine,  2-isoamyl-6-methylparazine,  and  3-isopentyl-2,5-  dimethylparazine), 2-ethyl-3,6-dimethylpyrazine was the most powerful odour-active compound to contribute to the aroma profile of the SBF. Cerny and Grosch (1992) stated that the character impact compounds with high aroma value from roasted beef  included 2-ethyl-3,5-  dimethylpyrazine and 2,3-diethyl-5-methylpyrazine, while Specht and Baltes (1994) suggested that the differences in pleasant flavour qualities in shallow-fried beef samples were mainly due to the combination of 2-ethyl-3,5-dimethylpyrazine and 2-propyl-3-methylpyrazine, and 2-ethyl3,6-dimethylpyrazine with lesser importance.  It can be speculated that 2-methyl-3-furanthiol, which has been reported to be responsible for beef broth or roasted meat odour (Gasser and Grosch, 1988; Macleod, 1994; Kerscher and Rosch, 1997), plays a major role in conferring beef-like characteristics to the SBF, while 2-ethyl-3,6dimethylpyrazine along with other pyrazines appeared to provide the roasted note in the SBF. The importance of 2-methyl-3-furanthiol to contribute to meaty note in meat-like process flavouring has also been demonstrated by several other research groups, which examined the odour-active compounds of the process flavouring. Baek et al. (2001) reported 2-methyl-3furanthiol as the most intense compound with a cooked rice/vitamin-like/meaty aroma note in a beef-like process flavour, which was produced from hydrolysed vegetable proteins. Schieberle (1998) described that 2-methyl-3-furanthiol  Munch and  along with 2-furfurylthiol, 3-  methylbutanal, and methional showed the highest odour activity values in a thermally treated 87  commercial yeast extract and a self-prepared bakers' yeast extract possessing roasty, meat-like odours. In addition, W u and Cadwallader (2002) indicated that 2-furfurylthiol followed by 2methyl-3-furanthiol,  3-mercapto-2-pentanone,  and  3-(methylthiol)propanal were the  most  important odour-active compounds in the overall aroma o f the meatlike process flavouring from hydrolysed vegetable protein.  On the other hand, along with many other alicyclic hydrocarbons found in S B F , delta-3-carene and alpha-terpinene, both o f which are known to have a lemon odour (Acree et al., 2004), might contribute to some perceived difference between the S B F and cooked beef flavour. Moreover, the aldehydes and ketones that were found in the cooked beef but absent in S B F , may also be responsible for the different odour profile of S B F from that o f cooked beef. According to Gasser and Grosch (1988), 2-octenal, 2-nonenal, 2,4-nonadienal, 2,4-decadienal (fatty, like fried potato), l-octen-3-one (mushroom-like), 2-octanone, 2-decanone and 2-dodecanone (musty, fruity) and phenylacetaldehyde (honey-like) in addition to sulfur containing compounds were included in the 17 odour compounds having high aroma values found in cooked beef. 2-Nonenal and 2,4decadienal were aldehydes with odour activity reported in roasted beef (Cerny and Grosch, 1992) and 3-mercapto-2-pentanone, l-octen-3-one and 2-nonenal were indicated to be a part of the most potent odourants in boiled beef (Kerscher and Grosch., 1997). Although these aldehydes and ketones may not be directly responsible for beefy flavour in cooked beef, they might more than likely contribute to the fatty notes o f the cooked beef, resulting in subtle differences between S B F and cooked beef.  3.4. Conclusion  Application o f G C - F I D , G C - M S and G C - 0 to the research o f volatile compounds in S B F revealed an intricate combination o f aroma compounds from several sub-classes. A total of 70 aroma compounds were identified in the S B F by combined use o f S P M E headspace analysis with G C - F I D and G C - M S . Moreover, G C - 0 was useful to identify the main odour-active compounds which contribute to the aroma profile o f S B F . Among these, several sulfur and nitrogen containing compounds as well as various terpenoids were proposed to be o f essential importance for the flavour profile o f the S B F . The most powerful odour-active compounds identified in S B F  88  were 2-methyl-3-furanthiol, delta-3-carene, alpha-terpinene, and 2-ethyl-3,6-dimethylpyrazine. It was postulated that the meaty characteristic of the SBF was provided by 2-methyl-3-furanthiol with its meat-like odour, along with various pyrazines contributing roasted notes, while other aroma compounds including terpenoids, along with the absence of various aldehydes and ketones, resulted in the subtle difference between the SBF and cooked beef.  It is anticipated that the  knowledge gained from this study on the potential impact odourants in SBF and comparison with odour-active compounds reported in cooked beef may provide some foundation for further research or applications by scientists and food or flavour companies that are interested in the chemistry and aroma characteristics of vegetarian simulated meat flavours.  89  £2 . O &  ca -=H  1  . <3  1  . X2 «?-• - 4  X  . O  <=» -  S  O  o  o  e  o  o  o  4  o  o  o H  s  o o o  H S  O  £3 Retention time  17 18  O  o  es o  (min)  24  (b) o fl <u  I  67  11  35  27  71  6  M  PH  fl  49  o +-»  o  €  5  34 33 3d  10  43  78  44  20  Figure 3.1. ( a )  GC-FID  50  (min)  c h r o m a t o g r a m o f s i m u l a t e d b e e f f l a v o u r a n d (b) a r o m a g r a m  volatile c o m p o u n d s  o f simulated beef flavour expressed  f r e q u e n c y b y GC-O  ( n u m b e r o f panelists  in Table  40  30 Retention time  =  8).  as r e t e n t i o n t i m e  and  of  detection  P e a k n u m b e r s c o r r e s p o n d to the p e a k  numbers  3.1.  90  Table 3.1. Volatile compounds in simulated beef flavour tentatively identified by mass spectral-search (MS) and linear retention index (LRI). Peak No.  Volatile compound  1 2  3-Methylfuran 2-Ethylftiran  646 713  MS MS, LRI  Previously reported in cooked beef 1 1,2,3  3 4  Methylbenzene 2-Methylthiophene  762 819  MS, LRI MS  1,3,4,5 1,3,6,7  5 6 7 8 9  2-Methyl-3-furanthiol 2,5-Dimethylthiophene 2-Acetylfuran alpha-Phellandrene 4,5-Dimethylthiazole  849 881 888 907 910  MS, MS, MS, MS MS,  1,6,8,9 3 1,3,10 1,11,12 6  10 11  2-Pinene Benzaldehyde  912 938  MS, LRI MS, LRI  5 1,2,3,4,13  12 13 14 15  Sabinene 7-Methyl-3-methylene-l,6-octadiene 2-Ethyl-6-methylpyrazine Trimethylpyrazine  957 959 986 990  MS, MS, MS, MS,  1,4,6,10,14 1,4,10,14,15  16  Octanal  995  MS, LRI  1  LRI  2  Basis for identification  LRI LRI LRI LRI  LRI LRI LRI LRI  3  Odour descriptor reported in the literature A c i d , sour, whey butter-like Paint Sulfur, green, sweet Meaty Balsamic Fresh Earthy, roasted, nutty, green Pine Almond, musty, sweet, metallic Woody Balsamic Fruity Nutty, roasted  Odour descriptor from panelists during G C - O  4  Stinky, candle, beef Perfume, lemony, beef  Floral, grass, burnt beef Meaty (2), oxo Beef, sweet, ham, rancid Sulfur, sweet, toffee, rancid Meaty, sulfur, caramelized, floral Pop corn, caramel, herby, sulfur, chemical, spicy  Burnt, smokey, painty, solvent  1,2,4,6,13,16, Soapy 77  1/  17  delta-3-Carene  1000  MS, LRI  18  alpha-Terpinene  1007  MS, LRI  1  Lemon  19 20  4-Isopropyltoluene Limonene  1014 1018  MS, LRI MS, LRI  1 1,3,5,11  Lemon  Table 3.1. continued  Lemon  Floral (3), sugar, honey, smokey, beef Grassy, plastic, smokey, cod liver oil. beef Grassy, rancid, rubbery  Basis for identification  Peak No.  Volatile compound  21 22 23  Phenylacetaldehyde gamma-Terpinene Acetylpyrrole  1030 1049 1052  MS, LRI MS, LRI MS, LRI  24  2-Ethy 1-3,6-dimethylpyrazine  1069  MS, LRI  26 27  Linalool Nonanal  1093 1097  MS, LRI MS, LRI  28 29  1105 1130  MS MS  30  3-Methyl-2-thiophenecarboxaldehyde 2,3-Dihydro-3,5-dihydroxy-6-methyl4//-pyran-4-one 2,3-Diethyl-5-methylpyrazine  1148  MS, LRI  32 33  4-Terpeneol beta-Fenchyl alcohol  1165 1179  MS MS, LRI  34 35  2,5-Dimethyl-3-isobutylpyrazine Decanal  1193 1198  MS MS, LRI  36 37 38  2,6-Diethyl-3,5-dimethylpyrazine 2,5 -D iethyl -3,6-dimethylpyrazine 2-Isoamyl-6-methylpyrazine  1221 1225 1242  MS MS MS  1  LRI  2  Odour descriptor Previously reported in the reported in literature cooked beef 1,2,10,13,16 Rosy, perfume Gasoline 1,11,12 1,4,10,11,18 Unpleasant, plastic, antiseptic Nutty, roasted, 1,4,6,8,14 potato Lemon, floral 1,2,4,6,13,16, Soapy, green, moldy 17 3  1,4,6,14,15  Nutty, roasted, potato Camphor  1,2,4,6,10,16, Soapy 17  Odour descriptor from panelists during G C - O  4  Herb, oil, burning  Meat, smokey (2), sulfur, natural gas, soil Grassy, tea, vegetable, lemony, sour, beefy Ham, sweet, beefy, savoury  Burnt, cured meat, popcorn (2), buttery (2) Fried, barbecue, beefy Bleach, chlorine, wood, dandelion, meat Roasted meat, floral Rubber tubing, cooked veggie, sweet popcorn, smokey  1,14 Rubbery (2), cabbage, sweet, mint iiimt  39  frvms-Anethole  1276  MS  40 41 43  Bornyl acetate Dihydrothienothiophene 3-Isopentyl-2,5-dimethylpyrazine  1278 1285 1310  MS MS MS  Table 3.1. continued  Sweet, sulfur, hot wing, beefy  10  Yeasty (2), fermented, floral  Previously ./ Basis for reported i n LRI identification *\ , . ,3 cooked beef  Peak No.  Volatile compound  44  delta-Elemene  1331  MS  45 48 49  alpha-Cubebene alpha-Copaene beta-Cubebene  1342 1367 1381  MS MS MS  50  beta-Elemene  1384  MS  51 52 53 54 55 56 57 58 61 62 64 65 66 67 68 69 70 71  1-Tetradecene c w-Caryophyl lene Tetradecane beta-Caryophyllene epi-Bicyclosesquiphellandrene allo-Aromadendrene alpha-Amorphene alpha-Humulene gamma-Cadinene beta-Selinene 5-Methyl-2-phenyl-2-hexenal alpha-Selinene gamma-Muurolene alpha-Muurolene Pentadecane beta-Bisabolene 4-Methyl-2,6-di-ter//-butylphenol Calamenene  1390 1395 1398 1411 1419 1427 1440 1443 1469 1476 1482 1485 1488 1494 1500 1505 1508 1515  MS MS MS, LRI MS, LRI MS MS MS MS MS MS MS MS MS MS MS, LRI MS MS MS  72 73  delta-Cadinene 1,2,3,4,4a,7-Hexahydro-1 ;6-dimethyl-  1518 1524  MS MS  1  Table 3.1. continued  2  Odour descriptor , . ^ _ ,• . Odour descriptor from reported i n the .. . , . ^ ^ ^ 4 ,. panelists during GC-0 literature perfume Barn yard, smelly socks, freshly cut wood, sulfur, burnt A  Rotten, dusty, yeasty, rubbery Sewage, green tea, meaty (2), MSG 1 Ul  1,3 11,19  Alkane Woody  Woody Herbaceous  Soil, meaty Ashy, yeast, sulfur  Herb, burnt, sulfur Wax, cured meat, roasted  1,3  Alkane  1,3,12,20  Herbal, savoury, spicy, yeasty, MSG, oxo, beef Woody Medicine, caramelized,  Peak No. 1  Volatile compound  LRI  2  Basis for identification  75 76 77 78  4-( 1 -methylethyl)-naphthalene Elemol Nerolidol Ionol Caryophyllene oxide  1542 1559 1562 1571  MS MS MS MS  79  cis-\ 1-Hexadecen-l-yl acetate  1576  MS  Previously reported in cooked beef  3  Odour descriptor reported in the literature  Odour descriptor from panelists during G C - O  4  roasted  Matches, musty, earthy, butter, meaty  'Peak numbers refer to all the peaks detected by G C - M S , as shown in Figures 3.1(a) and 3.1(b); only the 70 identified compounds are shown in this table. Linear retention index on D B - 5 capillary column. References for volatile components previously reported in cooked beef: 1 MacLeod and Seyyedain-Ardebili (1981); 2 Drumm and Spanier (1992); 3 M i n et al. (1979); 4 MacLeod and Ames (1986); 5 Ramarathnam et al. (1993); 6 Madruga and Mottram (1995); 7 Wilson et al. (1973); 5 Grosch (1993); 9 Werkhoff et al. (1993); 10 Cerny and Groschl. (1994); 11 Peterson and Chang. (1982); 12 Chang and Peterson (1977); 13 Moody (1983); 14 Mussinan and Katz. (1973); 15 Kerler and Grosch (1996); /6~Guth Grosch. (1993); 77 Ramarathnam et al. (1991); 18 Misharina et al. (1994); 19 Shahidi et al. (1986); 20 Hirai et al. (1973). Odour descriptor expressed by panelists at a given retention index in G C - O which was the same as that in G C - M S . Numbers in parentheses refer to the number o f panelists using the same descriptor.  Table 3.2. Comparison o f volatile compounds in simulated beef flavour, boiled beef and roasted beef tentatively identified by G C - M S coupled with H S - S P M E . Peak size Compound  Simulated beef flavour  Roasted beef  Boiled beef  1. Aliphatic and aromatic hydrocarbons 7-Methyl-3-methylene-1,6-octadiene Dodecane Tridecane Tetradecane Tetradecane 1- Tetradecene Pentadecane 2- Isopropyl-5-methyl-9-methylene-bicyclodec-1 -ene Toluene 4-Isopropyl toluene Calamenene  M  M S S xS S S xS  trans-Anethole  xS  alpha-Amorphene  xS  xS S  s  s s  s s  xS  2. Alicyclic hydrocarbons alpha-Phellandrene 2-Pinene Sabinene delta-3-Carene alpha-Terpinene Limonene gamma-Terpinene delta-Elemene alpha-Cubebene alpha-Copaene beta-Cubebene beta-Elemene cw-Caryophyllene beta-Caryophyllene epi-Bicyclosesquiphellandrene allo-Aromadendrene alpha-Humulene gamma-Cadinene beta-Selinene alpha-Selinene gamma-Muurolene alpha-Muurolene beta-Bisabolene Table 3.2. continued  S S xS S M M S L M L S M M xL M S L M M M xS M M 95  Peak size Compound delta-Cadinene 1,2,3,4,4a,7-Hexahydro-1,6-dimethyl-4(1 -methylethyl)-naphthalene  Simulated beef flavour  1  Roasted beef  Boiled beef  M  3. Alcohols 2,3-Butanediol l-Penten-3-ol Pentanol 1 -Hexanol 1 -Heptanol l-Octen-3-ol 1-Octanol Linalool 4-Terpeneol beta-Fenchyl alcohol 4-Methyl-2,6-di-ter//-butylphenol Elemol Nerolidol Ionol  xS S  xS S  s s s M M  M M M M M xS M  4. Aldehydes Acetaldehyde Butanal 3- Methyl butanal 2-Methyl butanal Pentanal Hexanal Heptanal Octanal Octenal Nonanal Nonenal 2,4-Nonadienal Decanal 2-Decenal (E, £)-2,4-Decadienal Undecanal Undecenal Dodecanal 4- Dodecen-l-al Hexadecanal Benzaldehyde Phenylacetaldehyde Table 3.2. continued  S S S S  s M M  M M M xS M S  s  L M M S L M S S S  S  s s s s s s  M  M  xS xS S M S  M  S 96  Peak size  1  Compound 5-Methyl-2-phenyl-2-hexenal  Simulated beef flavour M  5. Ketones 2-Propanone 2- Butanone 3- Hydroxy-2-butanone 2- Heptanone 2,3-Octanedione 2,3-Dihydroxy-6-methyl-4//-pyran-4-one 6. Carboxylic acids and esters Acetic acid Hexanoic acid Octanoic acid Butyl butyrate Benzyl acetate Bornyl acetate cis-l 1-Hexadecen-l-yl acetate 7. Furans and S-heterocyclic compounds 3- Methylfuran 2-Ethylfuran 2-Acetylfuran 2-Pentylfuran Methanethiol 2-Methylthiophene 2- Methyl-3-furanthiol 2,5-Dimethylthiophene 4.5- Dimethylthiazole 3- Methyl-2-thiophenecarboxaldehyde Djhydrothienothiophene 8. N-heterocyclic compounds 2.6- Dimethyl pyrazine Acetylpyrrole 2-Ethyl-3,6-dimethyl pyrazine 2-Ethyl-6-methylpyrazine Trimethylpyrazine 2,3-Diethyl-5-methylpyrazine 2.5- Dimethyl-3-isobutylpyrazine 2.6- Diethyl-3,5-dimethylpyrazine 2,5 -D iethyl -3,6-dimethylpyrazine 2-Isoamyl-6-methylpyrazine Table 3.2. continued  Roasted beef  s M S xS S  Boiled beef  M M S M  S xS xS xS M S  M S M M S S S S S M S  S M S S  xS S S  s s s s s  97  Peak size  1  Compound 3-Isopentyl-2,5-dimethyl-pyrazine 9. Miscellaneous Caryophyllene oxide  Simulated beef flavour M  Roasted beef  Boiled beef  M  Peak size was considered to be "xS" (extra small) when the log value of the peak area was below 6, "S" (small) between 6 and 7, "M" (medium) between 7 and 8, "L" (large) between 8 and 9, "xL" (extra large) above 9.  98  3.5. References Aaslyng, M. D., Elmore, J. S., and Mottram, D. S. (1998). Comparison of the aroma characteristics of acid-hydrolyzed and enzyme-hydrolyzed vegetable proteins produced from soy. Journal ofAgricultural and Food Chemistry, 46, 5225-5231. Acree, T. E., Barnard, J., and Cunningham, D. G. (1984). A procedure for the sensory analysis of gas chromatographic effluents. Food Chemistry, 14, 273-286. Acree, T., and Arn, H. (2004). FlavorNet. Gas chromatography-olfactometry (GC-O) of natural products, <http://www.flavornet.org/index.html> accessed on July 14, 2005. Baek, H. H., Kim, C. J., Ahn, B. H., Nam, H. S., and Cadwallader, K. R. (2001). Aroma extract dilution analysis of a beeflike process flavor from extruded enzyme-hydrolyzed soybean protein. Journal of Agricultural and Food Chemistry, 49, 790-793. Blank, I. (1997). Gas chromatography-olfactometry in food aroma analysis. In R. Marsili, Techniques for Analyzing Food Aroma, (pp. 293-329). New York: Marcel Dekker. Cerny, C , and Grosch, W. (1992). Evaluation of potent odorants in roasted beef by aroma extract dilution analysis. Zeitschrift fur Lebensmitteluntersuchung und -Forschung A, 194, 322-325. Cerny, C , and Grosch, W. (1994). Precursors of ethyldimethylpyrazine isomers and 2,3-diethyl5-methylpyrazine formed in roasted beef. Zeitschrift fur Lebensmitteluntersuchung und Forschung A, 198,210-214. Chang, S. S., and Peterson, R. J. (1977). Recent developments in the flavor of meat. Journal of Food Science, 42, 298-305. Drumm, T. D., and Spanier, A. M. (1992). Changes in the content of lipid autoxidation and sulfur-containing compounds in cooked beef during storage. Journal of Agricultural and Food Chemistry, 39, 336-343. Hirai, C , Herz, K. O., Pokorny, J., and Chang, S. S. (1973). Isolation and identification of volatile flavor compounds in boiled beef. Journal of Food Science, 38, 393-397. Hofmann, T., and Schieberle, P. (1995). Evaluation of the key odorants in a thermally treated solution of ribose and cysteine by aroma extract dilution techniques. Journal of Agricultural and Food Chemistry, 43, 2187-2194. Hofmann, T., and Schieberle, P. (1997). Identification of potent aroma compounds in thermally treated mixtures of glucose/cysteine and rhamnose/cysteine using aroma extract dilution techniques. Journal ofAgricultural and Food Chemistry, 45, 898-906.  99  Gasser, U . , and Grosch, W . (1988). Identification of volatile flavour compounds with high aroma values from cooked beef. Zeitschrift fur Lebensmitteluntersuchung und -Forschung A, 186, 489-494. Grosch, W. (1993). Detection of potent odorants in foods by aroma extract dilution analysis. Trends in Food Science and Technology, 4, 68-73. Giintert, M . , Briining, J., Emberger, R., Kopsel, M . , Kuhn, W., Thielmann, T., and Werkhoff, P. (1990). Identification and formation of some selected sulfur-containing flavor compounds in various meat model systems. Journal of Agricultural and Food Chemistry, 38, 2027-2041. Guth, H . , and Grosch, W . (1993). 12-Methyltridecanal, a species-specific odorant o f stewed beef. Lebensmittel Wissenschaft und Technologie, 26, 171-177. Guth, H . , and Grosch, W. (1994). Identification of the character impact odorants o f stewed beef juice by instrumental analyses and sensory studies. Journal of Agricultural and Food Chemistry, 42, 2862-2866. Kerler, J., and Grosch, W . (1996). Odorants contributing to warmed-over flavor (WOF) o f refrigerated cooked beef. Journal of Food Science, 61, 1271-1274, 1284. Kerscher, R., and Grosch, W. (1997). Comparative evaluation o f potent odorants of boiled beef by aroma extract dilution and concentration analysis. Zeitschrift fur Lebensmitteluntersuchung und -Forschung A, 204, 3-6. Le Guen, S., Prost, C , and Demaimay, M . (2000). Critical comparison o f three olfactometric methods for the identification of the most potent odorants in cooked mussels (Mytilus edulis). Journal of Agricultural and Food Chemistry, 48, 1307-1314. Linssen, J. P. FL, Janssens, J. L . G . M . , Roozen, J. P., and Posthumus, M . A . (1993). Combined gas chromatography and sniffing port analysis of volatile compounds of mineral water packed in polyethylene laminated packages. Food Chemistry, 46, 367-371. Machiels, D . , Istasse, L . , and van Ruth, S. M . (2004). Gas chromatography-olfactometry analysis of beef meat originating from differently fed Belgian Blue, Limousin and Aberdeen Angus bulls. Food Chemistry, 86, 377-383. Machiels, D., van Ruth, S. M . , Posthumus, M . A . , and Istasse, L . (2003). Gas chromatographyolfactometry analysis o f the volatile compounds of two commercial Irish beef meats. Talanta, 60, 755-764. MacLeod, G . (1994). The flavour of beef. In F. Shahidi, Flavor of Meat and Meat Products, (pp. 4-37). Glasgow: Chapman & Hall. MacLeod, G . , and Ames, J. M . (1986). Capillary gas chromatography-mass spectrometric analysis o f cooked ground beef aroma. Journal of Food Science, 51, 1427-1434.  100  MacLeod, G . , and Seyyedain-Ardebili, M . (1981). Natural and simulated meat flavors (with particular reference to beef). CRC Critical Reviews in Food Science and Nutrition, 14, 309437. Madruga, M . S., and Mottram, D . S. (1995). The effect o f p H on the formation of Maillardderived aroma volatiles using a cooked meat system. Journal of the Science of Food and Agriculture, 68, 305-310. Marsili, R. T., and Miller, N . (2000). Determination of major aroma impact compounds in fermented cucumbers by solid-phase microextraction-gas chromatography-mass spectrometry-olfactometry detection. Journal of Chromatographic Science, 38, 307-314. May, C . G . (1974). A n introduction to synthetic meat flavour. Food Trade Review, 44(1), 7, 9-10, 12-14. M i n , D . B . S., Ina, K . , Peterson, R. J., and Chang, S. S. (1979). Preliminary identification of volatile flavor compounds in the neutral fraction of roast beef. Journal of Food Science, AA, 639-642. Miranda-Lopez, R., Libbey, L . M . , Watson, B . T., and McDaniel, M . R. (1992). Odor analysis of Pinot noir wines from grapes of different maturities by a gas chromatography-olfactometry technique (Osme). Journal of Food Science, 57, 985-993, 1019. Misharina, T. A . , Golovnya, R. V . , Strashnenko, E . S., and Medvedeva, I. B . (1994). Sorptionstructural and mass spectrometric characteristics o f the volatile components of model systems and flavor compounds with meat odor. Journal of Analytical Chemistry, 49, 653659. Mistry, B . S., Reineccius, T., and Olson, L . K . (1997). Gas chromatography-olfactometry for the determination o f key odorants in foods. In R. Marsili, Techniques for Analyzing Food Aroma, (pp. 265-292). N e w York: Marcel Dekker. Mitiku, S. B . , Ukeda, H . , and Sawamura, M . (2001). Enantiomeric distribution of a-pinene, Ppinene, sabinene and limonene in various citrus essential oils. In A . M . Spanier, F. Shahidi, T. H . Parliament, C . Mussinan, C . T. Ho, and E . Tratras Contis, Food Flavors and Chemistry, (pp. 216-231). Cambridge: Royal Society o f Chemistry. Moody, W. G . (1983). Beef flavor - a review. Food Technology, 37(5), 227-232, 238. Moon, S.-Y., and Li-Chan, E . C . Y . (2004). Development of solid-phase microextraction methodology for analysis o f headspace volatile compounds in simulated beef flavour. Food Chemistry, 88, 141-149. Mottram, D . S. (1991). Meat. In H . Maarse, Volatile Compounds in Foods and Beverages, (pp. 107-177). N e w York: Marcel Dekker.  101  Mottram, D . S. (1994). Meat flavour. In J. R. Piggott and A . Paterson, Understanding Natural Flavors, (pp. 140-163). Glasgow: Chapman & Hall. Mottram, D . S., and Nobrega, I. C . C . (2002). Formation of sulfur aroma compounds in reaction mixtures containing cysteine and three different forms o f ribose. Journal of Agricultural and Food Chemistry, 50, 4080-4086. Munch, P., and Schieberle, P. (1998). Quantitative studies on the formation of key odorants in thermally treated yeast extracts using stable isotope dilution assays. Journal of Agricultural and Food Chemistry, 46, 4695-4701. Mussinan, C . J., and Katz, I. (1973). Isolation and identification of some sulfur chemicals present in two model systems approximating cooked meat. Journal of Agricultural and Food Chemistry, 21, 43-45. Mussinan, C . J., and Walradt, J. P. (1974). Volatile constituents o f pressure cooked pork liver. Journal ofAgricultural and Food Chemistry, 22, 827-831. Mussinan, C . J., Wilson, R . A . , and Katz, I. (1973). Isolation and identification of pyrazines present in pressure-cooked beef. Journal ofAgricultural and Food Chemistry, 21, 871-872. Persson, T., and von Sydow, E . (1973). Aroma of canned beef: gas chromatographic and mass spectrometric analysis of the volatiles. Journal of Food Science, 33, 377-385. Peterson, R. J., and Chang, S. S. (1982). Identification of volatile flavor compounds of fresh, frozen beef stew and a comparison of these with those of canned beef stew. Journal of Food Science, 47, 1444-1448, 1453. Pollien, P., Ott, A . , Montigon, F., Baumgartner, M . , Munoz-Box, R., and Chaintreau, A . (1997). Hyphenated headspace-gas chromatography-sniffing technique: Screening o f impact odorants and quantitative aromagram comparisons. Journal of Agricultural and Food Chemistry, 45, 2630-2637. Ramarathnam, N . , Rubin, L . J., and Diosady, L . L . (1991). Studies on meat flavor. 2. A quantitative investigation of the volatile carbonyls and hydrocarbons in uncured and cured beef and chicken. Journal ofAgricultural and Food Chemistry, 39, 1839-1847. Ramarathnam, N . , Rubin, L . J., and Diosady, L . L . (1993). Studies on meat flavor. 4. Fractionation, characterization, and quantitation of volatiles from uncured beef and chicken. Journal ofAgricultural and Food Chemistry, 41, 939-945. Ruiz del Castillo, M . L . , and Dobson, G . (2002). Varietal differences in terpene composition of blackcurrant (Ribes nigrum L ) berries by solid phase microextraction/gas chromatography. Journal of the Science of Food and Agriculture, 82, 1510-1515.  102  Shahidi, F., Rubin, L. J., and D'Souza, L. A. (1986). Meat flavor volatiles: A review of the composition, techniques of analysis, and sensory evaluation. CRC Critical Reviews in Food Science and Nutrition, 24, 141-243.  Shibamoto, T. (1980). Heterocyclic compounds found in cooked meats. Journal of Agricultural and Food Chemistry, 28, 237-243.  Shu, C-K., Hagedorn, M. L., Mookherjee, B. D., and Ho, C-T. (1985). Volatile components of the thermal degradation of cystine in water. Journal of Agricultural and Food Chemistry, 33, 438-442. Specht, K., and Baltes, W. (1994). Identification of volatile flavor compounds with high aroma values from shallow-fried beef. Journal of Agricultural and Food Chemistry, 42, 2246-2253. van den Ouweland, G. A. M., and Peer, H. G. (1975). Components contributing to beef flavor. Volatile compounds produced by the reaction of 4-hydroxy-5-methyl-3(2//)-furanone and its thio analog with hydrogen sulfide. Journal of Agricultural and Food Chemistry, 23, 501505. van Ruth, S. M., and O'Connor, C. H. (2001). Evaluation of three gas chromatographyolfactometry methods: comparison of odour intensity-concentration relationships of eight volatile compounds with sensory headspace data. Food Chemistry, 74, 341-347. van Ruth, S. M., Roozen, J. P., Cozijnsen, J. L., and Posthumus, M. A. (1995). Volatile compounds of rehydrated French beans, bell peppers and leeks. Part II. Gas chromatography/sniffing port analysis and sensory evaluation. Food Chemistry, 54, 1-7. Warriss, P. D. (2000). Meat Science, (p. 260). Oxon: CABI publishing. Wasserman, A. E. (1979). Chemical basis for meat flavor: A review. Journal of Food Science, 44, 6-11. Watanabe, K., and Sato, Y. (1972). Shallow-fried beef: Additional flavor components. Journal of Agricultural and Food Chemistry, 20, 174-176.  Werkhoff, P., Briining, J., Emberger, R., Giintert, M., and Hopp, R. (1993). Flavor chemistry of meat volatiles: New results on flavor components from beef, pork, and chicken. In Recent Development in Flavor and Flagrance Chemistry: Proceedings of the 3 Haarmann and Reimer Symposium, (pp. 183-213). Holzminden: VCH.  rd  International  Wilson, R. A., Mussinan, C. J., Katz, I., and Sanderson, A. (1973). Isolation and identification of some sulfur chemicals present in pressure-cooked beef. Journal of Agricultural and Food Chemistry, 21, 873-876. Wu, Y-F. G., and Cadwallader, K. R. (2002). Characterization of the aroma of a meatlike process flavoring from soybean-based enzyme-hydro lyzed vegetable protein. Journal of Agricultural and Food Chemistry, 50, 2900-2907. 103  Zhang, Y., and Ho, C-T. (1989). Volatile compounds formed from thermal interaction of 2,4decadienal with cystine and glutathione. Journal of Agricultural and Food Chemistry, 37, 1016-1020.  104  CHAPTER  4. CHANGES IN PERCEIVED AROMA CHARACTERISTICS OF SIMULATED BEEF FLAVOUR BY ADDING SOY PROTEIN ISOLATE AS ASSESSED BY DESCRIPTIVE SENSORY ANALYSIS AND GAS CHROMATOGRAPHY-OLFACTOMETRY 3  4.1. Introduction A characteristic beef odour is one of the most important parameters to determine the quality of beef analogue products such as vegetable protein based meat substitutes for vegetarian consumers. Among the vegetable protein materials used in meat substitutes, soy protein in particular has become more and more popular due to its reported health benefits such as hypocholesterolemic effects (Anderson et al., 1995) and cancer prevention (Fournier et al., 1998). More interest has been gained after the approval of a health claim by the U.S. Food and Drug Administration stating the cardiovascular advantages of soy consumption (USFDA, 1999).  Flavour related problems including "beany" odour (Boatright and Lei, 1999; Lei and Boatright, 2001; Wolf 1975) and off-flavour (Inouye et al., 2002; Maheshwari et al., 1995; McDaniel and Chan, 1988) have created technical obstacles to be overcome for the increased usage of soy proteins in human foods (Maheshwari et al., 1995). Aside from these undesirable yet hard-toremove soy aromas, the interactions of soy proteins with desirable aroma components of added flavour formulations have presented a different challenge concerning soy based products. Gremli (1974) reported that the presence of soy protein in aqueous systems increased the retention of volatile components in samples, while Malcolmson and McDaniel (1987) observed the suppression of chicken flavour in a formulated soup at high levels of soy protein.  Thermally produced simulated meat flavours, so called "reaction flavours", are often employed in vegetarian products to more closely simulate meat-like flavour. Generally, this type of flavour ingredient can be generated by heating amino acids or protein hydrolysates with various sugars to construct varying degrees of meaty aroma according to the reactant ratio, heating temperature and A version of this chapter will be submitted for publication. Moon, S. -Y., and Li-Chan, E. C. Y. Changes in perceived aroma characteristics of simulated beef flavour by adding soy protein isolate as assessed by descriptive sensory analysis and gas chromatography-olfactometry. The 3  Journal of the American Oil Chemists' Society.  105  heating time (Wasserman, 1979). M a y (1974) proposed the main reactions occurring in the various model systems to synthesize meat flavours, namely condensations, rearrangements, dehydrations and degradation between reducing sugars and amino acids including cysteine as a sulfur donor. Different types o f protein hydrolysates as starting materials for meat flavour were reviewed by the same author (May, 1974).  Due to the complexity o f the reactions involved in their creation, simulated meat flavours used in vegetarian products are likely to have a multifaceted aroma profile.  Nevertheless, it is  imperative to elucidate the influence of other ingredients such as soy protein on the sensory characteristics o f the simulated meat flavours, in order to gain an understanding and provide potential strategies for overcoming the diminution in meaty aroma intensity observed in the presence of soy proteins.  In a previous study, development o f a solid phase microextraction ( S P M E ) technique for gas chromatographic ( G C ) analysis o f simulated beef flavour was achieved (Moon and Li-Chan, 2004). Odour-active components in the simulated beef flavour were identified by the combination of GC-mass spectrometry ( G C - M S ) and GC-olfactometry (GC-O) (Moon et al., 2006).  The  present research was undertaken as an extension o f these earlier studies, in order to identify "indicator peaks" in the G C profiles that could be used to monitor soy-protein induced changes in the aroma o f simulated beef flavour, which would be correlated to the perceived aroma characteristics o f soy protein-simulated beef flavour mixtures as assessed by panelists.  Therefore, the specific objectives o f this study were (1) to apply descriptive analysis ( D A ) to express the aroma characteristics o f the simulated beef flavour (SBF) and soy protein isolate (SPI), (2) to monitor changes in these sensory attributes as a function o f different ratios of S B F and SPI, (3) to elucidate the relationship between the sensory response and G C data for the various mixtures o f the S B F and SPI, and (4) to select G C indicator peaks that are correlated to beefy characteristics. Successful investigation of changes in aroma profile o f S B F upon addition of SPI and selection of indicator peaks could lead to potential application to monitor the retention or release o f beef flavour due to SPI-SBF interactions in samples containing SPI.  106  4.2. Experimental methods 4.2.1. Materials The simulated beef flavour (SBF; Mastertaste, Arlington Heights, IL) was a commerciallyavailable blended flavour, as described in Section 2.2.1 of this thesis. The soy protein isolate (SPI, lot # 02060631-532) used in this study was a commercially available product from Solae (St. Louis, MO). Protein content of the SPI analyzed by the nitrogen combustion method using a LECO FP-428 (LECO Corporation, Joseph, MI) was 89.3 ± 0.1 % using 6.25 as a conversion factor (Renkema et al., 2002; Puppo et al., 2004). The solid phase assembly holder, 50/30 pm stableflex divinylbenzene/carboxen/polydimethylsiloxane (DVB/CAR/PDMS), 15mL capacity GC sample vials and polypropylene hole cap with PTFE/silicone septa were purchased from Supelco (Sigma-Aldrich Canada, Oakville, ON).  4.2.2. Sensory analysis Descriptive analysis was conducted by adapting the method of Zook and Pearce (1988) to obtain data describing the sensory attributes of SBF and SPI.  4.2.2.1. Panelist training Ten subjects consisting of 8 women and 2 men with an interest in descriptive sensory evaluation were selected from students in the Food Science graduate program at UBC and from food development staff at a food company in Vancouver producing soy-based products for vegetarians and consumers preferring meatless products. The panelists received 8 hours of training, consisting of four 2-h sessions conducted over 2 weeks.  Training session I The objectives of the first training session were to discuss the aroma characteristics of the SBF and SPI and to select potentially appropriate attributes describing them through open discussion. Two SBF samples, which were low SBF (150 mg in 5 g water) and high SBF (500 mg SBF in 5 g water), one SPI sample (500 mg SPI in 5 g water) and one mixture sample (500 mg SBF with 500 mg SPI in 5 g water) were prepared. Samples were provided in capped 15 mL vials covered with aluminum foil to prevent any potential bias from the appearance of the samples and held at 60 °C  107  for 20 minutes before serving. The aroma descriptors for the SBF suggested by the panel at the first training session were (1) brothy/oxo/miso-like, (2) roasted/dry/cooked/barbecue, (3) beefy (identity of beef), (4) rare/raw/bloody/uncooked, (5) browned caramel/sweet/candy-like and (6) cardboardy/rancid/off-flavour/yeasty. The aroma descriptors suggested for the SPI were (1) soymilk-like, (2) cooked cereal, and (3) straw/hay-like.  Training session II The aim of the second training session was to reach consensus among panelists on the aroma descriptors. The explored aroma terms, along with their characteristics and reference materials prepared for the training session, are presented in Table 4.1. OxoBeef (Knorr OXO Beef Bouillon, Toronto, Ontario) solution was prepared as specified in Table 4.1 to explain the attribute of "brothy/oxo beef-like/miso-like".  Fresh beef (1.25 kg of Canada AAA steak, 2.5 cm thick,  purchased at Safeway) was cooked with 5 g of Club House seasoned salt (McCormick Canada Inc., London, Canada) in a conventional oven as indicated in Table 4.1. The surface crust (2 mm thickness) of the roasted beef was separated from the inner part of the beef; the surface crust was used for the attributes of "roasted/dry cooked beef/barbecue" while the interior of the cooked beef was employed for the "beefy" attribute. Fresh extra lean ground beef (Safeway) was employed for the attribute of "rare/raw/uncooked". As very little odour was perceived from fresh raw beef, 100 g of the ground beef was put in each 125 mL Erlenmeyer flask to amplify the odour intensity of raw beef; the flasks were covered with Parafilm and aluminum foil and kept at 4 °C before serving. Werther's Original Caramel (Storck, Mississauga, Ontario) was used for the attributes of "browned caramel/sweet/candy-like". Wet cardboard (15 cm X 30 cm) as well as yeast extract powders (supplied by the Hain Celestial Group, Delta, B.C., Canada) were used as references for the "cardboardy/rancid/off-flavour/yeasty" attribute.  Soy milk (Sunrise Soya Beverage, unsweetened) was used as a reference standard for "soymilklike" attribute. Cooked oatmeal (Quaker Oatmeal Regular) prepared as shown in Table 4.1 was used for the attribute of "cooked cereal". Hay supplied from the Dairy Education and Research Center of UBC at Agassiz was put in a 200 mL beaker covered with Parafilm and aluminum foil and used for "straw/hay-like" attribute.  108  Except for the reference  standards for "rare/raw/bloody/uncooked", "cardboardy/rancid/off-  flavour/yeasty" and "straw/hay-like", which were provided at room temperature due to the sample characteristics, 20 g o f each o f the reference standards was placed in 15 m L vials, providing about half headspace volume, and held at 60 °C for 20 minutes. After discussing the odour characteristics o f the references, the panelists were asked to perceive and differentiate those attributes in prepared samples consisting o f 500 mg S B F in 5 g water and 500 mg SPI in 5 g water. B y the end o f the training session, the panelists agreed to select 3 attributes for S B F ("roasted", "beefy", and "yeasty") and 2 attributes for SPI ("soymilk-like" and "cereal") and recognized these attributes in S B F and SPI sample mixtures.  Training session III The objectives o f the third training session were to train the panelists to assess the 5 selected aroma attributes for samples containing various concentrations o f S B F with or without SPI, and to discuss the results in order to reach a consensus on the intensity o f the perceived attributes. The definitions which were agreed upon for the 5 selected descriptors, namely "roasted", "beefy", "yeasty", "soymilk-like", and "cereal", are listed in Table 4.2. Six samples containing 0, 250 or 500 mg o f S B F with or without 250 mg o f SPI were prepared as training samples and evaluation sheets with unstructured 15 cm line scales anchored at the ends by marks 1.3 cm from either end were used. Open discussion proceeded during evaluation o f each sample, until consensus was reached among the panelists.  Training session IV The objective o f the fourth training session was to practice assessing various mixtures o f S B F and SPI. A total o f 12 samples (3 sets each with 4 samples) were assessed using the same sensory sheets to be used in the main sensory evaluation. A reference comprised o f 150 mg o f SBF and 250 mg o f SPI was provided with each sample set. After the individual assessment for the samples, open discussion was held to assess the evaluation results and reach agreement for perceived aroma intensity of the provided samples.  109  4.2.2.2. Descriptive sensory analysis Eight o f the 10 trained panelists were able to participate in the main evaluation, which was carried out in individual sensory booths in a sensory room with air circulation. A total of 24 test samples were evaluated in two blocks o f replications consisting o f 12 samples, which were randomly ordered within the replication blocks. Four samples and a reference were provided for each set of the 6 evaluations. The 24 test samples (12 different samples in duplicate, coded with 3-digit random numbers) and the references (designated "R") were covered with aluminum foil and held at 60 °C for 20 minutes before serving. The 12 samples examined were 150 or 500 mg of SBF (i.e. " l o w " or "high" SBF) with 0, 250 or 500 mg o f SPI (i.e. "no", "medium" or "high" SPI) in 5 g of distilled water, each with or without heat treatment (98 °C, 30 minutes), as shown in Table 4.3. The panelists were asked to evaluate the samples for the 5 aroma attributes that had been developed during the 4 training sessions. The intensity o f each attribute in a sample was rated on an unstructured 15 cm line scale anchored at 1.3 cm from each end and labeled with "Slight" on the left, "Moderate" in the center, and "Intense" on the right. Panelists were requested to mark the intensities of the attributes in comparison with the reference, which were identified as "moderate" in the unstructured line scale. Data were quantified by measuring the distance of the panelists' mark from the origin in centimeters. In addition, a container including coffee beans was provided for each panelist, who was asked to sniff the coffee beans to minimize sensory adaptation.  4.2.3. Gas chromatography In a 15 m L capacity G C sampling vial with a magnetic stirring bar, 150 or 500 mg o f SBF and 0, 250 or 500 mg o f SPI in 5 g o f distilled water were mixed. The vial was tightly capped with a polypropylene hole cap with a PTFE/silicone septum. A 50/30 um D V B / C A R / P D M S S P M E fibre was then exposed to the headspace above the sample solution for 60 minutes at 60 °C in a thermostat controlled water bath ( ± 2 °C) to extract headspace volatile compounds. Stirring with a magnetic stirring bar was consistently applied. Detailed information on the headspace solid phase microextraction ( H S - S P M E ) methodology and analysis conditions were described in earlier reports (Moon and Li-Chan, 2004; Moon et al., 2006). A l l treatments were prepared in triplicate and results were expressed as the average of the triplicate values.  110  4.2.4. Statistical analysis Data from the descriptive analysis was evaluated by analysis of variance (ANOVA), Pearson correlation analysis, and principal component analysis (PCA) using Minitab software (version 13.30, Minitab inc. PA USA).  ANOVA with Duncan's multiple comparison tests were  performed to determine whether there were differences among individual samples and panelists for each sensory attribute.  When the interaction between sample and replication was found to  be significant, an adjusted F test was subsequently conducted based on using mean square of the interaction instead of mean square of error as the denominator (O'Mahony, 1986). Panel performance was also reviewed in terms of concordance among the panelists and repeatability of the individual panelists for each attribute using Pearson correlation and ANOVA analysis, respectively. In addition, Pearson correlation analysis on mean sensory scores for each attribute in the DA and mean values of peak area in GC chromatogram was conducted to determine correlation among panelists and between peak areas and sensory scores. PCA was performed on the mean sensory scores of the 12 samples for 5 attributes. The vector coefficients for each sensory attribute were rescaled by a factor of 5 to depict PCA bi-plots, and ellipses were employed to provide a visual aid for describing samples. 4.3. Results and Discussion  4.3.1. Sensory analysis The results of the ANOVA of the attribute ratings across the 12 samples for 8 panelists are summarized in Table 4.4. Samples were highly significantly different at p < 0.001 for all attributes.  Although "panelist" was a significant source of variation in all attributes except  "beefy", this result is not unusual in descriptive analysis, and may mean that panelists were using different parts of the scale due to physiological differences in perceived intensity or differences in personal style of scoring, such as central or extreme raters. A significant replication effect was found in the "beefy" attribute, which might be partly influenced by the block effect of the replications in this study.  There was no significant interaction between sample and panelist,  indicating that the panelists were scoring consistently for each attribute. In addition, no significant interaction between panelist and replication was found, showing that the panelists on the whole were reproducible in the duplicate trials for each attribute. However, a significant ill  interaction between sample and replication was observed for the soymilk-like and cereal attributes.  This implies that intensities o f soymilk-like and cereal in the samples were not rated  similarly when they were replicated.  Because of the significant replication effect on beefy  attribute and significant interaction effect between sample and replication in soymilk-like and cereal attributes, an adjusted F test was performed using the mean square o f sample x replication interaction instead of mean square of error as a denominator, dealing with replications as a random effect (O'Mahony, 1986), as shown in Table 4.5.  The results showed that samples were  still significantly different in all attributes after taking the variation o f replication into account, even though the level o f significance of both soymilk-like and cereal attributes changed from p < 0.001 t o p < 0.01.  4.3.1.1. Panel performance The mean scores o f each panelist for the 5 attributes and the grouping o f the panelists by Duncan's multiple tests are shown in Table 4.6. The ranges between lowest and highest mean scores for soymilk-like (4.41), cereal (4.41), and yeasty (5.69) notes were relatively large compared to those o f roasted (2.23) and beefy (1.97) notes. This result may be due to either different perception o f panelists in each of the attributes or difficulty  of consensus in  characterization o f soymilk-like, cereal, and yeasty notes compared to roasted and beefy notes. Obviously, panelist 4 evaluated the samples in a different way from the other panelists, giving low scores for all the samples especially in soymilk-like and yeasty notes.  Panelist 5 may be a  potential outlier in soymilk-like, cereal, and beefy notes.  In sensory analysis, panelists are considered as an instrument to measure sensory attributes of the samples. The degree of reliability o f the panel immensely affects the results of the sensory evaluation and further data interpretation; therefore, panel performance was reviewed (Jeong et al., 2004; Labbe et al., 2004) in terms of concordance among the panelists and repeatability of the individual panelist. Pearson correlation coefficients and probabilities o f sample mean for each sensory attribute between individual panelist and the other 7 panelists are presented in Table 4.7. The correlation coefficients between each panelist 1, 2, 3, 6, 7 or 8 with the rest o f the panel were generally very high for each sensory attribute with the exception of the "yeasty" attribute. However, correlation coefficients corresponding to panelists 4 and 5 were relatively low (p > 112  0.05) for most of the attributes. In addition, the F ratio and probability for the replication effect for each of the panelists shown in Table 4.8 indicate that Panelists 1, 2, 3, 6, 7, and 8 were highly reliable in terms of repeatability, while the replicates were significantly different for panelist 4 in the cereal attribute (p < 0.05) and for panelist 5 in beefy and yeasty attributes (p<0.1). Therefore, further analysis was conducted without including the ratings from panelists 4 and 5.  4.3.1.2. Sensory characteristics of the samples Correlation between the 5 sensory attributes and the concentration of SBF and SPI in the samples are shown in Table 4.9. Roasted, beefy, and yeasty attributes were positively correlated with SBF concentration (p < 0.05) and negatively correlated with SPI concentration (p < 0.05) while soymilk-like and cereal notes were highly positively correlated with SPI (p < 0.01) and negatively correlated with SBF concentration (p < 0.05).  ANOVA indicated significant differences (p < 0.001) among different samples in the intensity of all 5 attributes. The mean intensity values of the 5 attributes and the results of Duncan's multiple comparison tests are shown in Table 4.10. For more effective visual comparison, changes in the 5 sensory attributes in SBF by adding SPI are illustrated in Figure 4.1(a).  Samples SI, S2, and S4 were strong in beefy, roasted and to a lesser extent, yeasty notes, while they were weak in soymilk-like and cereal notes. Samples S3 and S6 were relatively even in all the 5 attributes. Sample S5 was very strong in soymilk-like and cereal notes but very weak in roasted, beefy, and yeasty notes. In particular, S2, S4, and S6, showed stronger roasted, beefy, and yeasty characteristics compared to SI, S3, and S5, respectively, due to increased SBF concentration at the same SPI content. Among high SBF samples the 3 beef flavour attributes (roasted, beefy, and yeasty) decreased while soymilk-like and cereal notes increased significantly, in the order of S2, S4, and S6, corresponding to increasing SPI content. A similar trend was observed as a function of increasing SPI content in the low SBF samples SI, S3, and S5, but with even more extensive changes (decreasing roasted, beefy and yeasty notes, and increasing soymilk-like and cereal notes). Although there were some minor differences, the general trends  113  for heat treated samples were very similar to those observed for the non-heat treated samples, as shown in Figure 4.1(b).  P C A was performed and the principal component (PC) loadings and scores of the sensory attributes and the samples are depicted in Figure 4.2(a) and Figure 4.2(b). P C I , P C 2 and PC3 explained 97.8 %, 1.0 % and 0.7 % of the total variance, respectively. The eigenvalues of P C 2 and PC3 were relatively small, 0.0478 and 0.0342, respectively, compared to P C I , which had an eigenvalue of 4.8892. A l l the attributes were highly loaded on P C I , which was characterized by positive loadings for cereal and soymilk-like attributes and negative loadings for roasted, yeasty, and beefy attributes. A s a result, the 5 sensory attributes can be largely divided into 2 groups. One group includes roasted, yeasty, and beefy notes, which have negative values for P C I and are attributed to S B F , while the other group contains cereal and soymilk-like notes, which come from SPI and have positive values for P C I . The 12 samples can be described by positive or negative P C I values, as shown in Figure 4.2(a). Roasted, beefy and yeasty characteristics were stronger while cereal and soymilk-like notes were weaker in the order o f S2, S4, S I , S6, S3, and S5. B y adding SPI, the roasted, beef and yeasty notes decreased while cereal and soymilk-like notes increased in both low S B F and high S B F samples. In fact, P C 1 could be used as a tool to differentiate the aroma characteristics among samples. From the comparison between SI (low SBF and no SPI) and S4 (high S B F with medium SPI), it was noted that although the amount of SBF in S4 was increased more than 3-fold compared to S I , the beefy characteristics of S4 were suppressed by the addition of SPI resulting in similar degree o f beefy characteristics between the two samples.  A similar although less marked trend was observed in the comparison between S3  (low S B F with medium SPI) and S6 (high S B F with high SPI). It can be postulated that interaction between S B F and SPI considerably hindered the beefy notes from releasing in the samples.  While all the attributes were negatively loaded in P C 2 , yeasty notes were positively loaded in PC3, as shown in Figure 4.2(b). Consequently, PC3 dimension was mainly defined by the yeasty note, which was described as being related to cardboardy/rancid/off-flavour o f the samples during the sensory training session, in contrast to the other 4 attributes as demonstrated in Figure 4.2(b). It was interesting to see that all the heat treated samples except sample S 2 H tended to be located  114  on upper position of the P C 3 in Figure 4.2(b), implying stronger yeasty note of heated samples compared to non heat treated samples.  4.3.2. Selection of indicator peaks (IPs) for the beef flavour It would be useful to select indicator peaks to represent the beefy attributes in samples in order to evaluate changes in beefy characteristics under varying treatment conditions. Indicator peaks have been applied in various ways. For instance, propanal analyzed by G C - M S was used as an indicator peak to determine antioxidant effect of spices in sardines (Kasahara 2004). Galactose determined by G C - F I D was used as an index of heat treatment in milk (Chiesa et al., 1999), and filbertone ((E)-5-methylhept-2-en-4-one) analyzed by G C - F I D was used to verify adulteration of olive oils in hazelnut oils (Blanch et al., 2000). Indicator peaks could be used as a multiplecompound-quality index, such as the application of several selected bacterial metabolites identified by G C - M S and G C - 0 to predict spoilage off-flavours in packed or smoked salmon (Jorgensen et al., 2001). In the study from Narzip et al. (1999), a group o f volatile compounds was selected as a "forcing index" that appeared or increased in the early stage o f beer ageing while several Strecker aldehydes and furfural were chosen as an "ageing index" that increased mainly late in ageing of beer. In this study, indicator peaks consisting o f multiple aroma compounds to represent beefy attribute in sample were considered since beefy characteristics arise from not one or a few compounds, but a combination of various subclasses of components.  To qualify as an indicator peak, first o f all, the peak appearing in the G C chromatogram should be positively correlated (at p < 0.05) with attributes related to the characteristic beefy flavour in descriptive sensory analysis. Correlation analysis was conducted between the D A scores of the 5 sensory attributes with the concentrations of S B F and SPI and the area o f peaks in G C chromatogram. The results are attached in Appendix I. Among 112 G C peaks, the areas of 75 peaks were significantly correlated with the specific sensory attributes (p < 0.05). O f these 75 peaks, 64 peaks were positively correlated with roasted, beefy, and yeasty notes while negatively correlated with soymilk-like and cereal notes. Eleven peaks were positively correlated with soymilk-like and cereal notes and negatively correlated with roasted, beefy, and yeasty notes.  115  However, even though the areas o f peaks give us information on the intensity of electrical response by flame ionization detector, the size of a peak according to the concentration of the compound does not necessarily reflect the aroma intensity o f the peak in a sample due to the different odour threshold and/or differences in detector sensitivity for different compounds. Hence, another criterion was added to select indicator peaks; it must be odour-active with greater than 50 % detection frequency as assessed by G C - O .  The present results were analyzed in the  context of the odour-active components in S B F previously tentatively identified by G C in conjunction with olfactometry (Moon et al., 2006). The results o f the analysis showed that among the 64 peaks that were positively correlated with roasted, beefy, and yeasty notes, only 15 peaks were identified as odour-active compounds in SBF, with greater than 50 % detection frequency (i.e. 4 or more detection frequency out of 8 panelists) by G C - 0 analysis. However, 5 out of the 15 peaks were not able to be identified by G C - M S due to either small peak area or low quality of identification.  Therefore, based on the two criteria, 15 peaks, namely peaks with retention times 1.717, 7.832, 12.046, 15.772, 16.133, 16.864, 18.199, 19.744, 22.062, 24.103, 25.374, 25.703, 28.204, 30.291, and 35.680 minutes, were selected from the 112 peaks detected in the sample including S B F (Table 4.11). Investigation of the selected indicator peaks under various environmental conditions may provide useful information to monitor changes in beef attribute related flavour binding or releasing in sample mixture of SPI and SBF.  4.4. Conclusion D A along with G C analysis was applied to investigate changes in aroma characteristics of S B F upon addition of SPI. Three attributes (beefy, roasted, and yeasty) for S B F and 2 attributes (soymilk-like and cereal) for SPI were selected by D A and used to assess various mixtures of SBF and SPI. The results from the sensory analysis confirmed that roasted, beefy and yeasty notes were highly positively correlated with S B F concentration in the samples containing mixtures of S B F and SPI, and the beefy related notes were severely suppressed by increasing SPI content. Heat treated samples showed generally similar tendency to those observed for the nonheat treated samples. Principal component analysis of the data revealed that the aroma  116  characteristics among samples could be differentiated by P C I , which was characterized by positive loadings for cereal and soymilk-like attributes and negative loadings for roasted, yeasty, and beefy attributes. Moreover, P C 3 was also useful to detect increased yeasty notes in heated samples compared to non heat treated samples.  Given the observation that addition of SPI to S B F significantly hindered the release of the beefy related aroma in the samples, indicator peaks were selected, which would represent beefy characteristics in the sample. B y considering information obtained by G C - O analysis on aromaimpact compounds in S B F , 15 peaks were categorized as indicator peaks, which were all recognized as odour-active compounds in G C - O analysis, and at the same time were significantly positively correlated with beefy characteristics analyzed by D A . The indicator peaks may form the basis of further research to elucidate the mechanism of SPI-SBF interactions to explain the suppression of beef volatile flavour components in samples containing SPI.  117  Roasted  Figure 4.1. Cobweb diagram o f the sensory scores from the descriptive analysis data for (a) 6 unheated mixtures of beef flavour and soy protein isolate (b) 6 heated (98 °C, 30 minutes) mixtures (n=12; 6 selected panelists with 2 replications). L - S B F and H - S B F represent low (150 mg) and high (500 mg) dose o f simulated beef flavour while N-SPI, M - S P I , and H-SPI symbolize no (0 mg), medium (250 mg), and high (500 mg) amount of soy protein isolate in the sample. Refer to Table 4.3 for the sample codes. 118  51 52 53 54 55 56  and and and and and and  SIH (L-SBF; N-SPI) S2H (H-SBF; N-SPI) S3H (L-SBF; M-SPI) S4H (H-SBF; M-SPI) S5H (L-SBF; H-SPI) S6H (H-SBF; H-SPI)  PC2  PCI  Figure 4.2(a). P C loadings and scores of the sensory attributes and the sample mixtures by principal component analysis; P C I versus P C 2 . L - S B F and H - S B F represent low (150 mg) and high (500 mg) dose o f simulated beef flavour while N-SPI, M - S P I , and H-SPI symbolize no (0 mg), medium (250 mg), and high (500 mg) amount o f soy protein isolate in the sample. Refer to Table 4.3 for the sample codes.  119  Yeasty  PC3  (b)  Soymilk-like  S5H PCI  S5  Cereal Beefy  Roasted  51 52 53 54 55 56  andSlH(L-SBF;N-SPI) and S2H (H-SBF; N-SPI) and S3H (L-SBF; M-SPI) and S4H (H-SBF; M-SPI) and S5H (L-SBF; H-SPI) and S6H (H-SBF; H-SPI)  Figure 4.2(b). P C loadings and scores of the sensory attributes and the sample mixtures by principal component analysis; P C I versus PC3. L - S B F and H - S B F represent low (150 mg) and high (500 mg) dose o f simulated beef flavour while N-SPI, M - S P I , and H-SPI symbolize no (0 mg), medium (250 mg), and high (500 mg) amount o f soy protein isolate in the sample. Refer to Table 4.3 for the sample codes.  120  Table 4.1. Description and references for aroma terms used in the sensory training session for descriptive analysis of simulated beef flavour and soy protein isolate.  Terms Brothy, oxobeef-like, miso-like  Characteristics  Reference  The smell associated with beef broth or oxo-beef solution.  4.5 g of OxoBeef powder (Knorr OXO Beef Bouillon) in 175 mL water  Roasted, dry cooked beef, barbecue  The smell associated with roasted or barbecued beef.  Surface crust of roasted beef cooked with Club House seasoned salt (McCormick) at 190 °C for 25 minutes and roasted 10 more minutes after turning over  Beefy  The smell associated with beef, as differentiated from pork or chicken meat.  Interior of roasted beef cooked at 190 °C for 25 minutes and cooked 10 more minutes after turning over  Rare, raw, uncooked  The smell associated with bloody raw beef.  Fresh extra lean ground beef  Cardboardy, rancid, off-flavour, yeasty  The smell associated with off flavour when beef has gone bad.  Wet cardboard (15 cm X 30 cm) and yeast extract powder  Browned caramel, sweet, candy-like  The smell associated with caramel.  Melted Werther's Original Caramel  Soymilk-like  The smell associated with volatile matter from soymilk.  Soymilk (Sunrise Soya Beverage, unsweetened)  Cooked cereal  The smell associated with hot cooked cereal.  Instant oatmeal (Quaker Oatmeal Regular; 1 packet in 150 mL boiling water)  Straw, hay-like  The smell associated with dry straw or hay.  Hay supplied from the Dairy Research Center at UBC  121  Table 4.2. Definition of the 5 selected sensory attributes agreed upon by panelists through the training sessions in the descriptive analysis for simulated beef flavour and soy protein isolate.  Descriptors  Definition  Roasted  ^ beef-related top note, associated with the degree of "roast beef crusty surface" that the sample represented.  R  f y  Yeasty  Soymilk-like  Cereal  The lingering characteristic note, associated with as "beef in contrast to pork or chicken meat.  The off-flavour agar-like note and as a roasted smell gone bad.  The green note associated with cooked soybeans.  The note associated with cooked oatmeal or cooked pasta.  122  Table 4.3. Samples used in the sensory evaluation and G C analysis.  Code  SBF  2  SPI  3  Distilled de-ionized water  Heat Treatment  SI  150 mg  Omg  5g  No  S1H  150 mg  Omg  5g  Yes  S2  500 mg  Omg  5g  No  S2H  500 mg  0 mg  5g  Yes  S3  150 mg  250 mg  5g  No  S3H  150 mg  250 mg  5g  Yes  S4  500 mg  250 mg  5g  No  S4H  500 mg  250 mg  5g  Yes  S5  150 mg  500 mg  5g  No  S5H  150 mg  500 mg  5g  Yes  S6  500 mg  500 mg  5g  No  S6H  500 mg  500 mg  5g  Yes  1  'Sample vials were heated in water bath at 98 °C for 30 minutes. 2  150 mg S B F and 500 mg S B F were also expressed as low S B F and high S B F as well as L -  SBF and H - S B F throughout this thesis. 3  0 mg of SPI, 250 mg SPI and 500 mg SPI were also expressed as no SPI, Medium SPI, and  high SPI as well as N-SPI, M - S P I , and H-SPI throughout this thesis.  123  Table 4.4. Summary o f analyses o f variance with F values, mean squares of error (MSE), and degrees o f freedom (df) for main effects and their interactions for each o f the five attributes.  F values  Sample (S) (df = 11)  Panelist (P) (df=7)  Roasted  21.06***  2.33*  Soymilk  17.31***  Cereal  Beefy  Yeasty  Replication (R) (df = 1)  MSE SxP (df=77)  PxR (df=7)  SxR (df = 11)  3.35  1.19  0.70  0.86  6.42  6.94***  0.09  1.18  0.67  3.16  5.19  14.18***  7.42***  2.49  0.96  0.86  2.32  15.74  1.54  4.34"  1.24  1.05  1.26  6.66  2.02  1.44  0.89  0.90  7.27  8.53***  11.66**"  s1  6.28  *, **, and ***, significant at p < 0.05, p < 0.01, and p < 0.001, respectively.  124  Table 4.5. Adjusted F values of the sample effects using mean squares of sample by replication instead of mean squares of error for each of the five attributes.  Original F value  Adjusted F value  Attributes  MSsample  MSerror  Roasted  135.24  6.42  5.52  21.06***  24.49***  Soymilk-like  89.94  5.19  16.41  17.31***  5.48**  Cereal  88.97  6.28  14.53  14.18***  6.12**  Beefy  104.77  6.66  8.39  15.74***  12.48***  Yeasty  61.98  7.27  6.57  8.53***  9 44***  MS plexrepIication sam  ** and ***, significant at p < 0.01, and p < 0.001, respectively.  125  Table 4.6. Results o f Duncan's multiple comparison test on mean sensory scores of each panelist for 5 attributes .  Roasted Panelist  Mean score  Soymilk-like  Cereal  Beefy  Yeasty  Panelist  Mean score  Panelist  Mean score  Panelist  Mean score  Panelist  Mean score  8  6.36 a  4  2.55 a  4  3.17a  5  5.98 a  4  2.63 a  6  6.41 ab  3  4.15 b  6  3.82 ab  6  6.79 ab  6  3.38 b  5  7.12 abc  7  4.90 bc  3  4.13 abc  3  7.41 ab  8  5.31 c  2  7.58 abc  6  5.13 bc  7  5.27 bed  2  7.47 ab  5  5.95 cd  3  7.77 abc  8  5.30 bc  8  5.52 cd  4  7.49 ab  2  6.76 cde  4  8.09 bc  2  5.83 cd  1  5.96 de  7  7.81 b  1  7.22 de  1  8.26 c  1  5.98 cd  2  6.45 de  1  7.90 b  3  7.80 e  7  8.59 c  5  6.96 d  5  7.58 e  8  7.95 b  7  8.31 e  'Mean scores for each attribute within a column with different letters are significantly different (p < 0.05) using Duncan's multiple comparison test. Scores are listed in ascending order (n=24; 12 samples with 2 replications).  ON  Table 4.7. Pearson correlation coefficients and probabilities of sample means for each sensory attribute between individual panelist and the other 7 panelists . 1  Attributes Panelist  Panelist 1  Panelist 2  Panelist 3  Panelist 4  Panelist 5  Panelist 6  Panelist 7  Panelist 8  Roasted  Soymilk-like  Cereal  Beefy  Yeasty  Coefficient  0.914  0.918  0.904  0.735  0.835  (P)  (0.000)  (0.000)  (0.000)  (0.006)  (0.001)  Coefficient  0.873  0.867  0.954  0.878  0.850  (P)  (0.000)  (0.000)  (0.000)  (0.000)  (0.000)  Coefficient  0.965  0.884  0.869  0.953  0.870  (P)  (0.000)  (0.000)  (0.000)  (0.000)  (0.000)  Coefficient  0.419  0.680  0.375  0.228  -0.098  (P)  (0.175)  (0.015)  (0.230)  (0.476)  (0.774)  Coefficient  0.726  0.670  0.517  0.493  0.478  (P)  (0.008)  (0.017)  (0.085)  (0.103)  (0.116)  Coefficient  0.944  0.681  0.777  0.904  0.137  (P)  (0.000)  (0.015)  (0.003)  (0.000)  (0.671)  Coefficient  0.907  0.845  0.867  0.896  0.875  (P)  (0.000)  (0.001)  (0.000)  (0.000)  (0.000)  Coefficient  0.808  0.744  0.820  0.896  0.470  (P)  (0.001)  (0.006)  (0.000)  (0.000)  (0.123)  ' B o l d letters indicate Pearson correlation coefficients greater than 0.600, which are statistically significant at p < 0.05 (n=12; means of 12 samples).  127  Table 4.8. F ratio and probability o f replication effect on the 5 attributes for individual panelist . 1  Attributes Panelist  Panelist 1  Panelist 2  Panelist 3  Panelist 4  Panelist 5  Panelist 6  Panelist 7  Panelist 8  Roasted  Soymilk-like  Cereal  Beefy  Yeasty  F ratio  0.13  0.34  0.00  0.16  0.28  (P)  (0.717)  (0.567)  (0.963)  (0.693)  (0.599)  F ratio  0.15  0.20  0.16  0.09  0.01  (P)  (0.698)  (0.657)  (0.691)  (0.768)  (0.934)  F ratio  0.73  0.07  0.06  1.17  0.69  (P)  (0.403)  (0.795)  (0.812)  (0.292)  (0.414)  F ratio  1.16  0.15  5.06  0.47  0.80  (P)  (0.294)  (0.706)  (0.036)  (0.502)  (0.383)  F ratio  0.01  2.23  0.75  2.99  3.04  (P)  (0.928)  (0.150)  (0.395)  (0.098)  (0.095)  F ratio  0.04  0.01  0.36  0.24  2.51  (P)  (0.847)  (0.918)  (0.558)  (0.632)  (0.128)  F ratio  0.58  0.14  0.26  1.34  0.40  (P)  (0.454)  (0.713)  (0.612)  (0.259)  (0.533)  F ratio  0.59  0.07  0.37  0.74  1.22  (P)  (0.451)  (0.788)  (0.549)  (0.399)  (0.2810  ' B o l d letters indicate statistically significant replication effects at p < 0.1 (n=24; 12 samples with 2 replications).  128  Table 4.9. Pearson correlation coefficients and probabilities between mean sensory score for each sensory attribute and the concentration o f S B F and SPI.  SBF  SPI  Coefficient  0.662*  -0.692*  (P)  (0.019)  (0.013)  Coefficient  -0.634*  0.714**  (P)  (0.027)  Coefficient  -0.638*  0.711**  (P)  (0.026)  (0.010)  Coefficient  0.626*  -0.696*  (P)  (0.030)  (0.012)  Coefficient  0.668*  -0.659*  (P)  (0.017)  (0.020)  Attributes  Roasted  Soymilk-like  Cereal  Beefy  Yeasty  (0.009)  * and **, significant at p < 0.05 and p < 0.01, respectively (n=12; means of 12 samples).  Table 4.10. The mean intensity values o f the 5 attributes for the 12 mixtures of SBF and SPI in descriptive sensory evaluation . 1  Roasted  Soymilk-like  Sample  Mean score  Sample  S5H  1.93 a  S2H  S5  2.66 a  S2  S3H  3.46 ab  S3  Cereal  Mean score 1.85  a  Sample  Beefy Mean score  Sample  Yeasty Mean score  Sample  Mean score 2.80 a .  S2H  1.73  a  S5  2.54 a  S5H  2.53 ab  S2  2.32  ab  S5H  2.69 a  S5  S4H  2.88 ab  S4H  2.33  ab  S3H  3.73 a  S3H  5.69 be  S4  3.25 ab  S4  3.57  ab  S3  6.71 b  S3  4.17  ab  S6H  6.83 cd  SI  3.52 ab  SI  3.65  ab  S6H  7.19 b  S6  6.32  be  S6  7.39 cde  S1H  4.02 ab  S1H  3.73  ab  S6  7.72 be  S6H  6.97 cd  S1H  8.53 def  S6H  4.73 be  S6H  4.69  be  SI  8.47 bed  SI  7.50 cd  3.11  a  3.20 a  SI  9.63 ef  S6  6.25 cd  S6  6.19  cd  S4H  9.92 cd  S4  7.97 cd  S4  10.23 fg  S3  6.73 cd  S3  6.84  cde  S4  9.93 cd  S1H  8.43 cd  S4H  10.63 fg  S3H  7.87 de  S3H  7.91  def  S1H  10.13 cd  S4H  8.43 cd  S2  10.82 fg  S5H  8.98 e  S5H  9.25  ef  S2H  10.78 d  S2  9.29 d  S2H  12.14 g  S5  9.98 e  S5  9.99  f  S2  10.88 d  S2H  9.41 d  'Mean scores for each attribute within a column with different letters are significantly different (p < 0.05) using Duncan's multiple comparison test (n=12; 6 panelists with 2 replications). Scores are listed in ascending order. Refer to Table 4.3 for the sample code.  Table 4.11. The 15 selected indicator peaks for the simulated beef flavour. Indicator peak  RT  IP1  1.717  IP2  Detection frequency  (%)  Pearson correlation coefficient for beefy attribute  3-Methyl furan (PI)  50  0.710*  7.832  2-Acetyl furan (P7)  75  0.658*  IP3  12.046  delta-3-Carene(P17)  88  0.872***  IP4  15.772  2-Ethyl-3,6-dimethyl pyrazine (P24)  88  0.601*  IP5  16.133  Unknown  100  0.852***  IP6  16.864  Unknown  100  0.827**  IP7  18.199  Unknown  75  0.605*  IP8  19.744  2,3-Diethyl-5-methylpyrazine (P30)  50  0.590*  IP9  22.062  Decanal (P35)  75  0.715**  IP10  24.103  2-Isoamyl-6-methylparazine (P38)  50  0.599*  IP11  25.374  Unknown  75  0.598*  IP12  25.703  Unknown  75  0.794**  IP13  28.204  delta-Elemene (P44)  50  0.931***  IP14  30.291  beta-Cubebene (P49)  63  0.850***  IP15  35.680  Calamenene (P71)  75  0.883***  Peak identification (ID N o . )  1  2  Peak numbers (P#) and tentative identification by G C - M S analysis as reported previously (Moon et al., 2006). frequency o f detection by 8 panelists at a sniffing port in G C - 0 as reported previously (Moon et al., 2006). *, **, and ***, significant at p < 0.05, p < 0.01, and p < 0.001, respectively.  4.5. References  Anderson, J. W., Johnstone, B . M . , and Cook-Newell, M . L . (1995). Meta-analysis of the effects of soy protein intake on serum lipids. The New England Journal of Medicine, 333(5), 276282. Blanch, G . P., Mar-Caja, M . , Leon, M . , and Herraiz, M . (2000). Determination of (E)-5methylhept-2-en-4-one in deodorised hazelnut o i l . Application to the detection o f adulterated olive oils. Journal of the Science of Food and Agriculture. 80, 140-144. Boatright, W . L . , and L e i , Q. (1999). Compounds contributing to the "beany" odor o f aqueous solutions of soy protein isolates. Journal of Food Science, 64(4), 667-670. Chiesa, L . M . , Radice, L . , Belloli, R., Renon, P., and Biondi, P. A . (1999). Gas chromatographic determination o f galactose in milk. Example o f a switching valve used for the protection o f the capillary column. Journal of Chromatography A, 847, 47-51. Fournier, D . B . , Erdman, J. W . Jr., and Gordon, G . B . (1998). Soy, its components, and cancer prevention: a review o f the in vitro animal and human data. Cancer Epidemiology Biomarkers and Prevention, 7, 1055-1065. Gremli, H . A . (1974). Interaction o f flavor compounds with soy protein. The Journal of the American Oil Chemists' Society, 51, 95A-97A. Inouye, K . , Shiihara, M . , Uno, T., and Takita, T. (2002). Deodorization o f soybean proteins by enzymatic and physicochemical treatments. Journal of Agricultural and Food Chemistry, 50(6), 1652-1658. Jeong, S. - Y . , Chung, S. -J., Suh, D . -S., Suh, B . - C . , and K i m , K . - O . (2004) Developing a descriptive analysis procedure for evaluating the sensory characteristics o f soy sauce. Journal of Food Science, 69, S319-S325. Jorgensen, L . V . , Huss, H . H . , and Dalgaard, P. (2001). Significance o f volatile compounds produced by spoilage bacteria in vacuum-packed cold-smoked salmon (Salmo salaf) analyzed by G C - M S and multivariate regression. Journal of Agricultural and Food Chemistry, 49,2376-2381. Kasahara, K . (2004). Antioxidative effect o f 9 spices on volatile components in boiled sardines. Journal of the Japanese Society for Food Science and Technology, 51(2), 98-101. Labbe, D., Rytz, A . , and Hugi, A . (2004). Training is a critical step to obtain reliable product profiles in a real food industry context. Food Quality and Preference, 15, 341-348. Lei, Q., and Boatright, W . L . (2001). Compounds contributing to the odor o f aqueous slurries of soy protein concentrate. Journal of Food Science, 66(9), 1306-1310. 132  Maheshwari, P., O o i , E . T., and Nikolov, Z . L . (1995). Off-flavor removal from soy-protein isolate by using liquid and supercritical carbon dioxide. The Journal of the American Oil Chemists' Society, 72(10), 1107-1115. Malcolmson, L . J., and McDaniel, M . R. (1987). Flavor protein interactions in a formulated soup containing flavored soy protein. Canadian Institute of Food Science and Technology, 20(4), 229-235. May, C . G (1974). A n introduction to synthetic meat flavour. Food Trade Review, 44(1), 7, 9-10, 12-14. McDaniel, M . R., and Chan, N . (1988). Masking of soy protein flavor by tomato sauce. Journal of Food Science, 53(1), 93-101. Moon, S.-Y., Cliff, M . A . , and Li-Chan, E . C . Y . (2006). Odour-active components of simulated beef flavour analyzed by solid phase microextraction and gas chromatography-mass spectrometry and -olfactometry. Food Research International, 39, 294-308. Moon, S.-Y., and Li-Chan, E . C . Y . (2004). Development o f solid-phase microextraction methodology for analysis o f headspace volatile compounds in simulated beef flavour. Food Chemistry, 88, 141-149. Narzip, L . , Miedaner, FL, and Lustig, S. (1999). The behaviour of volatile aromatic substances as beer ages. Monatsschrift fur Brauwissenschaft, 52(9/10), 164-175. O'Mahony, M . (1986). Fixed- and random- effects models. In S. R. Tannenbaum and P. Walstra (Eds.), Sensory Evaluation of Food - Statistical Methods and Procedures, (pp. 247-257). New York and Basel: Marcel Dekker, Inc. Puppo, E., Chapleau, N . , Speroni, F., de Lamballerie-Anton, M . , Michel, F., Anon, C , and Anton, M . (2004). Physicochemical modifications of high-pressure-treated soybean protein isolates. Journal of Agricultural and Food Chemistry, 52, 1564-1571. Renkema, J. M . S. and van Vliet, T. (2002). Heat-induced gel formation by soy proteins at neutral p H . Journal of Agricultural and Food Chemistry, 50, 1569-1573. U.S. Food and Drug Administration. (1999). Food labeling, health claims, soy protein and coronary heart disease. Federal Regulation, 57, 699-733. Wasserman, A . E . (1979). Chemical basis for meat flavor: A review. Journal of Food Science, 44, 6-11. Wolf, W. J. (1975). Lipoxygenase and flavor of soybean protein products. Journal of Agricultural and Food Chemistry, 23(2), 136-141. Zook, K . L . , and Pearce, J. H . (1988). Quantitative descriptive analysis. In H . Moskowitz (Ed), Applied Sensory Analysis of Foods, (pp. 43-71). Boca Raton: C R C Press, Inc. 133  CHAPTER  5.  ASSESSMENT OF ADDED INGREDIENT EFFECT ON THE INTERACTION OF SIMULATED BEEF FLAVOUR AND SOY PROTEIN ISOLATE BY GAS CHROMATOGRAPHY AND SPECTROSCOPIC TECHNIQUES 4  5.1. Introduction  Soy protein has gained increasing attention as a promising protein source with various health benefits in addition to functional properties. Due to the problematic retention o f indigenous offflavour, which has been attributed primarily to lipid oxidation products arising from the action of lipoxygenases present in soybean, much research has been conducted to investigate the interaction of soy protein with model compounds including aldehydes, ketones and alcohols. In addition to investigating indigenous, undesirable off-flavour related compounds o f soy products, research is required to elucidate the binding and release o f intentionally added, desirable flavour components by soy protein. For example, this type of interaction is important in formulation of soy-based products with simulated meat flavours for the vegetarian market (Malcolmson and McDaniel, 1987).  Interaction o f flavour compounds with soy proteins in model systems has been reported (Damodaran and Kinsella, 1981a and 1981b; Gremli, 1974). The types and degree of interaction between flavours and soy protein are dependent on characteristics o f the flavour compounds in addition to the environment o f the components. Wilson (1985) proposed that hydrogen bonding was involved in flavour binding to soy proteins in non-aqueous systems. Damodaran and Kinsella (1981a) reported that the interaction o f carbonyls with soy proteins in the native state was due to hydrophobic forces, and that the binding affinity for 2-nonanone was increased by partial denaturation o f soy protein, which might enhance the hydrophobicity o f the binding site via reorganization of the subunits. Gremli (1974) reported that aldehydes, especially unsaturated compounds, reacted more strongly with the proteins than ketones while alcohols did not interact with soy protein. N o binding affinity of carboxylic acid (Beyeler and Solms, 1974) or hexane A version of this chapter will be submitted for publication. Moon, S. - Y . , and Li-Chan, E . C. Y . Assessment o f added ingredient effect on the interaction o f simulated beef flavour and soy protein isolate by gas chromatography and spectroscopic techniques. The Journal of the American Oil Chemists' Society. 4  134  (Thissen, 1982) with soy protein was reported. Beyeler and Solms (1974) indicated that binding properties of model flavour compounds including ketones and aldehydes with soy protein isolate (SPI) were rather independent of pH and temperature, while binding of vanillin with soy protein was decreased with increasing temperature (Li et al., 2000). Moreover, Zhou and Cadwallader (2006) reported that hydrocarbons showed weak interaction with SPI whereas alcohol compounds strongly interacted with SPI, suggesting that hydrogen bonding was involved. Ester, ketone and aldehyde compounds interacted with SPI with similar forces and Zhou and Cadwallader (2006) proposed that hydrogen bonding, dipole forces, and van der Waals dispersion forces might be involved. Some conflicting data in terms of binding properties of flavour compounds to soy proteins were found in the literature, which may be attributed to the differences in methods used, origin and degree of denaturation of soy protein material during preparation and the experimental conditions applied in addition to different composition of buffer, which could affect the interaction between the protein and flavour compounds (O'Keefe et al., 1991).  Most of the studies conducted to date have used model systems of single ingredients or selected volatile model compounds such as series of aldehydes, ketones, alcohols, or alkanes. Although valuable thermodynamic information such as binding affinity or number of binding sites was acquired from the results of these studies, the knowledge may not be directly applicable to the real food system, in which the flavour ingredients usually contain intricate combinations of various subclasses of compounds. Soy protein could also bind with certain desirable flavour compounds. Depending on the nature and the strength of the binding, each aroma compound could be released at a different rate from SPI, which could have an impact on flavour suppression or alteration of flavour profiles in the final food products due to changes in the odour balance. Therefore, for the development of soy products with acceptable flavour quality it is imperative to elucidate the nature of the interactions of soy proteins with the mixture of compounds present in the actual flavour ingredients.  Recently, research has been conducted to elucidate the impact of SPI on aroma characteristics of a commercially produced simulated beef flavour (SBF) ingredient. By using headspace solid phase micro-extraction coupled with gas chromatography (GC), GC-mass spectrometry (GC-MS) 135  and GC-olfactometry (GC-O) analysis, the odour-active components were recognized, and several high aroma impact compounds from the volatile compounds in S B F were identified (Moon et al., 2006). Descriptive analysis ( D A ) and G C analysis were conducted to assess the aroma characteristics o f S B F in mixtures with SPI, and the volatile compounds in S B F that were related to a "beefy" note were identified. Several indicator peaks were selected to represent the beefy aroma notes. Addition o f SPI resulted in decreasing o f "beefy", "roasted", and "yeasty" aroma notes in the samples (Chapter 4 o f this thesis).  In this study, the interaction between SPI and S B F was examined by addition of compounds intended to produce changes in the potential forces which may affect SPI conformation. Specifically, alteration of several types o f interaction was carried out by adding a third component in order to explore the effect o f conformational changes o f SPI on its flavour holding capacity, particularly with regard to beef attribute related aroma compounds in S B F . Four components, i.e. glucosamine, sucrose, ascorbic acid, and polyethylene glycol, were selected based on their functional characteristics as food ingredients or food additives and potential to alter hydrogen bonding, hydrophobic interactions, and disulfide bonds in SPI. Sucrose (P-Dfructofuranosyl-a-D-glucopyranoside) is a disaccharide containing poly-hydroxyl groups, while glucosamine (2-amino-2-deoxy-D-glucose) is an amino sugar with amino as well as hydroxyl groups. Both sucrose and glucosamine can serve as hydrogen donors for hydrogen bonding through the hydroxyl group (-OH) and/or amino group (-NH ). Polyethylene glycol is a polymer 2  of ethylene oxide which may affect hydrophobicity of proteins through the repeating hydrophobic monomeric unit ( - C H - C H - ) . It can also act as a hydrogen bond donor but more likely serves as 2  2  a hydrogen acceptor, due to the - O H group and unshared electron pairs on the oxygen atom, respectively. Ascorbic acid is an organic acid with antioxidant properties. It is easily oxidized to dehydroascorbic acid, which is relatively stable. The application o f ascorbic acid as a reducing agent in biological systems was reported (Kashiba-Iwatsuki et al., 1997). Ascorbic acid may also induce disulfide-sulfhydryl interchange reactions through cleavage and reformation o f disulfide bonds (Dong and Hoseney, 1995). Moreover, it is known that reducing agents such as cysteine, ascorbic acid, P-mercaptoethanol, and dithiothreitol reduce disulfide cross-links and hence modify the conformation o f proteins (Cheftel et al., 1985). Therefore, ascorbic acid could affect the status o f disulfide bonds or sulfhydryl groups of SPI, which may lead to conformational  136  changes in the protein. In addition to the effect of each ingredient, the combined effect of ascorbic acid and polyethylene glycol on SPI was also investigated in this study to examine possible additive or synergistic effects since the reduction of disulfide bonds may enhance conformational changes induced by hydrophobic interactions, which generally tend to be located on interior part of proteins.  Among parameters affecting protein structures, hydrophobicity was reported to be extensively related to the functional properties of proteins (Nakai, 1983). Therefore afluorescenceprobe method using 6-propionyl-2(N,N-dimethyl-amino)naphthalene (PRODAN) was conducted to determine surface hydrophobicity of SPI affected by the ingredients, by measuring changes in fluorescence upon binding of the probe to accessible hydrophobic regions of the protein in aqueous solution. Compared to the anionic fluorescent probes such as l-anilinonaphthalene-8sulfonic acid (ANS) and cis-parinaric acid (CPA), PRODAN is a neutral probe, which could minimize possible contribution of electrostatic interaction between fluorescent probe and protein (Alizadeh-Pasdar and Li-Chan, 2000). In addition, conformational changes of SPI by addition of the ingredients were assessed by FT-Raman spectroscopy, which has been reported as a useful tool to investigate in situ protein structural changes under various conditions relevant to processing (Li-Chan, 1996).  Therefore, the objectives of the study in Chapter 5 were (a) to investigate the changes detected by GC on theflavourholding behavior of the SPI in terms of odour-active compounds contributing to beefy notes in SBF as affected by the ingredients such as glucosamine, sucrose, ascorbic acid, and polyethylene glycol, (b) to monitor changes in the protein structure in SPI induced by the ingredients by means of disulfide and sulfhydryl group measurement, fluorescence probe and FTRaman spectroscopy to understand the effect of each ingredient on the holding capacities of beef volatile flavour components in samples containing SPI and (c) to identify sensory impact on the mixture of SBF and SPI treated ingredient. The results from this study may provide a foundation for better understanding of the interaction between SPI and SBF.  137  5.2. Experimental methods 5.2.1. Materials Commercially available simulated beef flavour (SBF; Mastertaste, Arlington Heights, IL) and soy protein isolate (SPI; Solae, St. Louis, M O ) were used, as described in Section 2.2.1 and Section 4.2.1, respectively. Sucrose, ascorbic acid (min. 99 %), glucosamine (min. 99 %), and polyethylene glycol (average M w . 8,000) were purchased from Sigma Chemical Co. (St. Louis, M O ) and spectral-grade methanol was from Fisher Scientific (Fairlawn, NJ). The fluorescent probe, 6-propionyl-2-(dimethylamino) naphthalene ( P R O D A N ) , and Ellman's reagent, 5,5'dithio-bis-2-nitrobenzoic acid ( D T N B ) were obtained from Sigma (Sigma Chemical Co., St. Louis,  MO).  The  solid  phase  assembly  holder,  50/30  pm  stableflex  divinylbenzene/carboxen/polydimethylsiloxane ( D V B / C A R / P D M S ) , 15 m L capacity G C sample vials and polypropylene hole cap with PTFE/silicone septa were purchased from Supelco (SigmaAldrich Canada, Oakville, O N ) .  5.2.2. Differential scanning calorimetry (DSC) The thermal behavior of SPI was investigated by a multi-cell differential scanning calorimeter thermal analyzer (model 4207 M C - D S C , Calorimetry Science Corp). Three concentrations of SPI (0.2 %, 5 %, and 10 %) were prepared in p H 7.5 phosphate buffer (3.88 m M N a H P 0 , 15.5 m M 2  4  N a H P 0 ) . Approximately 0.05 g of each sample was placed in Hastelloy C ampoules, which 2  4  were sealed then weighed accurately. A sealed empty cell was used as a reference. For D S C analysis, each sample was held at 25 °C and equilibration time was set at 600 seconds. Heating scan was conducted from 25 °C to 105 °C at a rate of 1 °C/min for each sample.  5.2.3. Gas chromatography For headspace solid phase microextraction (HS-SPME),  100 mg o f each ingredient i.e.  glucosamine, sucrose, polyethylene glycol, or ascorbic acid, was first mixed with 250 mg SPI in a 15 m L capacity G C sampling vial with a magnetic stirring bar. Following addition of 5 g buffer (0.05 M Tris-HCl buffer p H 7.4), the vial was incubated at either room temperature (RT; 23 °C) or 60 °C for 20 minutes to compare potential changes due to heat treatment. To investigate the effects of a mixture of polyethylene glycol and ascorbic acid, the sample was first incubated with  138  100 mg o f ascorbic acid for 20 minutes at R T or 60 °C, and then 100 mg of polyethylene glycol was added and the mixture was incubated for another 20 minutes at the same temperature before adsorption.  SBF (500 mg) was added to the vial containing the incubated SPI and ingredients, and the vial was tightly capped with a polypropylene hole cap with a PTFE/silicone septum. A s a control, a sample containing SPI and S B F was prepared without any additional ingredients and analyzed under the same condition. A sample with only S B F (no SPI) was also analyzed.  Stirring with a  magnetic stirring bar was consistently applied.  To  extract  headspace  volatile  compounds, adsorption was performed  with  50/30 pm  D V B / C A R / P D M S S P M E fibre exposed to the headspace above the sample solution for 60 minutes at 60 °C in a thermostat controlled water bath ( ± 2 °C). Analysis o f the samples adsorbed at R T was also conducted to observe changes in volatile components induced by adding ingredients to SPI in the absence of heat. Analysis of volatile components was performed by gas chromatography - flame ionization detector as previously described by M o o n and Li-Chan (2004). The effect of each ingredient at each temperature condition was analyzed in triplicate with 3 independently prepared samples and the results were presented as the mean of the triplicate values. In addition, S B F alone without either SPI or ingredients as well as S B F with SPI but no ingredients were also analyzed in triplicate; the average coefficients of variation for individual peak areas o f these triplicates were 10.1 % and 10.6 %, respectively. The area of each individual peak in each sample was expressed as relative peak area (%) based on the peak area of SBF (100 %) for comparison purpose.  5.2.4. Sulfhydryl and disulfide content SPI (250 mg) and ingredient (100 mg o f glucosamine, sucrose, polyethylene glycol, or ascorbic acid) were placed in a 15 m L sample vial together with 10 m L p H 8.0 buffer (85 m M Tris, 100 m M glycine, 4 m M E D T A , 10 mg/mL SDS). The samples were heated in a water bath at 60 °C for 20 minutes with stirring. For the sample mixture of ascorbic acid and polyethylene glycol, SPI was first incubated at 60 °C for 20 minutes with ascorbic acid only, prior to addition of  139  polyethylene glycol and further incubation for another 20 minutes at 60 °C.  Ingredient blank  samples were prepared as above but without any SPI.  The content o f total sulfhydryl groups (SH) was determined with Ellman's reagent (5,5'-dithiobis-2-nitrobenzoic acid or D T N B ) (Ellman, 1959) by the method o f Beveridge et al. (1974) as modified by Chung et al. (2005). Freshly prepared Ellman's reagent solution (200 u L o f 4 mg/ml) was added to each tube containing 3 m L of sample and 3 m L o f the p H 8.0 buffer. The tubes were covered with aluminum foil and incubated in an oven at 40 °C for 15 minutes, then centrifuged at 14950 g for 30 minutes at 10 °C Absorbance at 412 nm o f each sample was measured  on  a  Shimadzu UV1700  UV/visible  spectrophotometer  (Shimadzu  Scientific  Instruments Inc., Columbia, M D ) . Net A 4 1 2 of sample was calculated as: Net A 4 1 2 = A 4 1 2 o f sample with Ellman's reagent - protein blank - ingredient blank, where protein blank contained only protein samples without Ellman's reagent and ingredient blank contained the ingredient in the buffer with the reagent but no protein. The sulfhydryl content was calculated based on an extinction coefficient o f 13,600 M " ' c m ' (Ellman, 1959), as 1  shown in the following equation: pmole SH/g protein = 73.53 x net A  4 1 2  (D/C)  [equation 11]  where D was the protein concentration in mg/mL and C was a dilution factor. Results were expressed as the mean o f duplicate analyses.  Determination o f total sulfhydryl and disulfide content (SH+SS) in the samples was conducted by the method o f Thannhauser et al. (1984), with incubation in the dark as suggested by Damodaran (1985). The 2-nitro-5-thiosulfobenzoate (NTSB) stock solution was prepared by dissolving 100 mg of D T N B in 10 m L o f 1 M sodium sulfite and adjusting to p H 7.5; oxygen was bubbled through the solution held in a water bath at 38 °C or 2 hours until the bright red colour turned into a pale-yellow colour. The N T S B stock solution was divided into 400 p L aliquots and stored frozen at -18 °C until used for analysis.  The N T S B stock solution was diluted in a ratio of 1 to 100 with a freshly made solution at p H 9.5 containing 2.0 M guanidine thiocyanate, 0.2 M Tris, 100 m M sodium sulfite and 3 m M E D T A to make N T S B assay solution. 200 \xL of protein sample in p H 8.0 buffer described above was 140  added to 3.0 mL of the NTSB assay solution. After incubating in the dark for 30 minutes followed by centrifugation at 14950 g for 30 minutes at 10 °C, the absorbance at 412 nm was recorded against a blank containing 3 mL of NTSB assay solution and 200 uL of dd-water. The content of (SH+SS) was calculated as described above in equation 11. The content of disulfide (SS) groups was determined by subtracting the content of total SH from the content of (SH+SS). Results were expressed as the mean of duplicate analyses.  5.2.5. Surface hydrophobicity Surface hydrophobicity of SPI samples was determined by the hydrophobic fluorescent probe method using PRODAN as described by Alizadeh-Pasdar and Li-Chan (2000). PRODAN stock solution was prepared by dissolving 3.2 mg PRODAN in 10 mL spectral-grade methanol; 1 mL aliquots in 1.5 mL Eppendorf tubes wrapped with aluminum foil were stored at -18 °C for future use. The concentration of the PRODAN stock solution was spectrophotometrically determined at 360 nm as 1.8  xlO"  3  M using a molar absorption coefficient  8360  of 1.8  x  10 M^cm" (Alizadeh4  1  Pasdar and Li-Chan, 2000). SPI stock solution was prepared to contain 1 % (w/v) SPI in Tris buffer (0.05 M, pH 7.4) with 0.02 % sodium azide. To prepare ingredient treated stock solutions, 10 mL SPI stock solution was mixed with the ingredient (40 mg of glucosamine, sucrose, polyethylene glycol, or ascorbic acid). Ingredient blank samples were prepared using buffer without any SPI. The samples were heated at 60 °C for 20 minutes. Additional heating at 60 °C for 20 minutes was employed after adding polyethylene glycol for the sample mixture of ascorbic acid and polyethylene glycol. After cooling down the samples in cold water for 15 minutes, serial dilution was performed to obtain 5 SPI concentrations ranging from 0.010 % to 0.0025 %.  For measurement of surface hydrophobicity, 10 uL PRODAN was added to 4 mL of each SPI dilution and vortexed. After incubating 15 minutes in the dark, the relative fluorescence intensity (RFI) of samples was measured by a Shimadzu RF-540 (Shimadzu Corp., Kyoto, Japan) spectrofluorometer.  The excitation/emission  wavelengths were 365  nm/465  nm and  excitation/emission slits were set at 5 nm/5 nm. All procedures dealing with PRODAN were conducted in a dark room due to its light sensitive nature and the analyses were carried out in triplicate. Standardization was performed to correct for day-to-day fluctuations of the instrument by measuring the RFI of 10 pL of PRODAN in 4 mL methanol and adjusting to a standard value 141  of 50. RFI was also measured for blanks, which contained SPI and/or ingredients but no PRODAN. Net RFI was obtained by subtracting the RFI of the blank sample without PRODAN from the RFI of the sample with PRODAN. The initial slope (So) of net RFI versus SPI concentration was calculated by linear regression analysis and reported as surface hydrophobicity of the sample.  5.2.6. FT-Raman spectroscopy Samples containing SPI with or without ingredient were prepared as described in Section 5.2.2. (GC analysis) and then freeze-dried due to the poor signal to noise ratio of the spectra of the aqueous samples observed in preliminary experiments. The freeze-dried samples and original (not freeze-dried) SPI powder were placed in NMR tubes and FT-Raman analyses were conducted using a Nexus 670 FT-Raman spectrometer equipped with CaF beam splitter and Ge 2  detector (Thermo Nicolet Corp., Madison, WI). The analyses for each sample were performed under the following conditions: laser power, 0.5 watt; scan number, 512; spectral resolution, 4 cm". The spectrum of polystyrene standard was used to adjust sample position for maximal 1  signal intensity to verify the system before analyzing samples. The spectrum of each ingredient was also analyzed and subtractedfromthe sample spectrum. Spectra of samples were obtained by OMNIC Version 6.0a (Thermo Nicolet Corp., Madison, WI). Spectral data processing including baseline correction, subtraction of ingredient spectra and normalization to the intensity of phenylalanine peak at 1003 cm" was performed using GRAMS/AI Version 7.02 (Thermo 1  Galactic, Galactic Industries Corp., Salem, NH). Assignments of the bands in the Raman spectra to the specific vibrational modes of amino acid chain or polypeptide backbone were based on the literature (Howell and Li-Chan, 1996; Li-Chan, 1996). Secondary structure analysis based on the amide I region (Williams and Dunker, 1981) using the Raman Spectral Analysis Package (RSAP) program (version 2.1) of Przybycien and Bailey (1989) was conducted with ingredient subtracted Raman spectral data. FT-Raman spectroscopic analyses were conducted in duplicate; coefficients of variation of peak intensities for duplicate spectra were 1 % or less.  5.2.7. Sensory evaluation Five panelists from the previous study (Chapter 4 in this thesis) and two additional panelists recruited from graduate students in the Food Science program at UBC and from staff at a food 142  company producing meat substitute products in Vancouver were invited to participate in descriptive sensory evaluation. To remind the panelists of the 5 sensory attributes selected during the previous training session, a 20 minute-training session was held using a hand-out with the previously defined attributes (Table 4.2). For the training session, three different samples (SBF only, SBF with SPI, and SBF with ascorbic acid treated SPI) were prepared. The sample incubation, serving, evaluating, and discussion methods as well as sensory evaluation sheet were as described in Chapter 4 of this thesis.  For the main descriptive sensory analysis, 3 different samples were prepared using the procedure described in section 5.2.3., as follows: 250 mg SPI in 5 g dd-water with 150 mg SBF but no ingredient added (SPIF), 250 mg ascorbic acid treated SPI in 5 g dd-water with 150 mg SBF (AF), and 250 mg ascorbic acid and polyethylene glycol treated SPI in 5 g dd-water with 150 mg SBF (APF). The 3 samples (coded with 3-digit random numbers) and a reference (250 mg SPI in 5 g dd-water with 150 mg SBF) were covered with aluminum foil and held at 60 °C for 20 minutes before serving. Refer to Section 4.2.2.2 of this thesis for detailed description of the sensory procedure.  5.2.8. Statistical analysis  General linear model of analysis of variance (ANOVA) was performed using Minitab software (version 13.30, Minitab Inc. PA) to determine different ingredient effects, and Fisher's least significant difference (LSD) test was performed to compare samples at the 95 % confidence level. For FT-Raman spectral data, the mean values from duplicate analyses were reported. To verify significant difference among samples, lower and upper 95 % confidence limits were calculated based on the assumption of 5 % coefficient of variation for the FT-Raman analysis (Badii and Howell, 2003).  143  5.3. Results 5.3.1. Differential scanning calorimetry The  t h e r m a l b e h a v i o u r o f the  SPI was  assessed using D S C between 25  1 ° C / m i n . T h e t h e r m o g r a m s o f S P I at t h r e e d i f f e r e n t c o n c e n t r a t i o n s phosphate changes  b u f f e r , p H 7.5  associated  are s h o w n  in Figure  5.1.  ° C and  (0.2  Generally, native  105  ° C at a rate  % , 5 % , a n d 10 % S P I i n  proteins  w i t h the rupture o f h y d r o g e n bonds or exothermic changes  show  endothermic  upon weakening  hydrophobic interactions and aggregation o f proteins by thermally i n d u c e d denaturation and  M a , 2002). T h e absence  the heating scans o f the  o f endothermic or exothermic peaks  SPI used  considerably denatured most  of  (Li-Chan  in the t h e r m o g r a m s f r o m a n y  in this study, suggests that the proteins  in the  of  SPI were  of  already  likely d u e to the treatment d u r i n g the p r o c e s s i n g o f S P I .  5.3.2. Gas chromatography Changes  in  influenced relative  areas by  %  of  peaks  addition  area  of  of  the  a n d at 6 0  which  significantly  to  peak  (p  <  0.05)  The  peaks  samples  8,  with  polyethylene  1,  4,  12,  glucosamine, glycol,  13  to  have  an  effect  under  these  each  in  investigated sample 5.1  S B F by  using  and Table  correlated with  in  the  presence  H S - S P M E  H S - S P M E 5.2.  G C  of  method.  adsorption  In these tables,  beefy  note  SPI,  at  only  in sensory  as  The room  peaks,  evaluation  and compared.  increased  polyethylene Although  a d s o r p t i o n at R T c a p t u r e d m u c h f e w e r interesting to note that g l u c o s a m i n e  were  in T a b l e  were  sucrose,  respectively.  characteristic  positively  shown  of  in  ° C are presented  (Chapter 4 o f this thesis), w e r e  areas  beefy  ingredients,  individual  temperature were  related  by  R T  glycol,  ascorbic  incubation  volatile  of  compounds  in terms  of  the  in the  and ascorbic acid alone  conditions  adsorption  60  and  mixtures headspace  ° C  was  the  beefy  not shown  adsorption  ascorbic  at  R T  o f the  or with polyethylene  alleviating  h o l d i n g p r o p e r t y o f S P I . H o w e v e r , the effect o f g l u c o s a m i n e  acid,  or  acid  with  followed samples,  glycol  related  in  by it  is  seemed  compounds  u n d e r the  condition  of 60 ° C incubation or 60 ° C adsorption.  Generally, affected  the  addition  of  ascorbic  acid  either  by  itself  flavour holding property o f SPI. Under  or  the  with  polyethylene  glycol  significantly  condition of R T adsorption (Table  5.1),  144  ascorbic acid significantly increased areas of 4 peaks (29.971, 30.729, 31.271, and 33.306 minutes) when added on its own, but it increased areas of 6 peaks (29.971, 30.729, 31.271, 31.881, and 33.306 minutes) when added with polyethylene glycol. Under 60 °C adsorption (Table 5.2), the areas of 8 peaks (1.734, 12.132, 13.915, 20.608, 25.794, 26.954, 30.377, and 31.317 minutes) were significantly increased by addition of ascorbic acid only while the areas of 8 peaks (1.734, 12.132, 13.915, 20.608, 30.377, and 31.317 minutes) were increased by ascorbic acid with polyethylene glycol. A number of the peaks which were increased by ascorbic acid, such as peaks with retention time of 29.971, 30.729, 31.981 and 33.306 minutes in Table 5.1 and 12.132, 30.377, and 31.317 minutes in Table 5.2, were highly positively related to the beefy note from SBF (p < 0.001). In addition, some of these peaks (e.g. with retention time at 1.734, 12.132, 25.794, and 30.377 minutes) were associated with high detection frequency (greater than 50 %) in GC-O and may therefore be expected to have an impact on perceived aroma profile in the mixture. Fifteen peaks had been selected as indicator peaks in Chapter 4 based on their potential contribution to beefy note in the mixture. Among the 15 peaks, the areas of 2, 4 and 4 peaks were significantly increased in samples containing polyethylene glycol, ascorbic acid, and ascorbic acid with polyethylene glycol, respectively (Table 5.2).  5.3.3. Sulfhydryl and disulfide content The contents of total SH and disulfide in each sample are shown in Table 5.3. The results should be considered with caution as high protein blanks were observed due to the turbidity of the protein in the testing buffer for these assays, even after being centrifuged at 14950 g for 30 minutes. The total SH content of SPI was 0.6 umole/g protein. No or only small differences in the content of total SH were observed by addition of glucosamine, sucrose, and polyethylene glycol to SPI, while considerable difference resulted in the presence of ascorbic acid, or ascorbic acid with polyethylene glycol. In addition, the content of SS was determined as 36.7 umole/g protein and a significant decrease of disulfide content was observed in samples containing ascorbic acid and ascorbic acid with polyethylene glycol.  5.3.4. Surface hydrophobicity The surface hydrophobicity of SPI and ingredient added SPI determined by the neutral fluorescent probe, PRODAN, are shown in Table 5.4.  The So of SPI was determined to be 145  145 %" and was not affected by addition of glucosamine or sucrose. However So was significantly increased by addition o f polyethylene glycol or ascorbic acid as shown in Table 5.4. The increase in So was amplified in the sample treated by the combination o f ascorbic acid and polyethylene glycol implying additive effects o f ascorbic acid and polyethylene glycol in exposing surface hydrophobic groups in SPI.  5.3.5. FT-Raman spectroscopy A typical FT-Raman spectrum of SPI (freeze-dried powder from 5 % w/w of SPI solution in 0.05 M Tris-HCl buffer p H 7.4) is shown in Figure 5.2. Effects of additional ingredients on the F T Raman spectrum are depicted in Figure 5.3. The tentative assignments of the major bands are listed in Table 5.5.  Comparison of the tryptophan band near 760 cm" in each sample spectrum shows a significant 1  decrease in peak intensity in the sample containing ascorbic acid with polyethylene glycol compared to SPI, as shown in Table 5.6. Further changes in aromatic amino acid residues may be noted in the intensity ratio o f the doublet bands at 850 and 830 cm" , which are assigned to 1  tyrosine residues. There were significant decreases in the tyrosine doublet ratio for SPI in the presence o f ascorbic acid and ascorbic acid with polyethylene glycol compared to SPI alone (Table 5.6). The microenvironment o f aliphatic amino acid residues was examined by Raman bands at 1450 cm" ( C - H bending), 1465 cm" ( C - H bending) or 2935 cm" ( C - H stretching). 1  1  1  2  Although there were no significant differences in the intensities at 2935 cm" , a significant 1  increase in the intensity at 1450 cm" was observed in samples containing polyethylene glycol 1  and ascorbic acid with polyethylene glycol. The SS-stretching band near 510-550 cm" clearly 1  indicated an effect o f addition of ingredients on the disulfide bonds, as shown in Table 5.6. Significant increase in intensity at 540 cm" was shown in the sucrose added sample while 1  significantly decreased intensity at 537 and 539 cm" was observed in samples containing 1  ascorbic acid and ascorbic acid with polyethylene glycol, respectively.  Among several distinct vibrational modes o f the - C O - N H - amide or peptide bond, the amide I and III bands are the most useful to determine the secondary structure of proteins (Li-Chan, 1996). In this study, compositions o f the secondary structure for SPI and SPI with added 146  ingredient were analyzed by least squares analysis of the Amide I band. The secondary structure compositions of the original (not freeze-dried) SPI powder (NOFD) and freeze-dried SPI (SPI) were very similar, as shown in Table 5.7. There was no significant difference in the fraction of ahelix, p-reverse turn and unordered structure, while a slightly increased proportion of p-sheet was observed in SPI after freeze-drying. The added ingredients led to considerable changes in the secondary structure of SPI. When SPI was treated with glucosamine, the proportion of P-sheet increased while total unordered structure decreased. Sucrose seemed to have the least effect on changing the secondary structure of SPI, with the only difference being a decrease in unordered structure. Addition of polyethylene glycol caused an increase in the proportion of a-helix and a decrease in P-sheet compared to SPI. The greatest change in terms of propensity of secondary structure of SPI was detected by addition of ascorbic acid, observed in both the amide I (1650-1685 cm") and amide III (1235-1305 cm") 1  1  bands (Figure 5.3). Table 5.7 shows that the a-helix and unordered structure were significantly increased by more than 5 % and 12 %, respectively, while p-reverse turn was drastically decreased by about 17 % when SPI was treated with ascorbic acid. Similar changes occurred in SPI treated by ascorbic acid with polyethylene glycol, but to a lesser extent. Considerable differences in the unordered structure were found between SPI and ingredient added SPI. The fraction of unordered structure was significantly increased in the samples containing ascorbic acid or ascorbic acid with polyethylene glycol, and considerable decrease was observed in samples containing sucrose or glucosamine, while no significant difference was shown in samples containing polyethylene glycol.  5.3.6. Sensory evaluation The mean sensory scores of 7 panelists for the 5 sensory attributes and the results of Fisher's LSD tests are shown in Table 5.8. The comparison of the 5 attributes for the 3 samples, (i.e. SPI without any ingredient added (SPIF), SPI containing ascorbic acid (AF) and SPI containing ascorbic acid with polyethylene glycol (APF)) is illustrated in Figure 5.4.  The SPI containing ascorbic acid with polyethylene glycol showed significantly higher score for roasted attribute and significantly lower scores for soymilk-like and cereal attributes compared to 147  SPI without any ingredient. A higher score for beefy attribute and lower score for yeasty attribute were also noted for SPI containing ascorbic acid with polyethylene glycol, but these differences were not statistically significant (p > 0.05). In general the intensities of SPI with ascorbic acid for all the sensory attributes except yeasty note were located between those of SPI without ingredient and of SPI containing ascorbic acid with polyethylene glycol. 5.4. Discussion 5.4.1. Changes in peak area by the ingredients in GC analysis The effect of each added ingredient on beefy flavour retention in SPI with regard to changes in beefy characteristics was investigated using GC-FID. Through preliminary work, the ratio of SPI to ingredient was adjusted from 50:1 to 2.5:1. This ratio was required in order to detect changes in the area of individual peaks in the GC chromatogram, beyond the signal noise and above the area reject limit value of 1000. Although this ratio was rather high by comparison to the real food system, the results could provide useful information to monitor the effects of different types of ingredients, which could affect binding property of SPI with flavour compounds. The flavour holding capacity should be distinguished from flavour binding property, where investigation of binding affinity and number of binding sites on SPI with individual flavour component (ligand) are focused.  The ability of SPI to suppress release of beefy related volatile compounds in the headspace, as indicated by the GC peak areas, was noticeably reduced in SPI containing glucosamine at RT incubation and RT adsorption condition, as shown in Table 5.1. However, the effect of glucosamine disappeared under the condition of 60 °C incubation or 60 °C adsorption. The observed result may be attributed to the changes in hydrogen bonding interactions of the SPI affected by the hydrogen in the amino group of glucosamine. The energy of hydrogen bond was reported as 8-40 kJ/mol and it is readily weakened by heating while hydrophobic interaction is enhanced (Cheftel et al., 1985). Ionic interactions between SPI and glucosamine were not considered to contribute significantly in this study where Tris buffer at pH 7.4 was used, since these forces would only be expected to play an important role in the pH range between 4.6 and 148  p H 6.9, i.e. between the p i o f SPI and p K a of glucosamine where SPI would be negatively charged and glucosamine would be positively charged.  Moreover, addition of ascorbic acid alone or with polyethylene glycol showed a potential to increase the areas of beef attribute related flavour compounds and some o f the peaks were significantly recovered toward the level found in S B F (Table 5.1 and Table 5.2). Although these results cannot provide information on whether the increase o f the peak areas was due to alleviating the binding constants between the flavour compounds and SPI, decreasing the number of binding sites for the flavour compounds on SPI, or reducing binding affinity due to possible changes in the flavour compounds by the ingredients, it clearly demonstrated that among the ingredients added, ascorbic acid had the biggest effect on flavour holding behaviour, in terms of decreasing suppression of beef attribute related aroma components by SPI in the mixture. Therefore, it can be postulated that addition of ascorbic acid alone or with polyethylene glycol could increase the beefy aroma characteristic of the sample mixture at the usual serving temperature (60 °C), although it might result in concomitant distorted beefy profile due to the unbalanced increase among beefy related aroma compounds.  5.4.2. Changes in sulfhydryl and disulfide content by the ingredients The increased areas o f G C peaks related to the beefy note seem to be closely related to the changes in protein structure of SPI. The analysis of S H and SS content demonstrated a significant increase in S H content and decrease in SS content by the addition of ascorbic acid alone or with polyethylene glycol (Table 5.3). The increased content of total S H in ascorbic acid added samples might arise from cleavage o f disulfide bonds in the samples with added ascorbic acid, a potent reducing agent. Therefore, it can be postulated that addition of ascorbic acid to SPI may lead to reduction of the disulfide linkage leading to possible changes in the conformation of SPI. Differences in total SH+SS among samples were observed in this study. The decreases of total SH+SS in SPI containing ascorbic acid or SPI containing ascorbic acid with polyethylene glycol may be partly attributed to possible involvement of the thiol groups in other bonds upon reduction of the disulfide bonds by ascorbic acid.  149  A wide range o f SH+SS contents of soy proteins have been reported. Boatright and Hettiarachchy (1995) reported that total S H content of SPI was 8.3 pmole/g protein and it was increased to 10.1 pmole/g protein after adding the antioxidant Tenox 22, while the total S H + SS content of 52.9 pmole/g protein was decreased to 43.9 pmole/g protein by antioxidant addition. Total S H contents o f 0.29 and 5.4 pmole/g protein were observed in SPI in buffer at p H 8 and p H 3, respectively, and these values were decreased with high pressure treatment between 400 and 600 M P a (Puppo et al., 2004). The average values o f surface, internal, and total S H contents of glycinin from five soybean cultivars were reported to be 0.6, 1.3, and 1.9 mole SH/mole glycinin, respectively (Nakamura et al., 1984), or 1.9, 4.1, 5.9 pmole SH/g glycinin, respectively. Determination o f 15 freeze-dried preparations of glycinin showed 0.6 - 2.2 mole S H / mole glycinin with an average of 1.4 mole SH/mole glycinin, which is equivalent to 1.9 - 6.9 pmole/g glycinin (Wolf, 1993) and the S H content increased by treatment with reducing agents such as 2mercaptoethanol, dithiothreitol, or sodium borohydride.  In the present study, the S H content in SPI was 0.6 pmole/g protein, which is less than values reported in the literature for glycinin. The discrepancy may be partly due to the presence of 0conglycinin with glycinin in the ratio of about 1:1 in SPI (Martins and Netto, 2006). Both a' and a subunits were reported to include low levels of cysteine and methionine while P subunit did not contain any methionine. In comparison each acid polypeptide in glycinin is linked with the basic polypeptide through a disulfide bond (Liu, 1997). Moreover, partial denaturation of the SPI during processing, as shown in Figure 5.1 and discussed in Section 5.3.1, may affect the total S H content of the SPI used in this study. In addition, the turbidity observed in the protein blanks could imply potential inaccessibility of interior S H that could not be accessed by Ellman's reagent due to the reported cross-link formation of unnatural covalent bonds during processing such as lysinoalanine in SPI (Wu et al., 1999), or due to SS bonds that would limit unfolding and exposure of interior S H in the SPI, even in the presence of SDS.  5.4.3. Changes in surface hydrophobicity by the ingredients Surface hydrophobicity was also affected by adding ingredients into SPI as shown in Table 5.4. Although results of So have been published using anionic fluorescence probes such as A N S and  150  C P A , to the author's knowledge, this thesis is the first to report So of SPI using the neutral uncharged probe, P R O D A N , which measures surface hydrophobicity based on the binding between probe and protein primarily from the hydrophobic interaction with minimal interference from electrostatic interaction.  While no significant difference was detected by addition o f sucrose or glucosamine to SPI, significant increases in surface hydrophobicity were observed when polyethylene glycol, ascorbic acid, or the combination o f the two ingredients were added to SPI. Sucrose is a disaccharide with eight hydroxyl groups and glucosamine is an amino sugar including four hydroxyl groups with an amino group. Polyols, which are ingredients with poly hydroxyl groups, have been reported to contribute to the stabilization o f the structure of protein molecules (Gekko et al., 1999). Various concentrations o f glycerol and sorbitol enhanced thermal stability o f soy protein, p-conglycinin and glycinin in aqueous solution. Addition of polyols was assumed to strengthen intra-molecular hydrophobic interaction due to the preferential solvent interaction with the protein molecules. According to the mechanism described by Gekko and Timasheff (1981), polyols such as glucosamine and sucrose in this study are fundamentally hydrophilic compounds, which can occupy a part of the solvation sheath around SPI with concomitant stabilization of solvent structure. Polyols were reported to show a stabilization effect on hydrophobic interaction of protein molecules more effectively compared to peptide-peptide hydrogen bonds (Gekko, 1981).  In contrast, surface hydrophobicity of SPI with polyethylene glycol was significantly increased. M i l d denaturing treatment by polyethylene glycol may lead to increasing surface hydrophobicity on SPI. Polyethylene glycol is highly soluble in water but includes a repeating hydrophobic region (-CH2-CH2-) per monomer unit, which could evoke relocation o f the hydrophobic side chains of non-polar amino acid residues in SPI. It can be hypothesized that addition of ascorbic acid increased the surface hydrophobicity through structural changes in SPI accompanying the reduction o f disulfide bonds to sulfhydryl groups (Table 5.3), and the increasing proportion of unordered structure as analyzed by FT-Raman spectroscopy (Table 5.7). The breakage of disulfide bonds in SPI could enable buried amino acid residues with non-polar side chains to become exposed to the surface o f the protein molecule, thus increasing the binding affinity or number o f binding sites for the hydrophobic fluorescence probes such as P R O D A N . The  151  denaturing effect by polyethylene glycol seemed to be enhanced by reduction o f disulfide bond by ascorbic acid, resulting in considerable increase of surface hydrophobicity in SPI treated with both ascorbic acid and polyethylene glycol.  5.4.4. Changes in FT-Raman spectra by the ingredients Addition of ingredients also affected the FT-Raman spectra of SPI. The most significant changes were observed in SPI containing ascorbic acid with polyethylene glycol, as shown in Table 5.6, implying the presence of additive effect between those two ingredients. Compared to SPI, the F T Raman spectrum o f SPI containing ascorbic acid with polyethylene glycol showed decreased intensity at 760 cm" indicating decrease of "buriedness" or hydrophobic microenvironment of the 1  tryptophan indole ring (Li-Chan, 1996). The decreased intensity suggested that when the ascorbic acid and then polyethylene glycol was added to SPI, the tryptophan residues became exposed from a buried, hydrophobic interior to a more polar microenvironment. A decrease in the intensity ratio of the 850 and 830 era" doublet could indicate intermolecular interactions of the 1  tyrosine residues as a strong hydrogen donor or increased buriedness around the residues (Howell and Li-Chan, 1996). The observed decrease in the ratio of the doublet in the samples containing ascorbic acid alone or with polyethylene glycol might suggest the possibility of interaction between tyrosine residues and polyethylene glycol, which could serve as a hydrogen bond acceptor. Increase o f the intensity at 1450 cm" in the samples including polyethylene glycol 1  alone or with ascorbic acid indicate decreased interior hydrophobic interactions due to increasing polarity o f the environment around hydrocarbon chains, which implies unfolding of proteins in those samples. Changes in intensity of the SS stretching band (510-550 cm" ) could be explained 1  by possible formation or stabilization o f disulfide bonds in SPI by addition o f sucrose, in contrast to disulfide bond breakage in samples containing ascorbic acid or ascorbic acid with polyethylene glycol, probably due to reduction by ascorbic acid, which is in good agreement with the results of S H and SS determination as discussed in the previous section (5.3.3 o f this thesis). In addition, the Raman wavenumber shift of the SS-stretching band suggests conformational changes in disulfide bonds, from a gauche-gauche-trans  conformation in SPI to a  trans-gauche-trans  conformation when the ingredient was added to SPI. However, no peaks were found in the region of 2550-2558 cm" , which could give information on S-H stretching, probably due to the low 1  concentration of sulfhydryl groups in these samples as shown by analysis with Ellman's reagent. 152  Analysis of the Amide I band o f the FT-Raman spectrum indicated that the SPI in this study contained 14.4 % a-helix, 55.2 % P-sheet, 19.0 % p-turns, and 11.5 % unordered structures, and was not changed by freeze-drying. Rickert et al. (2004) reported that a SPI produced by pilotplant-scale process contained 47.6 % o f glycinin and 47.2 % o f p-conglycinin. Based on its amino acid sequence, the composition of secondary structure of glycinin was predicted to contain 25 % a-helix, 25 % p-sheet, 42 % turns, and 8 % unordered structures (Argos et al., 1985). Abbott et al. (1996) investigated the secondary structure o f unheated glycinin using infrared spectroscopy having 25 % a-helix, 33 % P-sheet, 31 % turns, and 12 % unordered structures in aqueous buffer and found that glycinin in solution had the same secondary structure as glycinin in hydrated solids. The observed secondary structure composition o f SPI shown in Table 5.7, i.e. higher P-sheet and lower a-helix than the reported values for SPI in the literatures, is consistent with the results from D S C indicating denaturation of the SPI used in this study. Increase in the Psheet at the expense o f existing a-helix has been reported to be associated with denaturation and/or aggregation o f proteins (Herald and Smith, 1992; Przybycien and Bailey, 1991; Puppo et al., 2004).  The secondary structure composition o f SPI was changed by the addition of ingredients, as shown in Table 5.7. The decrease of unordered structure by addition o f glucosamine and sucrose might be explained by the reported roles o f sugars, which have been used as a protein stabilizer in food industry, with the hydrating ability through the poly-hydroxyl groups. Increased unordered structure in the samples containing ascorbic acid alone or with polyethylene glycol might be related to their increased surface hydrophobicities (section 5.3.4) along with the postulated disulfide breakage discussed in section 5.3.3.  Generally, various functional properties o f globular proteins depend on their physico-chemical properties,  which  are  primarily  governed  by  structural  and  conformational  attributes  (Hettiarachchy and Kalapathy, 1997). Denaturation has been known to be important in gelation property of SPI, and has been regarded as a necessary step for gel formation (Renkema et al., 2002). A s a result of thermal denaturation during gelation, hydrophobic areas buried in the native conformation are exposed to surrounding solvent. This change in partially unfolded protein 153  creates an aggregation process due to the imbalance between attractive and repulsive forces of molecules (Berli et al., 1999). Similarly, denaturation or changes in molecular structure induced by added ingredients must have played an important role in changing the flavour holding behavior of SPI. In the present study, the flavour holding property of SPI was most affected by structural changes of SPI through disulfide bond breakage in addition to hydrophobic interaction. Breakage in disulfide bond might be necessary to facilitate conformational changes in SPI.  5.4.5.  Descriptive sensory evaluation  According to the results from GC analysis, the areas of beef attribute related peaks increased by addition of ascorbic acid alone or with polyethylene glycol. Furthermore, some of these peaks were associated with high detection frequency in GC-0 analysis, which might imply improvement of aroma perception of SBF in mixtures of SPI and SBF. Therefore, descriptive sensory analysis was conducted with the 3 samples, which were SPIF, AF and APF as described in section 5.2.6. The sample of SBF with SPI containing ascorbic acid with polyethylene glycol showed considerable changes in the sensory scores of all the attributes although the changes were not statistically significant (p > 0.05) in beefy and yeasty notes. The sample of SBF with SPI containing ascorbic acid showed a similar trend with the sample containing those two ingredients but it was not statistically significant.  Since 5 out of the 7 panelists in this study also participated in the previous study (Chapter 4 of this thesis) and the same sensory evaluation method including evaluation sheet, reference sample, and procedure for sample preparation, serving and evaluation was used, this result was compared to the sensory result from Chapter 4 of this thesis. The sample coded "SPIF" in this study has the same composition as sample "S3" in Chapter 4 shown in Figure 4.1, Figure 4.2 and Table 4.10, consisting of 250 mg SPI in 5 g dd-water with 150 mg SBF (medium SPI with low SBF). By comparison of S3 in Figure 4.1 with SPIF in Figure 5.4, all the attributes except yeasty note were similarly assessed by the panelists. Different perception in yeast note was probably due to poor consensus of the definition for yeasty note.  The result of PCA of all the samples including the 12 samples evaluated in Chapter 4 and the 3 samples assessed in this chapter is shown in Figure 5.5. Roasted and beefy attributes were 154  demonstrated to be strongest in the sample APF. It is interesting to note that the characteristic of APF was evaluated to be similar to sample SI, which contains the same amount of SBF but without any SPI or additional ingredient.  Table 5.9 compares the mean intensity values of the 5 attributes for the 3 samples (SPIF, AF, and APF) with the results from Table 4.10. The intensity of the sample APF for the each of the attributes for roasted, soymilk-like, cereal, and beefy, was similar to the intensity of sample SI, which contains SBF but no SPI. The results clearly demonstrated that sample APF fell into the same group as SI with no significant difference in roasted, soymilk-like, cereal, and beefy attributes, indicating that the roasted and beefy characteristics were recovered while soymilk-like and cereal notes were suppressed to a certain extent by adding ascorbic acid to SPI before mixing with SBF, and the effect was enhanced by treating with ascorbic acid and polyethylene glycol in SPI.  Overall, it could be postulated that adding ascorbic acid and/or polyethylene glycol in SPI prior to interacting with SBF resulted in an increased release of flavour compounds involving the beefy characteristic. The lack of a significant difference in beefy note may probably be attributed to the complexity of "beef aroma, which can not be constructed easily, and possible deformation of beefy profile due to the unbalanced increase among beefy related aroma compounds. Nevertheless, although significant difference was not found in beefy note among the samples under these conditions, beefy characteristic in sample APF was expressed by enhancement of roasted note and alleviation of soymilk-like and cereal notes in the mixture of SBF and SPI. The results therefore could be considered successful in terms of recovering some of the beef characteristic and suppressing the soy characteristics, but also indicate the need for further research.  5.5. Conclusion GC analysis of SBF in the presence of SPI alone and SPI with ingredients demonstrated that the mixture of SBF and SPI containing ascorbic acid alone or ascorbic acid with polyethylene glycol contributed to increasing peak area of individual beef attribute related volatile compounds in GC 155  chromatogram.  According  to  the  results  of  sulfhydryl  and  disulfide groups,  surface  hydrophobicity, and FT-Raman spectroscopy, it can be postulated that while glucosamine and sucrose had little effect, the addition o f ascorbic acid especially in conjunction with polyethylene glycol triggered conformational changes in SPI structure mainly through disulfide bond reduction, resulting in increased surface hydrophobicity and unordered structure on SPI, which in turn affected improvement o f S B F perception in the mixture of S B F and SPI. The results of hydrophobicity measurement and FT-Raman analysis indicated that the denaturing effect was likely to be enhanced by the combined treatment of ascorbic acid and polyethylene glycol. Polyethylene glycol induced interior hydrophobic side chains in SPI to be exposed, resulting in unfolding o f the SPI, in other words, relocation of the non-polar amino acid residues of SPI induced by changes  in hydrophobic interaction with the hydrophobic repeating unit of  polyethylene glycol, which can be enhanced by means of modification of disulfide bond by ascorbic acid. The reduction of disulfide bonds, increased surface hydrophobicity and increased unordered structure in SPI, along with increased G C peak areas o f indicator peaks in the SPI containing ascorbic acid alone or with polyethylene glycol, were found to be associated with an increase in the perceived beef characteristic attributes in descriptive analysis. These results provide the basis for further research to elucidate strategies maximizing perception of beefy aroma in soy products.  156  4 0 0 Q-  Figure 5.1. Differential scanning calorimetric (DSC) thermograms of SPI (0.2 %, 5 %, and 10 % in p H 7.5 phosphate buffer).  157  Figure 5.2. FT-Raman spectrum (400-1700 cm" ) of SPI powder obtained by freeze-drying SPI 1  solution (5 % w/w in 0.05 M T r i s - H C l buffer p H 7.4).  158  Figure 5.3.  FT-Raman spectra o f freeze-dried SPI and SPI treated with various ingredients (baseline corrected, ingredient spectrum subtracted, and normalized to the intensity o f phenylalanine peak at 1003 cm" ). G : glucosamine; S : sucrose; P : polyethylene glycol; A : ascorbic acid; A P : ascorbic acid with polyethylene glycol. 1  159  Figure 5.4. Cobweb diagram of the sensory scores from the descriptive analysis of simulated beef flavour in the presence of soy protein isolate (SPIF), SPI containing ascorbic acid (AF) and SPI containing ascorbic acid with polyethylene glycol (APF) (n=14; 7 panelists with 2 replications).  160  PC2  51 and S I H ( L - S B F ; N - S P I ) 52 and S 2 H ( H - S B F ; N - S P I ) 53 and S 3 H ( L - S B F ; M - S P I ) 54 and S 4 H ( H - S B F ; M - S P I ) 55 and S 5 H ( L - S B F ; H-SPI) 56 and S 6 H ( H - S B F ; H-SPI) SPIF ( L - S B F ; M-SPI) A F ( L - S B F ; M - S P I with A ) A P F ( L - S B F ; M - S P I with A P )  Roasted S2H»  S  4  H  »  S3H •  S5HTO  •  S2  S5  Beefy Cereal  Soymilk-like  Yeasty  Figure 5.5. PC loadings and scores of the sensory attributes and the sample mixtures by principal component analysis; PCI versus PC2. L-SBF and H-SBF represent low (150 mg) and high (500 mg) dose of simulated beef flavour while N-SPI, M-SPI, and H-SPI symbolize no (0 mg), medium (250 mg), and high (500 mg) amount of soy protein isolate in the sample. Refer to Table 4.3 and Figure 5.4 for the sample codes.  161  Table 5.1. Effect o f added ingredient in SBF-SPI mixture incubated at R T or 60 °C on the peak areas of volatile compounds captured by H S S P M E under the adsorption condition at R T ' . ___ 1  cgp jp #3 Retention time  Area  2  RT incubation - RT adsorption % Area proportion based on SBF %  SPIF  GF  SF  PF  AF  60 °C incubation - RT adsorption % Area proportion based on SBF  APF SPIF GF  SF  PF  AF  p k characteristics e a  APF Tentative peak identification  DF  5  Beefy  6  (%)  50  0.710*  1.734  11512 100  71  94  68  60  88  48  42  42  40  32  55  56  3-Methyl furan  13.271  2514 100  19  15  0  0  0  0  17  0  0  0  0  0  Limonene  IP13 28.288  13837 100  72 933 100 129 387 15016 100 48 73 1897 100 0 57  50  50  63  64  50  45  42  34  57  57  delta-Elemene  159  104  58  133  0  0  0  0  0  0  Unknown  0.713**  57  58  49  45  42  35  59  59  Unknown  0.928***  56  42  74 77 184  22  39  19  21  67  Unknown  0.929***  66  64  55  43  71 110  78  Unknown  0.638*  IPl  29.122  42  0.713** 50  31.271  4513 100  57  89  65  74  71 75 288  31.530  2735 100  24  83  29  50  70  66  49  53  0  26  87  78  Unknown  31.881  113227 100  49  78  58  61  63  76  54  52  46  41  67  68  Unknown  0.750** 0.924***  32.231  3743 100  53  79  59  61  70  77  61  52  53  38  73  75  Unknown  0.879***  33.166  3793 100  50  54  61  68  72  58  46  50  37  67  70  Unknown  0.826**  33.306  6149 100  50  75 79  58  63  73  83  56  55  49  45  71  72  Unknown  0.894***  34.691  828 100  0  52  0  0  50  106  0  0  0  0  0  0  Unknown  0.881***  35.365  1501 100  57  80  0  0  26  57  0  0  0  0  0  29  Unknown  0.823**  IP15 35.759  417 100  0  105  0  0  98  220  0  0  0  0  85  90  Calamenene  35.887  1624 100  0  40  32  215  58  393  81  181  77  147  82  296  Unknown  29.971 30.729  75  0.883*** 0.685*  0.822** Unknown 44 43 66 67 82 48 68 51 55 60 81 5318 100 46 36.322 SPIF : S B F in the presence o f SPI; G F , SF, P F , A F , and A P F : S B F with SPI containing glucosamine, sucrose, polyethylene glycol, ascorbic acid, and ascorbic acid with polyethylene glycol, respectively. Bold numbers indicate significant (p < 0.05) difference from SPI in peak area of G C chromatogram by Fisher's L S D test. Peak area and % area are the average values from 3 replicate G C analyses. Number o f indicator peak as selected from the previous study (Chapter 4 o f this thesis). Retention time in minutes on G C chromatogram of this study. Detection frequency (%) - percentages o f panel from a total o f 8 panels perceived the aroma compound in GC-olfactometry. Pearson correlation coefficient for beefy notes in descriptive sensory analysis performed in Chapter 4 of this thesis. 1  2  3  4  5  6  Table 5.2. Effect o f added ingredient in SBF-SPI mixture incubated at R T or 60 °C on the peak areas of volatile compounds captured by H S S P M E under the adsorption condition at 60 ° C ' ' . (Refer to Table 5.1 for sample codes) 2  CRF jp #3 Retention time IP1 IP2 IP3  IP4 IP5 IP6 IP8 IP9 IP10 IP12  IP13  IP14  OS  1.734 7.893 12.132 13.271 13.915 15.014 15.877 16.247 16.953 19.917 20.608 22.151 24.232 25.794 26.319 26.954 27.841 28.288 28.830 29.122 29.971 30.220 30.377 30.615 30.729 30.950 31.271 31.317 31.530  Area 6953 5537 3360 1929 1352 6302 1761 2543 1935 2881 3328 3951 468 8033 3792 2156 3710 19545 8259 7556 25200 1260 3031 1639 9478 721 1849 11730 4843  RT incubation-60 °C adsorption % Area proportion based on SBF %  SPIF GF  100 106 100 122 100 66 100 24 100 33 100 .84 100 54 100 49 100 20 100 0 100 47 100 57 100 0 100 41 100 19 0 100 100 16 100 48 100 44 100 131 100 46 100 0 100 16 100 0 100 34 0 100 100 88 100 54 100 44  SF  91 113 83 141 67 58 0 30 34 39 62 63 88 51 23 68 0 49 23 35 44 61 52 59 0 0 47 46 18 40 19 21 35 38 36 55 41 53 130 130 52 39 0 0 14 49 0 0 42 29 0 0 106 0 71 60 50 39  PF  60°C incubation-60°C adsorption % Area proportion based on SBF  A F A P F SPIF GF  94 163 158 74 89 83 54 62 49 0 0 42 31 240 174 89 73 27 87 47 80 82 49 75 38 0 39 62 0 28 59 99 81 77 64 58 0 0 0 40 50 51 34 41 38 19 63 99 22 95 56 62 42 55 28 35 61 128 158 109 54 62 53 0 0 0 60 62 50 0 0 0 42 39 46 0 0 0 102 0 0 66 158 131 98 76 56  80 77 20 18 0 67 64 45 20 46 49 54 0 38 0 24 40 50 43 116 51 0 40 0 40 0 34 69 50  83 62 31 0 82 73 50 35 18 0 46 55 0 42 0 24 22 42 41 149 45 0 17 0 37 0 0 77 74  AF  APF  SF  PF  85 75 32 0 25 68 53 36 0 53 45 50 0 45 17 0 35 47 47 117 49 0 15 0 40 0 46 67 72  78 131 146 65 59 63 17 43 45 20 0 0 0 133 148 60 67 67 52 48 51 32 43 49 18 18 0 37 0 0 71 48 75 47 64 75 0 0 0 43 50 71 19 0 16 43 0 0 0 21 0 42 46 46 50 25 26 151 115 123 44 50 53 0 0 0 19 36 36 0 0 0 35 37 43 0 0 0 0 41 0 57 94 104 82 74 73  Peak characteristics  Tentative peak identification  DF  Beefy  6  (%)  50 0.710* 3-Methyl furan 75 0.658* 2-Acetyl furan 88 0.872*** delta-3-Carene 0.713** Limonene 0.648* Unknown 0.773** Unknown 2-Ethyl-3,6-dimethylpyrazine 88 0.601* 100 0.852*** Unknown 100 0.827** Unknown 2,3-Diethyl-5-methylpyrazine 50 0.590* 0.678* 4-Terpeneol 75 0.715** Decanal 2-Isoamyl-6-methylpyrazine 50 0.599* 75 0.794** Unknown 0.684* Unknown 0.752** Unknown 0.887*** Unknown 50 delta-Elemene 0.678* Unknown 0.713** Unknown 0.928*** Unknown 0.617* Unknown beta-Cubebene 63 0.850*** 0.645* Unknown 0.929*** Unknown 0.868*** Unknown 0.638* Unknown 0.914*** Unknown 0.750** Unknown  SBF jp #3 Retention time  Area  incubation-60 °C adsorption % Area proportion based on SBF %  SPIF GF  60°C incubation-60°C adsorption % Area proportion based on SBF  SF  PF  A F A P F SPIF GF  SF  PF  A F APF  57 43 48 44 57 12 0 31 31 0 44 11 43 34 41 40 40 40 26 38 0 47 43 43  57 44 44 45 57 25 0 49 35 0 46 10 45 37 38 404 40 40 54 40 0 47 48 52  55 53 30 56 55 25 0 26 31 0 45 11 40 37 47 55 46 30 64 38 0 31 42 39  54 40 40 45 53 11 0 33 35 62 29 18 38 33 39 40 37 41 69 39 0 51 46 44  49 33 36 38 49 0 0 37 32 47 26 19 36 31 32 140 34 37 23 37 0 53 42 44  48 54 40 56 48 24 0 15 22 0 39 8 38 36 41 46 41 29 41 27 0 35 37 34  characteristics  Tentative peak identification  DF  Beefy  6  (%)  IP15  31.881 32.231 32.679 33.166 33.306 33.620 33.813 34.211 34.346 34.481 34.691 34.871 35.097 35.365 35.759 35.887 36.322 36.646 37.033 37.347 37.674 38.001 38.425 38.658  262026 7117 4244 6537 24572 4026 4478 7156 11082 2104 9569 4778 9752 15997 16917 20361 49502 11216 3176 12581 2119 12714 6531 13153  100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100  50 41 42 43 50 9 0 31 30 0 40 33 27 34 39 39 36 37 0 32 0 45 44 40  43 25 29 38 45 13 0 17 26 0 34 9 35 31 35 42 33 26 61 33 0 39 36 36  50 47 30 53 51 14 0 30 28 0 43 9 38 36 39 407 43 28 54 30 0 26 41 38  55 42 42 51 54 29 0 32 36 0 43 20 38 36 38 43 38 41 59 38 0 38 40 39  47 38 39 45 48 9 0 26 32 0 38 18 36 33 37 48 35 38 27 33 0 31 43 36  53 48 48 57 52 34 0 24 26 0 41 0 42 40 41 379 43 32 43 31 0 37 33 30  Unknown Unknown Unknown Unknown Unknown Unknown Unknown Unknown Unknown Unknown Unknown Unknown Unknown Unknown Calamenene Unknown Unknown Unknown Unknown Unknown Unknown Unknown Unknown Unknown  0.924*** 0.879*** 0.863*** 0.826** 0.894*** 0.779** 0.689* 0.775** 0.787** 0.600* 0.881*** 0.781** 0.872*** 0.823** 75 0.883*** 0.685* 0.822** 0.780** 0.800** 0.778** 0.815** 0.746** 0.634* 0.767**  B o l d numbers indicate significant (p < 0.05) difference from SPI in peak area of G C chromatogram by Fisher's L S D test. Peak area and % area are the average values from 3 replicate G C analyses. Number o f indicator peak as selected from the previous study (Chapter 4 of this thesis). IP7 and I P l 1 were not detected in this study due to the peak area being small than the applied area limit o f 1000. Retention time in minutes on G C chromatogram of this study. Detection frequency (%) = percentage of 8 panelists who perceived the aroma compound in GC-olfactometry. Pearson correlation coefficient for beefy notes in descriptive sensory analysis performed in Chapter 4 of this thesis. 1  2  3  4  5  6  Table 5.3. Sulfhydryl groups and disulfide bonds in SPI and SPI treated with various ingredients . 1  Sample  2  Total sulfhydryl group  Disulfide bond  (pmole SH/g protein)  (pmole SS/g protein)  SPI  0.6  b  36.7  b  G  0.6  b  36.8  b  S  0.8  c  38.6  b  P  0.5  a  36.8  b  A  2.0  d  31.0  a  AP  2.4  e  30.8  a  'Mean values (n=2) with different superscripts (a-e) within a column are significantly (p < 0.05) different. 2  G : glucosamine; S : sucrose; P : polyethylene glycol; A : ascorbic acid; AP : ascorbic acid with  polyethylene glycol.  165  Table 5.4.  Surface hydrophobicity (S ) 0  o f S P I a n d S P I treated w i t h various  ingredients.  Surface hydrophobicity Sample  2  1  (%-')  145  ±  4  a  G  146 ±  2  a  S  150  ± 2  a  P  170  ± 9  b  A  171  ± 6  b  SPI  198 ±  A P  G  : glucosamine;  polyethylene  2  S : sucrose; P : polyethylene  glycol; A  24  c  : ascorbic acid; A P : ascorbic acid with  glycol.  S a m p l e s w i t h different superscripts (a-c) are s i g n i f i c a n t l y (p < 0.05)  h y d r o p h o b i c i t y (So). V a l u e s s h o w n  are m e a n ± standard deviation  different in  surface  (n=3).  166  Table 5.5. Tentative assignment of major bands in the FT-Raman spectrum of SPI and SPI treated with various ingredients (Adapted from Li-Chan, 1996).  Wavenumber region 1 (± 2 cm" )  Tentative assignment  1  510, 522, 539  S-S stretching of cystine  760  Tryptophan indole ring  850/830  Tyrosine ring  1003  Phenylalanine ring  1450  C-H bending of aliphatic residues  1655 ± 5  Amide I C=0 stretch, N - H wag (a-helix)  1670 ± 3  Amide I C=0 stretch, N - H wag (anti-parallel P-sheet)  1665 ± 3  Amide I C=0 stretch, N - H wag (solvated disordered structure)  1685  Amide I C=0 stretch, N - H wag (non-hydrogen bonded disordered structure)  2550-2580  S-H stretching of cysteine  2800-3000  C-H stretching of aliphatic residues  167  Table 5.6. Normalized intensity values at selected regions o f the FT-Raman spectra of SPI and SPI treated with various ingredients.  Mean of normalized peak intensity ' [wavenumber (cm" )]  1 2  „ , . Band assignment  1  SPI  SS of cystine (Gauche-gauche-trans)  G  S  P  A  AP  0.48  a  0.48  [537]  [539]  0.76 [759]  0.58  0.60  0.85  0.55" [522] 0.60 [537]  0.64  [540]  0.52" [541]  b  (Trans-gauche-trans)  c  Tryptophan indole ring  0.74" [760]  0.80 [759]  0.76 [759]  0.74" [759]  Tyrosine doublet ratio  0.97" [850/830]  0.99 [850/830]  0.89 [850/830]  0.98 [850/830]  Aliphatic residues C - H bending  1.62 [1448]  1.71 [1450]  1.76 [1450]  1.80  Aliphatic residues C - H stretching  3.79 [2934]  3.79 [2935]  3.99 [2932]  a  a  b  b  a  a  b  b  a  a  b  b  a  a  a  [759] a  [850/830]  [850/830]  1.81  [1449]  1.64 [1449]  [1449]  4.10 [2933]  3.75 [2934]  4.06 [2932]  b  a  a  a  b  a  'Mean values are shown from duplicate analyses. Samples with different superscripts (a-c) within a row are significantly (p < 0.05) different. The numbers in bold letter indicate intensity difference (p < 0.05) in Raman intensity of ingredient added SPI from SPI. G : glucosamine; S : sucrose; P : polyethylene glycol; A : ascorbic acid; A P : ascorbic acid with polyethylene glycol.  2  oo  Table 5.7. Composition of secondary structure in SPI and SPI treated with various ingredients ' . 1 2  Structure  NOFD  SPI  G  s  P  A  AP  (%)  (%)  (%)  (%)  (%)  (%)  (%)  i3.r  15.0  ab  16.5  d  57.2  C  55.1  d  19.8  d  17.0  8.0  a  11.4  100  100  Total a-helix  14.4  a  13.6  Total P-sheet  55.2  ab  57.2  C  59.6  P-reverse turn  19.0  cd  18.6  cd  20.0  Unordered structure  11.5  Structure total  100  b  10.6  a  b  100  7.3  100  a  bc  18.9  a  56.8  C  1.4  b  d  17.5  bc  56.9  a  22.9  100  d  cd  C  7.6  b  18.0  C  100  'Mean values are shown from duplicate analyses. Samples with different superscripts (a-d) within row are significantly (p < 0.05) different. 2  NOFD : original SPI (not freeze-dried); SPI: freeze-dried SPI without any ingredient;  G : glucosamine; S : sucrose; P : polyethylene glycol; A : ascorbic acid; AP : ascorbic acid with polyethylene glycol.  169  Table 5.8. Results of Fisher's least significant difference test on mean sensory scores of each sample for 5 attributes.  Mean score  1  SPIF  AF  Roasted  6.81  794  Soymilk-like  6.88  b  5.34  ab  3.67  Cereal  6.41  b  4.85  ab  4.08  Beefy  6.71  a  8.03  8.79  Yeasty  7.09  a  6.93  6.02  2  a  APF  3  ab  a  a  9.49  4  b  a  a  a  a  'Mean scores for each attribute within a row with different letters are significantly different (p < 0.05) using Fisher's least significant difference test (n=14; 7 panelists with 2 replications). 2  Mixture containing SBF and SPI but without any ingredient  3  Mixture containing SBF and SPI treated with ascorbic acid  4  Mixture containing SBF and SPI treated with ascorbic acid and polyethylene glycol  170  Table 5.9. Location o f the mean intensity values o f the 5 attributes for the 3 samples in comparison with the samples in Table 4.10 . 1  Roasted  Soymilk-like  Cereal  Sample  Mean score  Sample  S5H  1.93 a  S2H  S5  2.66 a  S2  2.53 ab  S2  Mean score 1.85  a  Sample S2H  Beefy Mean score  1.73  a  Sample  Yeasty Mean score  Sample  Mean score  S5  2.54 a  S5H  2.80 a  2.32 ab  S5H  2.69 a  S5  3.11 a  S3H  3.46 ab  S4H  2.88 ab  S4H  2.33 ab  S3H  3.73 a  S3H  3.20 a  S3  5.69 bc  S4  3.25 ab  S4  3.57 ab  SPIF  6.71 -  S3  4.17 ab  SPIF  6.81 -  SI  3.52 ab  SI  3.65  S3  6.71 b  APF  S6H  6.83 cd  S6H  7.19 b  S6  S6  7.39 cde  S1H  4.02 ab  S6  7.72 bc  AF  7.94 -  S6H  4.73 bc  S1H  8.53 def  APF  9.49 -  S6  6.25 cd  S6  SI  9.63 ef  S3  6.73 cd  SPIF  S4  10.23 fg  SPIF  S4H  10.63 fg  S3H  7.87 de  S3H  7.91  S2  10.82 fg  S5H  8.98 e  S5H  9.25 ef  S2H  12.14 g  S5  9.98 e  S5  APF  AF  3.67 -  5.35 -  6.88 -  S1H  APF S6H  AF  S3  ab  3.73 ab  4.08  -  4.69 bc  4.58  -  6.19 cd  6.41  -  6.84 cde  9.99  def  f  AF SI  APF  8.03 8.47 bed  AF S6H  SPIF  6.02 6.32 bc  6.93 6.97 cd  7.09 -  8.79 -  SI  7.50 cd  S4H  9.92 cd  S4  7.97 cd  S4  9.93 cd  S1H  8.43 cd  S1H  10.13 cd  S4H  8.43 cd  S2H  10.78 d  S2  9.29 d  S2  10.88 d  S2H  9.41 d  'Mean scores for each attribute within a column with different letters are significantly different (p < 0.05) using Duncan's multiple comparison test (n=12; 6 panelists with 2 replications), which was performed for the 12 samples in Chapter 4 o f this thesis. Scores are listed in ascending order. Refer to Table 4.3 and Table 5.8 for the sample code.  5.6. References  Abbott, T. P., Nabetani, H., Sessa, D. J., Wolf, W. J., Liebman, M . N . , and Dukor, R. K. (1996). Effects of bound water on FTIR spectra of glycinin. Journal of Agricultural and Food Chemistry, 44, 2220-2224. Alizadeh, N . and Li-Chan, E. C. Y . (2000). Comparison of protein surface hydrophobicity measured at various pH values using three different fluorescent probes. Journal of Agricultural and Food Chemistry, 48, 328-334.  Argos, P., Narayana, S. V . L., and Nielsen, N . C. (1985). Structural similarity between legumin and vicilin storage proteins from legumes. EMBO Journal, 4, 1111-1117. Badii, F., and Howell, N . K . (2003). Elucidation of the effect of formaldehyde and lipids on frozen stored cod collagen by FT-Raman spectroscopy and differential scanning calorimetry. Journal of Agricultural  and Food Chemistry, 51, 1440-1446.  Berli, C. L. A., Deiber, J. A., and Anon, A . C. (1999). Heat-induced phenomena in soy protein suspensions. Rheometric data and theoretical interpretation, Journal of Agricultural and Food Chemistry, 47, 893-900.  Beyeler, M . , and Solms, J. (1974). Interaction of flavor model compounds with soy protein and bovine serum albumin. Lebensmittel Wissenschaft und Technologie, 7, 217-219. Beveridge, T., Toma, S. J., and Nakai, S. (1974). Determination of SH- and SS-groups in some food proteins using Ellman's reagent. Journal of Food Science, 39, 49-51. Boatright, W. L., and Hettiarachchy, N . S. (1995). Soy protein isolate solubility and surface hydrophobicity as affected by antioxidants. Journal of Food Science, 60, 798-800. Cheftel, J. C , Cuq, J.-L., and Lorient, D. (1985). Amino acids, peptides, and proteins. In O. R. Fennema (Ed.), Food Chemistry, 2 edition, (p. 279). New York: Marcel Dekker, Inc. nd  Chung, M . W. Y., Lei, B., and Li-Chan, E. C. Y. (2005). Isolation and structural characterization of the major fraction from NorMan flaxseed (Linum usitatissimum L.). Food Chemistry, 90, 271-279. Damodaran, S. (1985). Estimation of disulfide bonds using 2-nitro-5-thiosulfobenzoic acid: Limitations. Analytical Biochemistry, 145, 200-204. Damodaran, S., and Kinsella, J. E. (1981a). Interaction of carbonyls with soy protein: Thermodynamic effects. Journal of Agricultural and Food Chemistry, 29, 1249-1253.  Damodaran, S., and Kinsella, J. E. (1981b). Interaction of carbonyls with soy protein: Conformational effects. Journal of Agricultural and Food Chemistry, 29, 1253-1257. 172  Dong, W., and Hoseney, R. C . (1995). Effects of certain breadmaking oxidants and reducing agents on dough rheological properties. Cereal Chemistry, 72, 58-64. Ellman, G . L . (1959). Tissue sulfhydryl groups. Archives of Biochemistry and Biophysics, 82, 7072. Gekko, K . (1981). Mechanism o f polyol-induced protein stabilization: Solubility of amino acids and diglycine in aqueous polyol solutions, Journal of Biochemistry, 90, 1633-1641. Gekko, K . and Timasheff, N . (1981). Mechanism of protein stabilization by glycerol: Preferential hydration in glycerol-water mixtures, Biochemistry, 20, 4667-4676. Gekko, K . , L i , W., and Makino, S. (1999). Competing effect of polyols on the thermal stability and gelation o f soy protein. Bioscience, Biotechnology and Biochemistry, 63, 2208-2211. Gremli, H . A . (1974). Interaction o f flavor compounds with soy protein. The Journal of the American Oil Chemists' Society, 51, 95A-97A. Herald, T. J. and Smith, D . M . (1992). Heat-induced changes in the secondary structure of hen egg S-ovalbumin. Journal of Agricultural and Food Chemistry, 40, 1737-1740. Hettiarachchy, N . and Kalapathy, U . (1997). Soybean protein products. In K . L i u (Ed.), Soybeans: Chemistry, Technology and Utilization, (pp. 379-411). N e w York: Chapman and Hall. Howell, N . and Li-Chan, E . (1996). Elucidation of interactions o f lysozyme with whey proteins by Raman spectroscopy. International Journal of Food Science and Technology, 31, 439451. Kashiba-Iwatsuki, M . , Kitoh, K . , Kasahara, E., Y u , H . , Nisikawa, M . , Matsuo, M . , and Inoue, M . (1997). Ascorbic acid and reducing agents regulate the fates and functions o f S-nitrosothiols. Journal of Biochemistry, 122, 1208-1214. L i , Z., Griin, I.U., and Fernando, L . N . (2000). Interaction of vanillin with soy and dairy proteins in aqueous model systems: A thermodynamic study. Journal of Food Science, 65, 997-1001. Li-Chan, E . C . Y . (1996). The applications of Raman spectroscopy in food science. Trends in Food Science and Technology, 7, 361-370. Li-Chan, E . C . Y . , and M a , C . - Y . (2002). Thermal analysis of flaxseed (Linum usitatissimum) proteins by differential scanning calorimetry. Food Chemistry, 11, 495-502. L i u , K . (1997). Chemistry and nutritional value of soybean components. In Soybeans: Chemistry, Technology and Utilization, (pp. 25-113). N e w York: Chapman & H a l l .  173  Malcolmson, L . J., and McDaniel, M . R. (1987). Flavor protein interactions in a formulated soup containing flavored soy protein. Canadian Institute of Food Science and Technology, 20(4), 229-235. Martins, V . B . and Netto, F. M . (2006) Physicochemical and functional properties of soy protein isolate as a function of water activity and storage. Food Research International, 39, 145-153. Moon, S.-Y., Cliff, M . A., and Li-Chan, E. C. Y . (2006). Odour-active components of simulated beef flavour analyzed by solid phase microextraction and gas chromatography-mass spectrometry and -olfactometry. Food Research International, 39, 294-308. Moon, S.-Y., and Li-Chan, E. C. Y . (2004). Development of solid-phase microextraction methodology for analysis of headspace volatile compounds in simulated beef flavour. Food Chemistry, 88, 141-149. Nakai, S. (1983). Structure-function relationships of food proteins with an emphasis on the importance of protein functionality. Journal of Agricultural and Food Chemistry, 31, 676-  683. Nakamura, T., Utsumi, S., Kitamura, K., Harada, K., and Mori, T. (1984). Cultivar differences in gelling characteristics of soy bean glycinin. Journal of Agricultural and Food Chemistry, 32, 647-651. O'Keefe, S. F., Wilson, L. A., Resurreccion, A. P., and Murphy, P. A. (1991). Determination of the binding of hexanal to soy glycinin and P-conglycinin in an aqueous model system using a headspace technique. Journal ofAgricultural and Food Chemistry, 39, 1022-1028.  Przybycien, T. M . and Bailey, J. E. (1988). Structure-function relationships in the inorganic saltinduced precipitation of a-chymotrypsin. Biochimica et Biophysica Acta, 995, 231-245. Przybycien, T. M . and Bailey, J. E. (1991). Secondary structure perturbations in salt-induced protein precipitates. Biochimica et Biophysica Acta, 1076, 103-111. Puppo, E., Chapleau, N . , Speroni, F., de Lamballerie-Anton, M . , Michel, F., Anon, C , and Anton, M . (2004). Physicochemical modifications of high-pressure-treated soybean protein isolates. Journal ofAgricultural and Food Chemistry, 52, 1564-1571.  Renkema, J. M . S. and van Vliet, T. (2002). Heat-induced gel formation by soy proteins at neutral pH. Journal ofAgricultural and Food Chemistry, 50, 1569-1573.  Rickert, D. A., Johnson, L . A . , and Murphy, P. A . (2004). Functional properties of improved glycinin and P-conglycinin fractions. Journal of Food Science, 69, 303-311. Thannhauser, T. W., Konishi, Y., and Scheraga, H. A. (1984). Sensitive quantitative analysis of disulfide bonds in polypeptides and proteins. Analytical Biochemistry, 138, 181-188. 174  Thissen, J. A . (1982). Interaction of off-flavor compounds with soy protein isolate in aqueous systems: Effects of chain length, functional group and temperature. M.Sc. Thesis, Iowa State University, Ames, p . l 19. Williams, R. W., and Dunker, A . K . (1981). Determination of the secondary structure of proteins from the amide I band of the laser Raman spectrum. Journal of Molecular Biology, 152, 783-813. Wilson, L. (1985). Flavor binding and removal of flavors from soybean protein. In R. Shibles (Ed.), World Soybean Research Conference III: Proceedings, (pp.  158-165). Boulder:  Westview Press. Wolf, W. J. (1993). Sulfhydryl content of glycinin: Effect of reducing agents. Journal of Agricultural and Food Chemistry, 41, 168-176.  Wu, W., Hettiarachchy, N . S., and Kalapathy, U . (1999). Functional properties and nutritional quality of alkali- and heat-treated soy protein isolate. Journal of Food Quality, 22, 119-133. Zhou, Q., and Cadwallader, K . R. (2006). Effect of flavor compound chemical structure and environmental relative humidity on the binding of volatile flavor compounds to dehydrated soy protein isolate. Journal of Agricultural and Food Chemistry, 54, 1838-1843.  175  CHAPTER 6. CONCLUSION  6.1. Overall conclusions as related to the proposed hypotheses  A reliable method for analysis of the S B F and mixtures o f SPI and S B F using H S - S P M E coupled with G C - F I D was established. The study in Chapter 2 demonstrated a successful application of fractional factorial experimental design based on Taguchi's orthogonal array to efficiently screen significant factors for H S - S P M E method for G C analysis of volatile components in S B F . The establishment of this reproducible and representative method for analysis o f volatile compounds to monitor changes in the headspace above samples containing S B F provided a consistent analytical technique throughout the thesis.  Identification of volatile compounds and determination o f odour-active components in S B F was investigated by using G C - F I D , G C - M S and G C - 0 in Chapter 3. The results revealed an intricate blend o f aroma compounds from several sub-classes in the headspace o f S B F . Moreover, G C - 0 by detection frequency method was very valuable to quantify the main odour-active compounds which contribute to the aroma profile o f S B F . Several sulfur- and nitrogen-containing compounds as well as various terpenoids were proposed to be of essential importance for the flavour profile of the SBF. Although one of the most powerful odour-active compounds identified in S B F was 2methyl-3-furanthiol, which has been reported to possess meat-like flavour, it was not selected as one of the indicator peaks to represent beefy note in S B F due to its low concentration compared to other volatile compounds. Instead, furans and nitrogen containing compounds such as 3methyl furan, 2-acetyl furan, 2-ethyl-3,6-dimethylpyrazine, 2,3-diethyl-5-methylpyrazine, and 2isoamyl-6-methylpyrazine were selected as indicator peaks to monitor the release of beef attribute related aroma compounds by SPI under various conditions. It was also demonstrated that volatile compounds in S B F included less sulfur- and nitrogen-containing compounds but many more alicyclic hydrocarbons including a great number of terpenoids than boiled or roasted beef. This difference might contribute to differences in aroma characteristics o f S B F and the real beef system.  The results presented in Chapter 4 showed that descriptive analysis along with G C analysis was 176  useful to monitor and describe changes in aroma characteristics of S B F upon increasing SPI content in the sample mixture. Moreover, well trained panelists could be o f great importance in precisely analyzing the aroma profile of samples, which can not be achieved with any other analytical instrument. The results in Chapter 4 strongly demonstrated that "roasted", "beefy", and "yeasty" notes were highly positively correlated with S B F concentration in the S B F and SPI sample mixtures, and the beefy related notes were severely suppressed by increasing SPI content. Using P C A , differences in aroma profile o f the mixtures with various combinations o f SBF and SPI were effectively expressed. Moreover, increased "yeasty" notes in heated samples compared to non heat treated samples were detected by using PC3. Finally, selection o f indicator peaks was accomplished to represent "beefy" characteristics in the sample satisfying two criteria set by the analysis of G C - F I D and G C - O . A l l 15 selected indicator peaks were odour-active compounds in G C - O analysis, and at the same time were significantly positively correlated with "beefy" characteristics analyzed by D A , therefore the indicator peaks could monitor the changes in aroma perception by retention o f beef flavour in samples containing SPI due to SPI-SBF interactions.  Research results from Chapter 5 clearly showed that interaction of SPI with odour-active compounds in S B F was affected by addition of different ingredients. Addition of ascorbic acid led to reduction of disulfide bonds in SPI, increase in surface hydrophobicity, and relocation of interior hydrophobic side chain residues in SPI to the less hydrophobic exterior. The denaturing effect was enhanced by the combined treatment of ascorbic acid and polyethylene glycol. Conformational changes were induced by cleavage of the disulfide bond by a reducing agent, ascorbic acid, and by the interactions between hydrophobic side chains of aromatic and aliphatic amino acid residues and the repeating hydrophobic regions of polyethylene glycol. Possible intermolecular hydrogen bonding between tyrosine residues and polyethylene glycol was also proposed. In turn, the resultant structural changes in SPI seemed to affect the flavour holding capacity of SPI. Results of G C analysis showed that indicator peak areas of beef attribute related volatile compounds were increased by addition of ascorbic acid alone or with polyethylene glycol, and that these were also expressed by enhancement of roasted note and diminished soymilk-like and cereal notes in the mixture of S B F and ingredients added SPI. Therefore, it could be postulated that suppression of beef flavour in S B F due to binding o f odour-active components with SPI could be improved by addition of ascorbic acid especially in conjunction with 177  polyethylene glycol, resulting in maximizing of S B F perception in the mixture of S B F and SPI.  6.2. Significance of this thesis research to the field of study  McGorrin (1996) classified the nature of interactions between volatile aroma compounds and non-flavour food matrix components into three types, i.e., binding, partitioning, and releasing. Considerable research has been conducted and reported in the literature on the binding of volatiles to proteins using model systems with individual chemicals or a series of chemicals, including lipid oxidation products responsible for off-flavour o f soy product. Although these studies have provided useful information on binding properties such as binding constants and number o f binding sites o f each volatile compound by determination o f thermodynamic parameters related to binding, their application to the food system is limited.  A significant aspect o f the present research described in this thesis is that the interaction between SPI and S B F was investigated from the flavour "releasing" point of view, focusing on the "availability of flavour compounds from the bulk food into the gas phase for sensory perception" based on flavour holding capacity o f SPI, rather than flavour "binding", explained as "retention or absorption of volatile compounds onto non-volatile substrates" (McGorrin, 1996).  In the present study, commercially available S B F and SPI were used, rather than a model system with individual volatile compounds. Therefore, the results may be more practical and could easily be transferred to food processing applications. Furthermore, the results from this study could form a constructive basis to formulate strategies to enhance the beefy aroma perception in soy protein based products.  This thesis also provided information on the identity o f volatile compounds in S B F and, through the comparison o f the components in S B F with cooked beef such as roasted or boiled beef, similarities and differences between the simulated beef flavouring and authentic beef were discussed. Moreover, odour-active components contributing to the aroma profile of S B F were documented using G C - O . Knowledge regarding identification o f volatile compounds and odouractive components in S B F may provide the basis to develop effective flavouring products to 178  simulate beef flavour by the flavour industry.  In addition, selected indicator peaks should be useful in the future to monitor interaction between SBF and SPI in terms of beefy flavour releasing under various environmental conditions. The use of indicator peaks may reduce the number of sensory evaluations, which are generally time- and labour-consuming work and cannot be replaced easily. Furthermore, the procedure to select the indicator peaks may be applied to other food systems during food product development or processing.  6.3. Suggestions for future research  There is no doubt that flavour is one of the most important factors for consumers to select food products. Research over the last several decades on flavour and food component interactions in either model system or food matrix has shown that food constituents such as proteins, polysaccharides, and lipids interact with flavour compounds (Chevance and Farmer, 1998; Mottram, 1998; Hau et al., 1996).  Since attenuation of even a single odour-active component  may have a remarkable effect on flavour perception, the flavour-binding behavior of food components must be considered in the development of food products. In this thesis interactions of SPI in terms of flavour holding behavior toward volatile compounds in SBF were examined. However, in a real food system, proteins comprise merely one of the many components in a food product. Therefore, flavour interactions between flavours and lipids, carbohydrates, proteins, or complex food components as a function of different pH, temperature, humidity and pressure should be investigated.  When a food product is consumed, the aroma of the food is perceived in two different steps. Food flavour isfirstperceived by olfactory epithelia in our nose while a food is being served. The other route involves perception of aroma by releasing of the aroma compounds from food while the food is masticated with saliva at body temperature in our mouth. Volatile compounds should be transported from the saliva phase to the air phase in the mouth, and then to the olfactory receptors in the nose, which is affected by the rate of breathing, swallowing and salivation (Taylor, 2002). Therefore understanding of aroma perception relating to odour 179  compounds release would be imperative to improve flavour quality in food products. Nonvolatile compounds that contribute to the taste of the S B F and perception o f taste of the flavour in gustatory receptors o f taste buds would be another area to be elucidated, which can be explored using high performance liquid chromatography.  In addition, variance in composition o f SPI and S B F should be considered. Zhou and Cadwallader (2006) investigated the binding properties of three SPIs from different origins and reported that those SPIs showed similar flavour binding patterns although absolute flavour binding capacities were not the same. More research results on the effects of variability of SPI and S B F from different origins or processing methods should be gathered. A s indicated in Chapter 5 o f this thesis, the SPI used in this study was already denatured. It would be worth performing research using SPIs with different degrees of denaturation.  Some o f the odour impact compounds relating to beefy note in S B F were sulfur containing compounds. Given the knowledge that most sulfur containing compounds have low thresholds, i.e., high aroma impact, further investigations that focus specifically on changes of the sulfur containing compounds under various conditions would be valuable. Although G C - F I D is very sensitive for most volatile compounds, usually the concentration of sulfur containing compounds in food ingredients are extremely low. Gas chromatography coupled with flame photometric detector (FPD), which is a specific detector for sulfur- or phosphorus-bearing compounds (Patterson et al., 1978; Y a o et al., 2001) may be a complementary technique in investigations focused on sulfur containing compounds.  180  6.4. References  Chevance, F. F. V., and Farmer, L. J. (1998). Effect of fat on flavour release from frankfurters. In P. Schieberle (Ed.), Interaction of Food Matrix with Small Ligands Influencing Flavour and  Texture, Vol. 3, (pp.55-58). Food science and technology COST 96. Luxembourg: European Communities. Hau, M. Y. M , Gray, D. A., and Taylor, A. J. (1996). Binding of volatiles to starch. In R. J. McGorrin and J. V. Leland (Eds), Flavor-Food  Interaction, ACS Symposium Series 663,  (pp.109-117). Washington, D.C.: American Chemical Society. McGorrin, R. J. (1996). Introduction. In R. J. McGorrin and J. V. Leland (Eds), Interaction,  Flavor-Food  ACS Symposium Series 663, (pp.ix-xii). Washington, D.C.: American  Chemical Society. Mottram, D. S. (1998). The interaction thiol and disulfide flavour compounds with proteins in food. In P. Schieberle (Ed.), Interaction of Food Matrix with Small Ligands Influencing  Flavour  and Texture, Vol. 3, (pp.50-54). Food science and technology COST 96.  Luxembourg: European Communities. Patterson, P. L., Howe, R. L., and Abu-Shumays, A. (1978). Dual-flame photometric detector for sulfur and phosphorus compounds in gas chromatograph effluents. Analytical Chemistry, 50, 339-344. Taylor, A. J. (2002). Release and transport of flavors in vivo: Physicochemical, physiological, and perceptual considerations. Comprehensive Reviews in Food Science and Food Safety, 1,  45-57. Yao, Z., Jiang, G., Liu, J., and Cheng, W. (2001). Application of solid-phase microextraction for the  determination  of organophosphorus  pesticides in aqueous samples by gas  chromatography with flame photometric detector. Talanta, 55, 807-814. Zhou, Q. and Cadwallader, K. R. (2006). Effect of flavor compound chemical structure and environmental relative humidity on the binding of volatile flavor compounds to dehydrated soy protein isolate. Journal of Agricultural and Food Chemistry, 54, 1838-1843.  181  APPENDIX  182  Appendix I. Correlation matrix o f mean value o f the peak area in G C chromatogram and the mean scores of the 5 sensory attributes . 1  RT  Peak #  1.717  Pl  2  Roasted  Soymilk-like  Cereal  Beefy  Yeasty 0.757 ***  0.734 **  -0.742 **  -0.756 **  0.710 *  4.144  -0.735 **  0.740 **  0.746 **  -0.702 *  -0.824 **  4.188  -0.811 **  0.819 **  0.824 **  -0.793 **  -0.883 ***  6.212  P5  0.529  -0.502  -0.500  0.451  7.832  P7  0.701 *  -0.670 *  -0.673 *  0.658 *  9.833  Pll  10.250  12.046  0.312  -0.240  -0.389  0.435  -0.429  -0.436  0.433  0.463  0.514  -0.555  -0.573  -0.167  12.428 13.205  0.301  0.864 * * *  P17  P20  13.822  14.881 15.014 15.772  P24  16.133  (P81)  3  0.523 -0.877 *** 0.221  -0.858 * * * 0.209  0.736 **  -0.728 **  -0.707 *  0.615 *  -0.634 *  -0.633 *  -0.144  14.726  0.711 *  -0.283  -0.557  11.605  0.498  0.193  0.201  0.872 * * * -0.080 0.713 ** 0.648 * -0.125  0.797 ** -0.191 0.739 ** 0.633 * -0.236  0.804 **  -0.787 **  -0.787 **  0.773 **  0.811 **  0.745 **  -0.696 *  -0.696 *  0.696 *  0.721 **  0.686 *  -0.596 *  -0.598 *  0.601 *  0.664 *  0.895 * * *  -0.874 * * *  -0.873 ***  0.852 *** -0.645 *  0.914 *** -0.663 *  0.754 **  0.754 **  0.887 * * *  -0.834 **  -0.844 **  0.827 **  0.847 **  0.460  -0.474  -0.452  0.495  0.387  17.338  0.544  -0.561  -0.596 *  0.526  0.581 *  17.570  0.568  -0.619 *  -0.643 *  0.519  0.613 *  17.930  0.426  -0.394  -0.388  0.329  0.371  0.631 *  -0.635 *  -0.633 *  0.605 *  0.657 *  18.740  0.340  -0.224  -0.217  0.321  0.269  19.673  0.553  -0.565  -0.595 *  0.486  0.560  0.580 *  -0.570  -0.558  0.590 *  0.491  0.038  -0.155  -0.176  0.068  0.149  0.569  -0.623 *  -0.642 *  0.678 *  0.692 *  0.624 *  -0.617 *  -0.606 *  0.545  0.605 *  0.601 *  -0.581 *  -0.591 *  0.544  0.589 *  0.442  -0.339  -0.344  0.443  0.395  0.694 *  -0.709 *  -0.702 *  0.715 **  -0.636 *  16.526 16.864 17.064  18.199  19.774  P27  (P84)  3  P30  20.198 20.517  P32  20.770 21.196  P33  21.488 22.062  P35  -0.254  23.145 23.380 24.103  P38  0.318  0.322  -0.225  0.664 * -0.377  0.643 *  -0.624 *  -0.654 *  0.594 *  0.634 *  0.671 *  -0.634 *  -0.646 *  0.599 *  0.646 *  24.682  0.004  0.001  0.026  0.058  -0.076  25.034  0.599 *  -0.574  -0.564  0.537  0.572  25.374  0.596 *  -0.598 *  -0.594 *  0.598 *  0.593 *  0.805 **  -0.798 **  -0.803 **  0.794 **  0.823 **  0.722 **  -0.707 *  -0.730 **  0.697 *  0.733 **  25.703 25.948  (P87)  3  183  RT  Peak#  2  Roasted *  Soymilk-like *  Cereal *  Beefy  Yeasty  0.684 *  0.646 *  -0.527  0.496  0.539  -0.759 **  0.752 **  0.757 **  26.223  0.710  26.707  0.542  -0.526  26.866  0.747 **  -0.764 **  26.983  0.397  -0.403  -0.397  0.310  0.383  27.154  -0.563  0.475  0.476  -0.574  -0.550  -0.327  -0.469  27.209  P43  -0.355  -0.634  27.742  0.875 ***  0.442 -0.897 ***  27.968  0.515  -0.504  0.899 ***  -0.879 ***  28.204  P44  28.322  -0.396  0.267 -0.648 *  29.038  0.756 **  -0.685  *  29.384  0.068  29.519  0.556  29.767 29.888  P45  P48  30.140 30.291  P49  30.648  P50  0.255 -0.626 *  0.902 ***  0.487  0.526  0.931  ***  -0.388  0.928 *** -0.281  0.678  -0.688 *  0.713  **  -0.061  -0.039  0.127  -0.008  -0.545  -0.543  0.509  0.555  0.375  -0.239  -0.238  0.337  0.360  0.904 ***  -0.885 ***  -0.887 ***  0.928 ***  0.934 ***  0.626 *  -0.633 *  -0.633 *  0.617 *  0.666 *  -0.890 *** -0.702 *  -0.895 ***  0.850 ***  0.924 ***  -0.678 *  0.645 *  0.675 *  -0.905  ***  -0.906  ***  -0.866  ***  -0.871 ***  0.879  ***  0.914  *** ***  30.849  -0.881 ***  0.887 ***  *  0.723 **  30.525  0.443 -0.900 *** •-0.507  0.715 **  28.746  -0.640  0.681 * 0.738 **  ***  0.937 ***  ***  0.868 0.638 *  0.896 ***  0.914 *** 0.750 **  0.953 ***  0.929  31.162  0.890 0.674 *  31.238  0.881  -0.894  ***  31.438  0.736 **  -0.753  **  0.896 ***  -0.886 ***  -0.888 ***  0.924 ***  0.931 ***  32.148  0.867 ***  -0.868 ***  -0.870 ***  0.879  ***  0.900 ***  32.438  0.624 *  0.618  *  0.666 *  31.813  P54  ***  ***  32.598  0.859  33.083  0.813 **  33.225  P58  0.875 *** **  33.540  0.778  33.727  0.717 ** -0.704 *  33.935  **  34.130  0.782  34.261  0.779 **  34.403  0.585 *  34.610  P67  0.868  ***  -0.679 *  -0.630  *  -0.857  ***  -0.828 ** -0.872  ***  -0.772  **  -0.705 * 0.812  **  -0.770  **  -0.780 ** -0.589  *  -0.858  *»*  -0.677 * -0.911  ***  -0.753  **  -0.632 * -0.857  ***  -0.828 ** -0.873  ***  -0.771 ** -0.694 *  0.684 * 0.762 **  0.863 ***  0.896 ***  0.826 **  0.854 ***  0.894 ***  0.908 ***  **  0.817 **  0.689 * -0.715 **  0.727 ** -0.726 **  0.779  0.818  **  -0.766  **  0.775 **  0.811 **  -0.780 **  0.787 **  0.817 **  -0.593  *  -0.859  ***  0.600  *  0.632 *  0.881  ***  0.899 ***  0.781  **  0.816 **  ***  0.899 ***  34.780  0.792 **  -0.764 **  -0.762 **  35.011  0.858 ***  -0.864 ***  -0.864 ***  0.872  35.172  0.214  0.353  0.859 ***  -0.258 -0.859 ***  -0.278  35.283  -0.855 ***  0.823  35.562  0.544  -0.510  -0.512  0.503  0.541  0.869 ***  -0.866 **»  -0.870 ***  0.883 ***  0.905 ***  0.675 *  -0.735 **  -0.742 **  0.685 *  0.748 **  35.680  P71  35.816 36.093 36.242  -0.458 P72  0.810 **  0.496 -0.813 **  0.494 -0.816 **  0.359 **  -0.456 0.822 **  0.868 ***  -0.495 0.853 #**  184  RT  Peak #  Roasted  Soymilk-like  Beefy  Yeasty  36.565  0.792 **  -0.780 **  -0.774 **  0.780 **  0.813 **  36.755  0.533  -0.519  -0.521  0.495  0.537  36.948  0.789 **  -0.793 **  -0.798 **  0.800 **  0.842 **  37.071  2  0.254  -0.267  Cereal  0.248  -0.332  -0.232  0.793 **  -0.787 **  -0.789 **  0.778 **  0.819 **  37.582  0.805 **  -0.816 **  -0.809 **  0.815 **  0.803 **  37.921  0.757 **  -0.755 **  -0.759 **  0.746 **  0.790 **  38.081  0.153  -0.167  -0.167  0.169  0.251  38.125  0.253  -0.264  -0.265  0.276  0.315  0.668 *  -0.653 *  -0.658 *.  0.634 *  0.715 **  0.774 **  -0.756 ** 0.613 *  -0.756 ** 0.605 *  0.767 ** -0.632 *  0.788 **  -0.679 *  -0.675 *  0.672 *  37.263  38.342  P75  P78  38.548  39.030  -0.610 * 0.674 *  39.148  -0.600 *  0.604 *  0.596 *  -0.618 *  -0.575  39.281  0.782 ** 0.625 *  0.783 ** 0.618 *  -0.694 * -0.650 *  -0.647 *  39.462  -0.670 * -0.642 *  -0.620 *  39.758  -0.524  0.484  0.477  -0.518  -0.430  39.865  -0.719 **  0.693 *  0.674 *  -0.703 *  -0.669 *  40.126  -0.729 **  0.653 *  0.647 *  -0.753 **  -0.653 *  38.906  P79  40.261  0.160  40.403  0.721  40.512  -0.502  0.527  0.505  -0.535  40.666  0.362  -0.387  -0.389  0.371  **  40.887  -0.800 **  40.927  0.616 *  -0.570 0.713 **  -0.280  -0.272  0.217  0.236  -0.709 *  -0.706 *  0.710 *  0.753 **  0.774 ** -0.647 *  0.767 ** -0.653 *  -0.441  -0.801 **  0.419 -0.755 **  0.626 *  0.699 *  **, and ***, significant at p < 0.05, p < 0.01, and p < 0.001, respectively. The peaks were numbered based on G C - M S identification of volatile components in SBF (Moon et al., 2006). Peaks in this table without peak numbers were either not identified by G C - M S or were components derived from SPI. 3  The peak numbers P81, P84, and P87 were assigned to these peaks due to high aroma value  (over 75 % detection frequency) in GC-olfactometry analysis, although they were not identified by G C - M S analysis (Moon et al., 2006).  185  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0100487/manifest

Comment

Related Items