Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Assessment of bacteriophage-insensitive culture bacteria for cheddar cheese making and subsequent discriminant… Amantea, Gerald Fiore 1984

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

Item Metadata

Download

Media
831-UBC_1984_A1 A48.pdf [ 6.87MB ]
Metadata
JSON: 831-1.0096248.json
JSON-LD: 831-1.0096248-ld.json
RDF/XML (Pretty): 831-1.0096248-rdf.xml
RDF/JSON: 831-1.0096248-rdf.json
Turtle: 831-1.0096248-turtle.txt
N-Triples: 831-1.0096248-rdf-ntriples.txt
Original Record: 831-1.0096248-source.json
Full Text
831-1.0096248-fulltext.txt
Citation
831-1.0096248.ris

Full Text

ASSESSMENT OF BACTERIOPHAGE-INSENSITIVE CULTURE BACTERIA FOR CHEDDAR CHEESE MAKING AND SUBSEQUENT DISCRIMINANT ANALYSIS FOR OBJECTIVE FLAVOUR EVALUATION by Gerald F io re Amantea B . S c , Un ivers i t y of B r i t i s h Columbia, 1971 M . S c , Un ive rs i t y of B r i t i s h Columbia, 1973 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES DEPARTMENT OF FOOD SCIENCE We accept th i s thes i s as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA SEPTEMBER 1984 © Gerald F io re Amantea, 1984 In presenting t h i s thesis i n p a r t i a l f u l f i l m e n t of the requirements for an advanced degree at the University of B r i t i s h Columbia, I agree that the Library s h a l l make i t f r e e l y available for reference and study. I further agree that permission for extensive copying of t h i s thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. I t i s understood that copying or publication of t h i s thesis for f i n a n c i a l gain s h a l l not be allowed without my written permission. Department of The University of B r i t i s h Columbia 1956 Main Mall Vancouver, Canada V6T 1Y3 Date E-6 (3/81) ABSTRACT The f i r s t part of t h i s two-part study deals with s ix def ined s t ra ins of bacter iophage- insens i t i ve Streptococcus cremoris used over a per iod of 10 months to produce more than 2 m i l l i o n kg of Cheddar cheese on continuous cheesemaking equipment. Flavour development in t h i s cheese was less than that in cheese made with conventional bulk s t a r t e r . P ro teo l ys i s and Theological propert ies were examined. Pattern recogni t ion techniques were used to analyze the mul t i va r i a te data . Texture was a f fec ted by the mechanical process , moisture content and y i e l d po in t . Casein p r o t e o l y s i s , age, cu l tu re type and firmness were the most d i sc r imina t ing var iab les a f f e c t i ng matur i ty . In part two, i t i s well recognized that a minimum maturation period is required before Cheddar cheese is acceptable to the consumer. This is a lengthy and cos t l y procedure in which qua l i t y i s based on subject ive eva lua t ion . C l a s s i c a l l y , t ra ined graders and sensory panels have performed t h i s duty, but because of the v a r i ab l e , subject ive and time consuming nature or organolept ic methods, a s imple , more object ive and r e l i a b l e method for accurate ly assessing cheese f lavour i s proposed. Sensory evaluat ions by t ra ined graders were done on more than 60 commercially produced cheese samples, and water soluble f rac t ions prepared from a l l samples. A ternary gradient system was used to e lute the non-vo la t i le f l avour components from an Adsorbosphere Cs reverse phase column. A new mapping simplex opt imizat ion technique was appl ied to the HPLC p r o f i l e s to optimize separat ion of the multi-component mixture. More than 45 peaks were obtained using an i n i t i a l solvent volume r a t i o of 44 .6 :0 :55 .4 of t r i f l u o r o a c e t i c ac id (0.1%), a c e t o n i t r i l e and water. Over 56.6 min the r a t i o was changed to 0:36.6:63.4 at a flow-rate of 0.97 mi/min. The i i i . optimized gradient system was super ior to previous i s o c r a t i c separations and the e lu t ion patterns d i f f e r ed for the various categor ies of cheese. S t a t i s t i c a l pattern recogni t ion techniques - p r inc ipa l component ana lys is and stepwise l i nea r d iscr iminant ana lys is - were used to in te rpre t the HPLC p r o f i l e s . Cheese samples were co r rec t l y c l a s s i f i e d according to t he i r d iscr iminant funct ions into groups. The technique d i f f e r en t i a t ed between f i r s t grade and downgraded samples and was capable of assessing cheese f lavour qua l i t y at an ear ly age. i v . TABLE OF CONTENTS Part I : Assessment of bacteriophage-insensitive culture bacteria for continuous Cheddar cheese making. Page INTRODUCTION 1 LITERATURE REVIEW 4 Propagat ion, se l ec t ion and commercial app l i ca t ion 6 Composition 8 Texture 9 MATERIALS AND METHODS 13 1. Cheese manufacture 13 2. Compositional ana lys is 13 3. Sensory evaluat ion 14 4. Culture propagation 14 5. S tar ter a c t i v i t y 15 6. Bacteriophage t e s t i ng 17 7. Assaying for free amino groups 17 8. Preparat ion of cheese samples 20 9. Textural evaluat ion 20 10. Sample preparat ion fo r HPLC ana lys is 21 11. S t a t i s t i c a l ana lys is 21 12. HPLC ana lys is 21 RESULTS AND DISCUSSION 23 Rheological propert ies 23 E f fec t of moisture content 28 V. Page E f fec t of p ro teo lys i s 33 E f fec t of heat treatment 37 Stepwise l i nea r d iscr iminant ana lys i s (age) 39 Stepwise l i nea r d iscr iminant ana lys is ( texture) 43 HPLC ana lys is 51 CONCLUSIONS 54 Part I I : Optimization of HPLC resolution and subsequent discriminant analysis for objective cheddar cheese taste evaluation. INTRODUCTION 57 LITERATURE REVIEW 61 Cheddar cheese f lavour 61 Amino acids 61 Free fa t ty acids 64 V o l a t i l e su l f u r compounds 66 Non-sulfur v o l a t i l e compounds 67 Ana ly t i ca l techniques 68 Reversed-phase l i q u i d chromatography 69 Mobile phase 70 Mu l t i va r i a te ana lys is 73 MATERIALS AND METHODS 77 1. Sample preparat ion 77 2. Nitrogen determinations 79 3. Compositional ana lys is 79 v i . Page a) Fat 79 b) Moisture 79 c) Sa l t 80 d) pH 80 4. Sensory evaluat ion 80 5. HPLC ana lys i s 81 6. Opt imizat ion techniques 82 a) Super modif ied simplex 82 b) Mapping super simplex and simultaneous fac tor s h i f t 84 7. S t a t i s t i c a l ana lys is 84 a) P r inc ipa l component ana lys is 85 b) Stepwise l i n ea r d iscr iminant ana lys i s 86 8. Amino ac id ana lys is 88 9. Ana lys is of HPLC f rac t ions 89 RESULTS AND DISCUSSION 89 Sample preparat ion 89 Nitrogen determinations 89 Compositional ana lys is 90 Sensory evaluat ion 92 HPLC ana lys i s , 92 Optimizat ion 99 a) Super modif ied simplex 99 b) Mapping super simplex 104 c) Simultaneous f ac to r s h i f t 109 v i i . Page P r inc ipa l component ana lys is 109 Regression on p r inc ipa l components 113 Stepwise l i nea r d iscr iminant ana lys is 113 Amino ac id ana lys is 119 Ana lys is of the HPLC f rac t ions 123 CONCLUSIONS 124 LITERATURE CITED 132 v i i i . LIST OF FIGURES Figure Page 1.1 Schematic ou t l i n ing the whey-based s ta r t e r system. 16 2.1 Method used to detect the presence of bacteriophage in whey with bromocresol purple as an i nd i c a to r . 19 3.1 Force-compression curves fo r a m i l d , medium and aged cheese (samples number 18, 19 and 17 respect i ve ly from Table 4 .1 ) . 24 4.1 Force-compression curves fo r cheese of approximately the same age, composition and degree of p ro teo l ys i s but varying in moisture content. Samples number 12, 6 and 7 with moisture of 36 .0 , 37.3 and 38.5% respec t i ve l y . See Tables 2 .1 , 3 .1 , 4 .1 . 29 5.1 Var ia t ion in the force at the y i e l d point for cheese samples number 1, 2, 4-14, 16 sub jec t i ve ly evaluated weak or s l i g h t l y weak tex ture . See Table 2.1 fo r cheese composition (r = 0.788). 30 6.1 Var ia t ion in the free amino groups for cheeses of d i f f e r en t ages commercially produced using e i ther S^ . cremoris or S^ . l a c t i s cu l ture s t r a i n s . See Table 4 .1 . ( r 2 = 0.862). 35 7.1 E f fec t of heat treatment for cheese milk on force-compression curves for commercially produced cheese of approximately the same age, composition and degree of p r o t e o l y s i s , a , 63°C/16s; b, 72°C/16s; (Samples no 3 and 5 respect i ve ly Table 4 .1 ) . 38 8.1 Canonical p lot fo r age fo r the 36 commercially produced cheeses using the mul t i va r i a te data in Table 4 .1 . A, X-aged; 0, o l d ; C, medium; M, m i l d . Group means ind icated by c i r c l e s . 44 9.1 Canonical p lot for texture for the 36 commercially produced cheeses using the mul t i va r i a te data in Table 4 .1 . F, f i rm ; S, s l i g h t l y weak; w, weak. Group means indicated by c i r c l e s . 47 10.1 Va r i a t ion in the force at the y i e ld-po in t for cheese samples number 1-11, 14, 15, 18, 24-29 grouped as mild by SLDA and approximately same composition but varying moisture. See Table 3.1 for composition (r = 0.853). 49 11.1 Var ia t ion in the force at the y i e ld-po in t for cheese samples number 1-16 ( r 2 = 0.817). Numbers 4, 7-11, 13 of approximately the same composition but varying moisture c l a s s i f i e d by SLDA as weak or s l i g h t l y weak tex ture . See Table 3.1 for composit ion. 50 i x . Figure Page 12.1 HPLC p r o f i l e of the water extract from a 199 day o ld cheese, made with _S. cremoris and sub jec t i ve ly graded m i l d . See Table 4.1 (sample no 2) . 52 13.1 HPLC p r o f i l e of the water extract from a 136 day o ld cheese, made with j>. l a c t i s and sub jec t i ve ly graded medium. See Table 4.1 (sample no 4 ) . 53 14.2 Schematic showing enzymatic hydro lys is of casein by rennet, milk proteinase and s t a r t e r prote inase . 63 15.2 Flow diagram of preparat ion of cheese f r a c t i o n s . 78 16.2 Schematic ou t l i n ing method used to ca l cu la te peak r e so lu t i on , adapted from Glajch et aj_. (1980). 84 17.2 RP/HPLC separat ion of the water-soluble f r a c t i on from a young cheese. 94 18.2 RP/HPLC separat ion of the water-soluble f r a c t i on from a mild cheese. 95 19.2 RP/HPLC separat ion of the water-soluble f r ac t i on from a medium cheese. 96 20.2 RP/HPLC separat ion of the water-soluble f r a c t i on from an o ld cheese. 97 21.2 RP/HPLC separat ion of the water-soluble f r ac t i on from a X-aged cheese. 98 22.2 RP/HPLC separat ion of the water-soluble f r a c t i on from a b i t t e r cheese. 100 23.2 Opt imizat ion of the mobile phase to increase peak reso lu t ion of the water-soluble extract from a X-aged cheese sample. An Adsorbosphere Cg column was used for a l l chromatographic runs. 103 24.2 Mapping resu l ts of experiments to maximize peak r e s o l u t i o n . Factor 1, flow rate with const ra in ts 0.5-2.0 mL/min. 105 25.2 Mapping resu l t s of experiments to maximize peak r e s o l u t i o n . Factor 2, a c e t o n i t r i l e concentrat ion with const ra in ts 10 - 40%. 106 X. Figures Page 26.2 Mapping resu l ts of experiments to maximize peak r e so lu t i on . Factor 3, time with const ra ints 30 - 60 min. 107 27.2 Mapping resu l ts of experiments to maximize peak r e so lu t i on . Factor 4, TFA concentrat ion 0 - 100%. 108 28.2 P r inc ipa l component plot of f ac to r 1 vs f ac to r 2. 112 29.2 Canonical p lot of the group means fo r the 60 cheese samples. 115 30.2 Canonical p lot of the 60 cheese samples. A, X-aged; 0, o l d ; C, medium; M, m i l d ; Y, young. Overlap ind icated by a s t e r i s k . 116 31.2 Canonical p lot of the cheese samples showing loca t ion of the unknown samples, X. Overlap indicated by a s t e r i s k . 118 32.2 Canonical plot of the cheese samples showing loca t ion of the b i t t e r samples. A, X-aged; 0, o l d ; C, medium; M, m i ld ; Y, young; B, b i t t e r . Overlap indicated by a s t e r i s k . 120 33.2 Amino acid p r o f i l e of water-extract from an X-aged cheese sample. 121 LIST OF TABLES Table Page 1.1 1 and 3% act ive cu l ture inoculated into reconst i tu ted non-fat-dry-milk powder conta in ing bromocresol purple and showing e f fec t on pH a f ter 1.5 and 3 h is incubat ion at 30°C r e spec t i v e l y . 2.1 Compositional analys is of the commercially produced 26 cheeses. 3.1 Rheological data fo r the commercially produced cheeses of 27 d i f f e r en t ages, composition and cu l ture type. 4.1 Rheo log i ca l , composit ional and sensory evaluat ion of the commercially produced cheese samples of various ages using S_. cremoris or S_. l a c t i s cu l tu re s t ra ins on both 31 conventional and automated cheese processing equipment. 5.1 F-matrix among the four age groups (mi ld , medium, o l d , X-aged) fo r the cheese samples af ter d iscr iminant 41 ana l y s i s . 6.1 C l a ss i f i c a t i on-mat r i x fo r age generated by stepwise d iscr iminant ana l ys i s . 100% of the cases were co r r e c t l y 42 c l a s s i f i e d using F-to-enter of 4 .0 . 7.1 F-matrix among the three texture groups ( f i rm, s l i g h t l y weak, weak) f o r the cheese samples af ter d iscr iminant 45 ana l ys i s . 8.1 C l a ss i f i c a t i on-mat r i x fo r texture generated by stepwise d iscr iminant ana l ys i s . 94.4% of the cases were co r r e c t l y 46 c l a s s i f i e d using F-to-enter of 4 .0 . 9.2 Nitrogen (N) d i s t r i b u t i o n of the various f r ac t ions from 91 mild and aged cheese. 10.2 Eigenvalue (VP), sum of VP, and cumulative proport ion in 110 to ta l var iance (%) in a p r inc ipa l component ana l y s i s . 11.2 Amino acid analys is of the tota l water-soluble extract 122 from various cheeses. x i i . LIST OF APPENDICES APPENDIX Page I. Factor loading and communality matr ix. 126 II. Table showing co r re l a t i on between p r inc ipa l components and dependent va r i ab l e , regression c o e f f i c i e n t s of p r inc ipa l components and c o e f f i c i e n t s of var iab les obtained from regression on p r inc ipa l components. 128 III. Eigenvalues for p r inc ipa l components 1 through 10 fo r each peak obtained from the HPLC p r o f i l e . 130 x i i i . ACKNOWLEDGMENT I would l i k e to express s incere graditude to my research superv i sor , Dr. S. Nakai , for his constant encouragement and guidance during the course of th i s study and a lso to members of my committee, Drs . W.D. Powrie, J . Sim, B. Skura and M.A. Tung, who allowed me th i s opportuni ty . I wish to thank my wife Barbara for her continued support during th i s d i f f i c u l t t ime, without which th i s work could not have been completed. F i n a l l y , I would l i k e to dedicate th i s thes i s in lov ing memory of my f a the r , Jack. -1-INTRODUCTION Improved s t a r t e r cu l tures using bacter iophage-insensi t ive mutants fo r Cheddar making have received much a t tent ion during the past ten years and major advantages have been c la imed. Moreover, i t i s genera l ly recognized that good qua l i t y Cheddar cheese i s l a rge l y dependent on the s t a r t e r s t ra ins se lected (McDowall and Whelan, 1933). It i s f o r t h i s reason and because of the economic s i gn i f i c ance of Cheddar cheese as well as consumer acceptance that the dairy industry has placed great emphasis on cheese f lavour qua l i t y c o n t r o l . The aim of the present study was to evaluate the performance of six defined s t ra ins of bacter iophage-insens i t i ve Streptococcus cremoris grown in whey-based s t a r t e r media and subsequently used fo r commercial cheese production on automated continuous equipment. The performance of these s t ra ins were compared to conventional _S. 1 a c t i s cu l tu re propagated and used in the same manner. Unlike other studies that assess cu l ture performance by evaluat ing the cheese using t r ad i t i ona l methods with c l a s s i c a l l y t ra ined graders or sensory panels with l i t t l e or no importance given to ana l y t i ca l methodology or s t a t i s t i c a l techniques, the present study proposes a more object ive and r e l i a b l e approach. Moreover, t h i s approach i s necessary in that i t has been shown that the sensory evaluat ion of cheese may not be r e l i a b l e and var ies from grader to grader (Le l ievre and G i l l e s , 1982). The primary ob jec t i ve , therefore was to assess the e f f ec t of s t a r t e r cu l ture on texture and degree of cheese r ipening while at the same time evaluat ing f lavour qua l i t y . The texture of a cheese i s one of the important cha r a c t e r i s t i c s that determines the iden t i t y and q u a l i t y . However, most studies re l a t ing to cheese texture involve cheese matur i ty , - 2 -tes t temperature, crosshead speed, sample height and surface area (Cu l i o l i and Sherman, 1976). Few studies have dealt with the i n t e r r e l a t i onsh ips among manufacturing parameters, compositional data and rheologica l p roper t i es . Lawrence and G i l l e s (1982) r e s t r i c t ed the i r study to only Cheddar cheese which had been manufactured to various composit ions. They found that the force compression curves were d i f f e ren t for d i f f e ren t cheese samples and were affected by the moisture content, pH and extent of casein p ro teo lys i s that had taken p lace . However, most studies are car r ied out on laboratory scale making i t sometimes d i f f i c u l t to re la te these resu l t s to commercial product ion. The present study was r e s t r i c t ed to only Cheddar cheese. Unl ike other s tud ies , the cheese was produced on continuous commercial scale equipment and varied in composit ion. Lawrence and G i l l e s (1980) showed that the qua l i t y of cheese was inf luenced by moisture in the non-fat-substance (MNFS), sa l t- in-moisture (S/M) l eve ls and pH and proposed optimum s p e c i f i c a t i o n l im i t s fo r production of qua l i t y cheese. Therefore , the present studies placed emphasis on quant i ta t i ve measurement of the cheese composit ion. Because of the subject ive nature of evaluat ing cheese f lavour by sensory analys is a procedure out l ined by McGugan et ^1_. (1979) was adopted. They proposed that the non-volat i le water extractable f r a c t i on was responsib le for the i n t ens i t y of cheese f l a vour . Therefore, ana lys is of these extracts from cheeses of approximately the same composit ion, age and manufacturing procedures but made by using the d i f f e ren t cu l tures would provide valuable information about the f lavour of the respect ive cheeses. Reversed phase high performance l i q u i d chromatography (RP/HPLC) shown to be - 3 -a valuable ana ly t i ca l tool in assessing cheese f lavour development (Pham and Nakai , 1984) was se lected to analyze the ex t r ac t s . The object ive was to quant i t a t i ve l y re la te p ro teo l y t i c cu l ture a c t i v i t y in the respect ive cheeses to t h e i r HPLC p r o f i l e . An addi t iona l aim of the study was to in te rpre t the mul t i va r i a te data by s t a t i s t i c a l pattern recogni t ion techniques. Although the technique i s well recognized and widely used in other d i s c i p l i n e s i t has only recent ly been appl ied to Food Science. Thus, the object ive was to se lec t from a l l the data the most d i sc r imina t ing factors or condit ions that s i g n i f i c a n t l y a f fected cheese texture and/or t a s t e . - 4 -LITERATURE REVIEW The basic process for making Cheddar cheese has changed l i t t l e : coagu la t ion , cooking, d r y - s t i r r i n g , matting or cheddaring, m i l l i n g , s a l t i n g and p ress ing , while the time required fo r making has remained about the same. Mechanization of the process for making Cheddar cheese to reduce labour costs and physical labour has received much a t t en t i on . Unfortunate ly , l i t t l e data ex i s t s for c r i t i c a l l y evaluat ing these systems with respect to qua l i t y of cheese and to uniformity of composition (Emmons, 1978). S ta r te r cu l tures remain a v i t a l necessi ty for manufacturing Cheddar cheese. Simple fool-proof methods of e l iminat ing bacter iophage, developing bacter iophage-insens i t ive s t ra ins or avoiding the consequences of phage i n f ec t i on have been e lus ive goals of the cheese indust ry . Research e f f o r t s to improve s t a r t e r cu l ture technology for Cheddar cheese making has received much at tent ion during the past several decades. The f lavour of a good qua l i t y Cheddar cheese is l a rge l y dependent on the s t a r t e r s t ra ins se lected (McDowall and whelan, 1933). Although i t is recognized that the primary funct ion of a s t a r t e r cu l ture is to produce l a c t i c a c i d , the se lec t ion of the cu l ture to be used i s very essent ia l to good cheese manufacturing procedures. The cu l ture must meet and sa t i s f y a number of important requirements inc lud ing essent ia l ac id production and f lavour development (Lawrence et aj_., 1976), phage i n s e n s i t i v i t y (Heap and Lawrence, 1976; Lawrence et a l_ . , 1978; Thunell et a l_ . , 1981), absence of bac te r ioc ins (Babel, 1977) and s e n s i t i v i t y to Cheddar cheese cook temperatures (Lowrie et aj_., 1974; McDowall and Whelan, 1933). - 5 -Since i t s development more than 40 years ago, cheese s t a r t e r systems based on da i l y rota t ion of phage-unrelated s ing le s t ra ins of l a c t i c organisms and aimed at prevention of bacteriophage attack during cheese making has been a p r inc ipa l concern of cheese manufacturers. Heap and Lawrence (1976) provided a better understanding of bacteriophage r e l a t i onsh ip s . They suggested that the major source of phages was the s t a r t e r cu l ture i t s e l f and phage leve ls in a commercial plant could be great ly reduced by avoiding the use of s t ra ins which could act as hosts for the phages present. Limsowtin and Terzaghi (1976) i so la ted three S^ . cremoris mutants based on t he i r s p e c i f i c phage res is tance and su i tab le ac id product ion . They were used in ro t a t i ons , each paired with another s ing le s t r a in and success fu l l y used in cheese production for more than nine months. A new mul t ip le s t a r t e r approach using s ix c a r e fu l l y se lected s t ra ins was developed (Limsowtin et a l_ . , 1977) and continuously used for eight months. Czulak and co-workers (1979) i so l a ted a s ing le phage res i s tan t s t r a in and reported i t s use in cheese manufacture for three years without f a i l u r e s caused by phage at tack . Thunell and co-workers (1981) developed phage i n s e n s i t i v e , mu l t i p l e-s t r a in s ta r te rs by plaquing cheese whey against potent ia l s ta r te r s t r a i n s . The s ix defined s t ra ins of Streptococcus cremoris have been used to produce more than 68 m i l l i o n kilograms of Cheddar cheese with no plants experiencing s ta r te r f a i l u r e from phage at tack. Phage-insensit ive mutant s t ra ins have been used success fu l l y s ing ly (Czulak et j i l_ . , 1979) in paired rotat ions (Limsowtin et^  a l_ . , 1977) and mul t ip le s t ra ins (Lawrence et aj_., 1978; Limsowtin and Te rzagh i , 1976; Thunell et a l_ . , 1981). - 6 -The major advantage of the mul t ip le s t r a in cu l tu re i s that i t can be used cont inuous ly , at a lower innoculum with cons is tent a c t i v i t y despite s t r a i n va r i a t i on and seasonal changes in milk composition (Limsowtin et a l . , 1977). Moreover, cheese f lavour uniformity i s obtained while the cheesemaker i s able to se lec t and mix compatible s t ra ins together to improve cheese qua l i t y by c o n t r o l l i n g s t a r t e r a c t i v i t y (Lawrence et a l . , 1978; Thunell et aj_., 1981). PROPAGATION, SELECTION AND COMMERCIAL APPLICATION The p r inc ipa l funct ion of l a c t i c ac id bacter ia during cheese making i s fermentation of lactose into l a c t i c a c i d . However, t h i s product of metabolism can reduce a c t i v i t y of l a c t i c bacter ia e spec i a l l y when the bacter ia l c e l l s are held at a pH below 5.0 f o r extended periods of t ime. Benef ic ia l e f f ec t s of neu t r a l i z i ng acids in l a c t i c cu l ture media have been recognized fo r some time, however, commercial app l i ca t i on has occurred only within the past f i ve years (Richardson ejt al_., 1981). A number of pH control systems have evolved: continuous n e u t r a l i z a -t ion during cu l tu re growth maintaining the pH between 5.9 to 6.1 by i n j e c t i on of l i q u i d or gaseous ammonia or ammonium hydroxide into the medium (Richardson et a]_., 1980); use of buffered s t a r t e r media, several which are ava i l ab le and proven successful commercially (Mermelstein, 1982). A va r i a t i on of pH contro l (Limsowtin et al_., 1980), in which cu l tures are grown in reconst i tuted nonfat-dry-milk (10 percent so l ids ) unt i l the pH reaches between 4.5 to 5.0, requires approximately 18 h of incubat ion . The s ta r te r i s then neutra l ized to pH 7.0 with sodium hydroxide and incubated for an addi t ional two hours. The pH f a l l s to about 5.0 during th i s per iod and remains s t a t i c f o r at l e a s t two hours. - 7 -Genera l ly , s t a r t e r cul tures produced with pH control and nutr ient exhaustion can be stored for in te rva l s that are considerably longer then have prev ious ly been poss ib le (Hong et_ al_. , 1977). These cul tures can be used at lower inoculum l e v e l s , 30-60% of that of conventional cu l tu res , and because of the high populat ion of uninjured c e l l s with a shorter lag time the cheese develops acid f as te r (Mermelstein, 1982). Procedures to determine the s u i t a b i l i t y of a s t r a in fo r use in mul t ip le s ta r te rs has been out l ined (Lawrence et , 1978). Cu l tura l c h a r a c t e r i s t i c s of phage-insensit ive mutants were compared to those of the i r respect ive parent s t ra ins and found to possess s im i l a r des i rab le cha r a c t e r i s t i c s and s t a b i l i t i e s (Thunell et^al_., 1981). Richardson and co-workers (1983) s e l e c t i v e l y grew proteinase-negative ( P r t - ) var iants of Streptococcus cremoris i n pH c o n t r o l l e d , buffered media and showed a reduced potent ia l for the development of b i t t e r f lavour when used exc lus i ve l y to manufacture Cheddar cheese. The P r t " var iants were more res i s tan t to bacteriophage and a n t i b i o t i c s than proteinase pos i t i v e (Pr t + ) s t r a ins as well as being less sens i t i ve to higher temperatures which retard normal cu l ture growth. Normal Cheddar f l avour developed when S_. cremori s numbers were cont ro l l ed (lower) without markedly re tard ing acid production (Lowrie et^  a l . , 1974). These authors suggested a c id i t y development by the s ta r t e r pa r t l y determined the environment in which the maturation process occurred as well as c o n t r o l l i n g the growth of non-starter bac te r ia and hence any o f f- f l avours that they might produce. Adda et _al_. (1982) showed that the pH of the cheese cont ro l l ed the r ipening process and i t together with the moisture content af fected texture . - 8 -In sp i te of the many advantages claimed for bacter iophage-insens i t ive mutants and pH cont ro l l ed cu l ture systems, bacteriophage i n s e n s i t i v i t y i s not a permanent property of these s t ra ins (Czulak et aj_., 1979; Lawrence et a l . , 1978). The s e n s i t i v i t i e s of the s t ra ins may change when subcultured over a long period of time (Limsowtin et a]_., 1980). A c r i t i c a l aspect therefore of th i s s t a r t e r system involves da i l y monitoring of the whey for bacteriophage against each of the s t ra ins and those showing phage s e n s i t i v i t y are replaced (Keogh, 1972; H u l l , 1977). App l i ca t ion and commercial izat ion of bacter iophage-insens i t ive mutant s t ra ins grown in pH cont ro l l ed s t a r t e r systems have been used success fu l l y in cheesemaking (Czulak et al_. , 1979; Limsowtin and Terzagh i , 1976; Lawrence et a l_ . , 1978; Thunell et , 1981; Danie l l and Sandine, 1981) and advantages to the industry has been reported (Richardson et al_. , 1983). Consistent s t a r t e r performance and economic savings due to improved qua l i t y have been reported (Thunell et al_. , 1981). Moreover, i t i s poss ib le to put together compatible mixtures of s t ra ins that w i l l resu l t in even better qua l i t y cheese (Lawrence et a l_ . , 1978) with reduced s t a r t e r v a r i a b i l i t y and more r e l i a b l e cheesemaking (Daniel l and Sandine, 1981). COMPOSITION It is general ly recognized that t r ad i t i ona l procedures for assessing cheese qua l i t y at an ear ly age are not a r e l i a b l e guide to i t s a c cep tab i l i t y by consumers when the cheese i s mature (McBride and H a l l , 1979). Lawrence and G i l l e s (1980), in an exce l lent review, showed that the qua l i t y of cheese was inf luenced most by three compositional f a c t o r s : MNFS, S/M leve ls and pH. The pH is inf luenced by the S/M and depends on the s t a r t e r cu l tures used, the manufacturing condit ions and the buf fer ing - 9 -capacity of the curd (Pearce and G i l l e s , 1979; Lawrence and G i l l e s , 1982). Sa l t content is an important determinant of Cheddar cheese q u a l i t y . High concentrat ions have a d i rec t e f f ec t on cu l ture growth, lac tose fermentation and subsequent acid production as well as p ro teo l ys i s (Thomas and Pearce, 1981). Cheese composition has been shown to inf luence y i e l d , grade and f i nanc i a l return when deviat ions from optimum s p e c i f i c a t i o n l im i t s (MNFS, 52-56%; fat-in-dry-matter (FDM), 52-55%; S/M, 4.0-6.0%; and pH 4.95-5.10) resu l t during manufacture ( L e l i e v r e , 1983; Le l i ev re and G i l l e s , 1982). TEXTURE Most studies invo lv ing cheese s t a r t e r cu l tures have deal t with s e l e c t i o n , i s o l a t i o n , rate of acid production of the s t ra ins and f lavour development in the cheese, while the qua l i t y of the cheese has been assessed by c l a s s i c a l subject ive eva lua t ion . It i s important to note, however, that the majority of cheeses are i d e n t i f i e d according to tex ture . The l a t t e r has been recognized as a multidimensional qua l i t y a t t r ibu te important for consumer preferences. In sp i te of the importance of cheese texture i t has not on the whole been very extens ive ly s tud ied . Since the development of ana l y t i ca l equipment al lowing texture p r o f i l e ana lys i s (TPA), (Szczesniak, 1975) a bet ter and more complete understanding of the textura l propert ies of the food i s poss ib le (Bourne, 1978). Important subject ive textura l evaluat ions of cheese inc lud ing f i rmness, sp r ing iness , smoothness and softness have been considered and measured ob j ec t i v e l y . Moreover, these resu l ts have been used to determine the iden t i t y and qua l i t y of the cheese (Baron and Scott B l a i r , 1953). - 10 -Prel iminary observations using ana ly t i ca l instrumentation indicated that normal curds for several types of cheese recovered e l a s t i c a l l y a very high proport ion of t he i r o r ig ina l compression deformation upon unloading. The e l a s t i c deformation was re la ted to the shear modulus and i n e l a s t i c deformation was reported to be re lated to the v i s cos i t y of the cheese curd . A value inverse ly proport ional to the shear modulus was suggested as the best s ing le c r i t e r i o n of firmness (Baron and Scott B l a i r , 1953). The texture of a large range of cheese samples and va r i e t i e s has been tested ob jec t i ve l y using a var iety of instruments. These ana ly t i ca l measurements provide object ive values of e l a s t i c deformation, p l a s t i c deformation, e l a s t i c i t y , forces required to compress or penetrate the cheese and energy required to rupture the ma te r i a l . Attempts to co r re l a te these object ive parameters with subject ive evaluat ions have had various degrees of success. Shama and Sherman (1973) reported that the Instron Universal Test ing Machine could be meaningfully employed to evaluate textura l propert ies of foods. However, i t is necessary to se lect the correct instrumental test condit ions in order to cor re la te with sensory responses. Many var iab les have been shown to inf luence the force compression behaviour of cheese ( C u l i o l i and Sherman, 1976). These include sample shape, dimensions, the loca t ion on the larger block of cheese, from which the sample was taken, rate that the compression force is app l i ed , the cohesiveness of the sample, age, the nature of the in te r face between the samples edge and the Instron plates as well as tes t temperature. Lee-et a l . (1978) determined the important sensory textura l cha r a c t e r i s t i c s of cheese and re la ted them to object ive measurements obtained with an Instron Universal Test ing Machine and found good co r re l a t i on with sensory panel - 11 -eva lua t ions . Sensory preference indicated that firmness was the important c h a r a c t e r i s t i c of the cheese. Carter and Sherman (1978) evaluated the firmness of Le i ces te r cheese and concluded that the force compression curves are inf luenced by the same fac tors prev iously found by C u l i o l i and Sherman (1976). Chen et al_. (1979) measured s ix textura l cha r a c t e r i s t i c s with an Instron Universal Test ing Machine for eleven va r i e t i e s of cheese and the object ive measurements cor re la ted with sensory panel r e s u l t s . Using s tep -wise mul t ip le l i nea r regression ana l y s i s , cheese texture was found to be inf luenced by the fo l lowing composit ional sequence: prote in > NaCl > water > pH > f a t . The e f f ec t of compression r a t i o on the mechanical propert ies of cheeses of various textura l cha r a c t e r i s t i c s was studied (Imoto et a l . , 1979; Lee et al_. , 1978) and co r re l a t ions between sensory evaluat ions of cheese texture and mechanical propert ies determined. The pattern of changes in mechanical propert ies in r e l a t i on to compression ra t io was unique for each type of cheese t e s t ed . Dickinson and Goulding (1980) studied the y i e l d behaviour of three cheeses (Cheddar, Cheshire and Le ices te r ) se lected for t he i r "crumbly" t ex -ture as a funct ion of deformation ra te , temperature and pre-y ie ld compres-sion h is tory and showed that simple non-linear v i s c o e l a s t i c models were not s a t i s f a c to r y to pred ic t y i e l d behaviour. Examination of the i n t e r r e l a t i onsh ips between composition and micro-st ructure in f u l l - f a t and reduced-fat cheeses have been studied (Emmons et a l . , 1980) and re lated to tex ture . E lectron microscopy and compositional ana lys is revealed about 30% more prote in matrix in the reduced-fat cheese which was responsible for the f i rmer and more e l a s t i c tex ture . - 12 -Green et aj_. (1981) studied the s t ructure and texture of Cheddar cheese made with milk concentrated to d i f f e r en t extents and found a d i r ec t r e l a t ionsh ip between concentrat ion fac tor and the Theological behaviour and s t ruc tu re . They found that prote in hydro lys is decreased with an increase in concentrat ion. Scanning e lect ron micrographs confirmed that the prote in const i tu ted a network in which the fat was entangled. Any modi f i ca t ion of the nature or the amount of the prote in present in the cheese would modify i t s t ex tu re . These resu l ts were in agreement with those of Emmons et a l . (1980). - 13 -MATERIALS AND METHODS 1. CHEESE MANUFACTURE Cheddar cheese was manufactured (1339 kg/h) with commercial sca le equipment using 8500 l i t r e OST vats and Alpha Matic continuous cheddaring equipment (Alpha-Laval Cheddar Systems L im i ted , 10 Oxford Road, Y e o v i l , Somerset, Great B r i t a i n BA21 5HR) or purchased from re t a i l ou t l e t s . Fresh, unstandardized raw milk was processed using high temperature short time (HTST) pas teur iza t ion (72 °C ; 16 sec) and f i l l e d d i r e c t l y into the vats at 30°C or heat t reated (63 °C ; 16 sec) and then f i l l e d into the va ts . Microbial rennet from Mucor miehei (Novo Indust r ies , Weston, Ont . , Can.) was used for a l l cheese manufactured and d i l u t ed 1:20 with water before adding to the milk under slow but constant a g i t a t i o n . The label dec la ra t ion on the purchased cheese was used as a parameter fo r age, assuming mild to be at least 100 days old and correspondingly medium 180 days and aged greater than 250 days. 2. COMPOSITIONAL ANALYSIS A l l cheese samples were analyzed in dupl icate for f a t , moisture , to ta l so l i ds and sa l t according to o f f i c i a l AOAC methods (AOAC, 1980). Samples fo r ana lys is were taken from the ins ide area of the piece of cheese and shredded f i n e l y using a 'home-style' g ra te r . Butter fat was determined by the Roese-Gottl ieb method. Moisture was determined by weighing 2-3 g of the shredded cheese sample in a weighed f l a t bottom metal d i s h . The sample was dr ied in a vacuum oven held constant at 100°C (ca 4 h) . Af ter coo l ing the dishes in a des i c ca to r , samples were quick ly weighed. Loss in weight was expressed as moisture and % residue as to ta l s o l i d s . - 14 -Sa l t was determined by a mod i f i ca t ion of the AOAC method for rout ine ana lys is during cheese manufacture. About 2 g of the shredded prepared cheese sample was accurate ly weighed into a 300 mL erlenmeyer and add 100 mL b o i l i n g H2O. The f lask was allowed to stand, sw i r l ing occas iona l l y for 5-10 min while coo l ing to 50-55°C. A f te r adding 2 mL 1^004 i nd i c a to r the mixture was t i t r a t e d with 0.IN AgN03 unt i l an orange-brown co lour pers is ted 30 seconds. % NaCl = mL 0.1N AgN0 3 X 0.585/g sample For accurate determination of NaCl in the cheese sample a spectrophotometry method was used (Kupke and Sauer, 1970). To 0.2 mL of the water por t ion containing the s a l t , 4.0 mL f e r r i c perchlorate working so lut ion was added and absorption a t 366 nm compared to a standard curve made with NaCl . [0.8g Fe(C104)3 was d isso lved in 1.0 mL deionized H2O and made up to volume (100 mL) with 70% HCIO4. Then 20 mL of 10% HCIO4 was added to 75 mL of t h i s so lu t ion to prepare the working so lu t i on . ] 3. SENSORY EVALUATION Approximately 21d a f te r product ion, samples were o f f i c i a l l y graded for f lavour and physica l cha r a c t e r i s t i c s by Federal Graders as out l ined by Agr i cu l tu ra l Canada Guidel ines for the examination of Dairy Products. Retai l purchased cheese was o f f i c i a l l y graded as soon as poss ib le a f t e r purchase. Sensory a t t r ibutes for a l l cheese inc luded : f l a vour , texture , co lour , moisture and s a l t . 4. CULTURE PROPAGATION Six se lected s t ra ins of bacter iophage-insens i t ive Streptococcus cremoris were purchased from Northwest Culture Tech. Inc. (P.O. Box 1991, - 15 -C o r v a l i s , Oregon, 97339) and propagated in a pH cont ro l l ed whey-based s ta r te r system (B io lac Inc . , 750 West 200 North, P.O. Box 3490, Logan, Utah 84321) as shown in F igure 1.1. The s ta r t e r media was made by mixing f r esh l y separated whey with water to adjust the lactose to 4.0%. Stimulant powder (Northwest Culture Tech. Inc . , P.O. Box 1991, C o r v a l l i s , Oregon, 97339) was added at a rate of 4.5 kg per 450 l i t r e of d i lu ted whey ( t i t r a t a b l e ac id i t y 0.55) . The mixture was then pasteurized at 85°C fo r 45 min, cooled to 27°C and a sep t i c a l l y inocu la ted . The inoculated whey-base was cont inuously s t i r r ed during incubat ion and automatical ly cont ro l l ed at 27°C by c i r c u l a t i n g hot or co ld water as required through the j acke t . The pH was cont ro l l ed between 5.9 and 6.1 by the addit ion of 29.4% aqueous ammonia. A f te r lactose deplet ion (approximately 18 h) the cul tured media was cooled to 4°C and used at an inocu la t ion level of 0.8%. S ta r te r a c t i v i t y was measured before i nocu l a t i on . 5. STARTER ACTIVITY Eleven grams of a n t i b i o t i c f ree non-fat-dry milk powder (NDM) was reconst i tu ted in 100 mL of water. Ten mL was added to a ser ies of 16 x 150 mm test tubes and s t e r i l i z e d (121°C; 15 min). To one 10 mL s t e r i l e NDM tube was added 0.1 mL of the s ta r te r to be tested and 0.3 mL of the s t a r t e r to another 10 mL s t e r i l e NDM tube. The procedure was repeated for each s ta r t e r to be tested and one 10 mL s t e r i l e NDM tube was retained as a blank. Af ter mixing, the tubes were incubated at 30°C fo r 1.5 h af ter which pH readings recorded. The d i f fe rence in pH between the blank and the inoculated tubes served as a quant i ta t i ve reference and was an i nd i ca t i on of the potent ia l a c t i v i t y of the s t a r t e r (Table 1.1). The procedure was Figure 1.1 Schematic ou t l i n i ng the whey-based s t a r t e r system. - 17 -Table 1 .1. 1 and 3% ac t i ve cu l ture inoculated into reconst i tuted non-fat-dry-milk powder conta in ing bromocresol purple and showing e f fec t on pH af ter 1.5 and 3.0 h incubation at 30°C r e spec t i ve l y . TIME (h) INOCULUM 1% 3% pH of blank 0.0 6.43 6.43 pH of 1% tube 1.5 6.17 pH of 3% tube 3.0 5.80 d i f f e r e n c e 0.26 0.63 - 18 -repeated a f te r 3 h for the 3% i nocu l a t i on . Only cul tures with the same a c t i v i t y were used for cheese making. 6. BACTERIOPHAGE TESTING To 500 mL of H 2 0, 55 g of NDM (10% so l i d s ) and 0.75 g bromocresol purple (F isher S c i e n t i f i c Company, Chemical Manufacturing D i v i s i o n , Fa i r Lawn, NJ) were added. Ten mL of t h i s i nd i ca to r medium was then added to a ser ies of tes t tubes 16 x 150 mm and s t e r i l i z e d (121°C; 15 min). A sample of cheese whey was f i l t e r e d through a 0.45 urn M i l l i p o r e f i l t e r to remove bac te r ia l c e l l s . Ser ia l d i l u t i ons of the f i l t e r e d whey were added to the s t e r i l e tubes containing 0.2 mL of act ive s t a r t e r cu l t u r e . The tubes were incubated fo r 5 h at 30°C before reading r e s u l t s . No colour change, a purpl ish-blue hue, indicated no ac id development whereas a greenish-yellow hue, was a resu l t of un inh ib i ted growth of the s t a r t e r cu l t u r e . Phage t i t r e was indicated by the degree of colour in tens i t y (Figure 2.1) and served as a q u a l i t a t i v e measurement. 7. ASSAYING FOR FREE AMINO GROUPS P ro teo l ys i s in the t r i c h l o r o a c e t i c ac id (TCA) so lub le cheese f r ac t i on was measured by a modif ied micromethod with f luorescamine (Kwan et a l . , 1983). Four grams of cheese was centr i fuged (15,000 X g; 30 min; 25 °C ) to remove f a t . P r e c i p i t a t e , 0.3 g was added to 10 mL of TCA buffer conta in ing 0.075M TCA, 0.15M sodium acetate and 0.225M acet i c acid (Nakai et al_. , 1964) A f te r mixing thoroughly the mixture was centr i fuged (10,000 X g; 5 min; 2 5 ° C ) . To 0.4 mL of the supernatant, 10 mL of TCA buf fe r was added. To 0.1 mL of the d i l u t ed supernatant 0.3 mL of 3M K2HPO4 was added fol lowed gure 2.1 Method used to detect the presence of bacteriophage in whey with bromocresol purple as an i nd i c a to r . - 20 -by 0.15 mL of 0.03% (wT/vol) f luorescamine (Chemical Dynamics Co rp . , South P l a i n f i e l d , NJ) in acetone. The mixture was immediately mixed. To adjust the volume to the cuvette s i z e , 3.0 mL of d i s t i l l e d water was added. A standard curve was made with tryptophan. Fluorescence ( exc i t a t ion = 395 nm; emission = 480 nm) was measured with an Aminco Bowman 4-8202 spectrophotof1 urometer (American Instrument C o . , Inc . , S i l v e r Spr ing , MD) and resu l t s reported in ug N^/gm cheese. 8. PREPARATION OF CHEESE SAMPLES Cheddar cheese samples were prepared from the central regions of whole blocks (19 kg) to minimize the e f fec t of moisture gradients which have been shown to exh ib i t non-symmetrical deformation on compression (Carter and Sherman, 1978; C u l i o l i and Sherman, 1976). Samples (4 °C ) were cut into 2 cm cubes (Creamer and O lson , 1982) with a wire cheese cu t t e r . A l l samples were weighed to confirm uniformity of s ize and placed in p l a s t i c sealed cups to prevent moisture losses . The cubes were allowed to equ i l i b r a t e to room temperature (ca 22°C) f o r ca 6 h p r io r to compression t e s t i n g . 9. TEXTURAL EVALUATION Ten rep l i ca tes of each sample of Cheddar cheese were compressed at ambient temperature (22 °C) between f l a t pa ra l l e l metal plates of a la rger c ross-sect iona l area on an Instron Model 1122 Universal Tes t ing Machine. Force-compression data were recorded with the standard Instron recorder at a chart speed of 200 mm/min. A l l data reported were obtained using a constant crosshead speed of 50 mm/min. - 21 -10. SAMPLE PREPARATION FOR HPLC ANALYSIS Commercially produced samples of Cheddar cheese made with e i the r _S. cremoris or _S. l a c t i s s t ra ins and of the correct composition (Lawrence and G i l l e s , 1982) and approximately the same age were se lected for ana l y s i s . Ext rac t ion of the non-volat i le water-soluble f r a c t i on was ca r r i ed out according to the method of McGugan et al_. (1979) except a smaller amount of sample was used (Pham and Nakai , 1984). The complete ext rac t ion procedure i s described in Sect ion 1 of Mater ia ls and Methods in Part II of t h i s t h e s i s . 11. STATISTICAL ANALYSIS Computer analyses were performed on an Amdahl 470 V/8. Programs used were: BMD: 4M - P r inc ipa l Component Ana l y s i s ; BMD: 7M - Stepwise L inear Discr iminant Ana l y s i s , BMDP S t a t i s t i c a l Software Inc . , 1981 Westwood B l v d . , Suite 202, Univers i ty of C a l i f o r n i a . Descr ip t ion for ana lys is is out l ined in Sect ion 7 of Mater ia ls and Methods in Part II of t h i s t h e s i s . 12. HPLC ANALYSIS A Spectra-Physics 8100 HPLC and 8400 var iab le wavelength detector (Spectra-Physics, Santa C l a r a , C a l i f o r n i a ) operated at a wavelength of 220 nm were used for the ana lys is with modi f i ca t ion discussed in Part II - 4. A reversed-phase column (250 x 4.6 mm I.D.) packed with Adsorbosphere Cs (5 u.m) purchased from Appl ied Sciences Laborator ies (State Co l l ege , Pennsylvania) was used for chromatographic runs. The volume of the sample 1oop was 50 u L. A l l chromatographic runs were performed at ambient temperature at a flow rate of 0.97 mL/min. An optimized gradient e lu t ion system was used as - 22 -descr ibed in Sect ion 5 in Part II of t h i s t h e s i s . The i n i t i a l solvent composition was 44 .6 :0 .0 :55 .4 for t r i f l u r o a c e t i c ac id (0.1%), a c e t o n i t r i l e and water. Over 56.6 min the r a t i o was changed to 0 .0 :36 .6 :63 .4 . Solvents were prepared as described in Sect ion 4 in Part II of th i s t h e s i s . - 23 -RESULTS AND DISCUSSION RHEOLOGICAL PROPERTIES The force-compression curves obtained for m i l d , medium and aged Cheddar samples sub jec t i ve l y evaluated f i r s t grade and with e s s e n t i a l l y the same moisture, f a t , sa l t and pH are shown in F igure 3.1. In a l l cases the cheese deformed on compression and the force on the cheese increased unt i l a y i e l d point was reached. The i n i t i a l port ion of the curves were s l i g h t l y convex then became concave. This is in agreement with Creamer and Olson (1982) who showed that low pH cheese has a steep convex force-compression curve r e l a t i v e to the Y-axis and crumbles at the y i e l d point while a high pH cheese has a less steep, concave, force-compression curve and at the higher y i e l d point i t s p l i t s into large fragments. The rheo log ica l propert ies of cheese have been shown to be re la ted to the composit ion. Moreover, the degree of p ro teo lys i s determines the s t ructure which in turn is responsib le for the texture . Any a l te ra t ions in the composition therefore can af fect the f i n a l rheologica l p roper t i es . Cheese is e s s e n t i a l l y made up of a network of fat surrounded by prote in ( case in ) . Emmons et al_. (1980) showed that reduced-fat cheese was considerably f i rmer and more e l a s t i c than f u l l - f a t cheese even though the moisture leve ls in the nonfat matter were the same. The textural d i f fe rences found were re la ted to moisture content, degree of p ro teo lys i s and pH of the cheese. The pH, brought about through the production of l a c t i c acid by s t a r t e r c u l t u r e , a f fec ts the physico-chemical propert ies of the cheese. A p laus ib le explanation of the d i s t i n c t l y d i f f e r en t textures of high and low pH cheeses was proposed by Creamer and Olson (1982). They concluded that fa t s t a b i l i t y in the prote in/ fa t matrix of the cheese is not be affected by pH, however, - 24 -5000 F igure 3.1 Force-compression curves fo r a m i l d , medium and aged cheese (samples number 18, 19 and 17 r espec t i v e l y from Table 4 .1 ) . - 25 -the prote in f r a c t i on would be in f luenced . The major caseins have i s o e l e c t r i c points near 4 .5. At t h i s pH the casein components form compact aggregates which are held together by hydrophobic intra-aggregate forces while the inter-aggregate forces are weaker. Most of the water in such a system is J-nert and i n t e r s t i t i a l . At the higher pH, the casein molecules acquire a net negative charge r e su l t i ng in repuls ion and the t i gh t prote in aggregates absorb water. The dramatic d i f ferences obtained in the force-compression curves (Creamer and Olson, 1982) between experimental ly produced high and low pH cheese were not found in t h i s study. As shown in Table 2.1, d i f fe rences in pH in commercially produced Cheddar was not as extreme. This was assumed to be due to the fact that a l l cheeses evaluated were within an intermediate pH range of 4.95-5.10 considered to be a low pH cheese ( Le l i ev re and G i l l e s , 1982) with the exception of samples 13 and 36. Sample 36 was an extra-aged sample approximately 400 days o ld which did show a convex force-compression curve typ ica l of low pH cheese and agreed with others . Voisey (1977) explained that the i n i t i a l concave then convex nature of the force-compression curve was due to adhesion between the sample and the c e l l surface in the case of the former. A f te r compression the slope changes becoming convex and continues at a steady rate unt i l the y i e l d po in t . His study made no reference to age, composit ion or pH of the sample cheese. It i s reasonable from the resu l t s of the present study and in agreement with others ( Le l i ev re and G i l l e s , 1982) that Vo i sey ' s work was done on a low pH cheese and the pa r t i cu l a r force-compression curve i s by no means representat ive of a l l Cheddar cheese. The y i e l d point for cheese samples occurred at greater force and greater compression for mild and medium cheeses (Figure 3 .1) . Rheological - 26 -CHEESE FAT MOISTURE pH SALT MNFS1 FDM2 S/M3 (sample no.) (%) (%) (%) 1 33.3 36.8 4.98 1.85 55.17 52.69 5.02 2 33.5 36.7 5.00 1.88 55.19 52.92 5.12 3 33.7 36.1 5.04 1.73 54.45 52.74 4.70 4 31.0 38.6 4.99 1.90 55.94 50.49 4.92 5 33.8 36.2 5.00 1.70 54.68 52.98 4.70 6 32.6 37.3 4.98 1.75 55.34 51.99 4.69 7 31.2 38.5 5.10 1.80 55.96 50.73 4.68 8 32.3 37.8 5.02 1.75 55.83 51.93 4.63 9 31.6 38.7 4.97 1.88 56.58 51.55 4.86 10 31.5 38.0 5.00 1.95 55.47 50.81 5.13 11 32.1 37.6 4.97 1.89 55.38 51.49 5.02 12 34.9 36.0 5.00 1.73 55.30 54.53 4.80 13 32.7 37.8 4.90 1.90 56.17 52.57 5.03 14 33.6 36.5 5.10 1.60 54.97 52.91 4.38 15 34.9 36.0 5.15 1.70 55.30 54.53 4.72 16 35.0 36.0 5.08 1.72 55.38 54.69 4.78 17 35.1 35.3 4.95 1.70 54.39 54.25 4.82 18 34.8 36.0 5.08 1.80 55.21 54.38 5.00 19 33.2 37.0 5.05 1.85 55.39 52.70 5.00 20 34.8 36.1 5.01 1.70 55.06 54.46 4.71 21 34.5 35.9 5.05 1.65 54.80 53.82 4.59 22 33.6 36.5 5.05 1.75 54.97 52.91 4.79 23 34.5 36.0 5.04 1.70 54.96 53.90 4.72 24 33.9 36.9 5.07 1.81 55.82 53.72 4.90 25 34.5 36.0 5.07 1.72 54.96 53.90 4.78 26 32.2 36.4 5.05 1.77 53.68 50.63 4.86 27 32.5 36.2 5.10 1.75 53.63 50.94 4.83 28 33.5 37.0 5.06 1.80 55.64 53.17 4.86 29 34.6 36.0 5.07 1.73 55.04 54.06 4.80 30 34.5 36.1 5.08 1.78 55.11 53.99 4.93 31 34.5 35.5 4.98 1.77 54.19 53.49 4.89 32 33.5 36.0 5.00 1.79 54.14 52.34 4.97 33 34.0 35.9 4.98 1.70 54.39 53.04 4.74 34 32.5 36.2 4.96 1.75 53.63 50.94 4.83 35 33.0 36.3 5.06 1.75 54.18 51.80 4.82 36 34.3 35.5 4.90 1.80 54.03 53.18 5.07 iMNFS - moisture in the non-fat substance; 2FDM - fat-in-dry-matter ; 3S/M - sa l t- in-mois ture . Table 2.1. Compositional ana lys is of the commercially produced cheeses. - 27 -CHEESE AGE YIELD POINT DISTANCE % COMPRESSION (sample no.) (days) (g) (nm) AT YIELD POINT 1 199 3139.17 24.46 30.55 2 199 3427.00 24.65 30.80 3 116 4763.30 33.44 41.80 4 174 2997.50 30.92 38.65 5 160 3874.00 25.90 32.40 6 139 3546.00 29.40 36.75 7 185 2863.00 30.30 37.90 8 276 2234.00 20.75 25.95 9 227 2906.00 38.64 48.30 10 212 2507.00 29.60 37.00 11 221 2797.00 27.60 34.50 12 136 3944.00 29.70 37.15 13 300 3212.00 27.65 34.60 14 115 4361.00 30.85 38.55 15 36 4680.00 29.80 37.25 16 100 3425.00 25.70 32.15 17 424 2752.00 17.30 21.65 18 96 4292.00 30.50 38.15 19 150 3409.00 32.05 40.05 20 180 3650.00 32.00 40.00 21 180 3650.00 32.00 40.00 22 180 3550.00 33.00 41.25 23 180 3450.00 32.00 40.00 24 100 4400.00 31.00 38.75 25 100 4500.00 31.50 39.40 26 100 4480.00 81.00 37.50 27 100 4580.00 31.00 38.75 28 100 4550.00 29.00 36.25 29 100 5300.00 28.00 35.00 30 300 3200.00 28.00 35.00 31 300 3053.00 28.00 35.00 32 300 3390.00 27.50 34.40 33 300 3250.00 26.50 33.15 34 300 3290.00 28.00 35.00 35 300 3350.00 28.50 35.65 36 424 2752.00 17.30 21.65 DEFORMATION (mm) FIRMNESS (g/mm) 6.11 514.15 6.16 556.21 8.36 570.03 7.73 389.42 6.48 598.63 7.35 483.47 7.58 378.24 5.19 430.97 9.66 302.50 7.40 339.93 6.90 405.89 4.73 531.89 6.92 465.54 7.71 567.44 7.45 631.22 6.43 535.32 4.33 639.00 7.63 563.18 8.01 426.55 8.00 456.25 8.00 456.25 8.25 430.30 8.00 431.25 7.75 567.74 7.88 571.06 7.50 597.33 7.75 590.00 7.25 627.59 7.00 757.14 7.00 457.14 7.00 435.72 6.88 492.73 6.63 490.19 7.00 470.00 7.13 469.85 4.33 639.00 Table 3.1. Rheological data for the corrniercially produced cheeses of different ages, composition and culture type. - 28 -data (Table 3.1) are cons is tent with the fact that younger cheese is more res i s ten t to deformation. This is discussed l a te r in ' E f f ec t s of P r o t e o l y s i s ' . Deformation curves of the cheese samples were concave during compression due to the nature of the contact surface between the cheese and the plates of the Instron. This was assumed to be due to the o i l (but ter fa t ) present in the cheese which lubr ica ted the metal plates of the Instron thus preventing a t yp i ca l barrel shape during compression found by others ( C u l i o l i and Sherman, 1976). EFFECT OF MOISTURE CONTENT Table 2.1 shows the composit ional ana lys is of the 36 commercially produced cheeses which were a l l sub jec t i ve ly evaluated by the Federal Graders as f i r s t grade. It can be seen that moisture in the non-fat-substance, fat- in-dry-matter , sa l t- in-moisture and pH were within optimum spec i f i c a t i ons for a qua l i t y cheese ( Le l i e v re , 1983; Le l i ev re and G i l l e s , 1982). Cheeses varying in moisture but of approximately the same composit ion, age and degree of p ro teo l ys i s (Figure 4.1) show force-compression curves that are a f fec ted by moisture content a lone. As the moisture content increases , the firmness of the cheese as indicated by the slope decreases. Figure 5.1 shows the y i e l d force vs moisture content for cheese samples 1, 2, 4-14, 16 that were sub jec t i ve ly evaluated (Table 4.1) as being weak or s l i g h t l y weak tex ture . The c o e f f i c i e n t of determination for the regression l i n e was 0.788. With the exception of samples number 12, 13 and 16 a l l others were evaluated as mild cheese regardless of age. It seems reasonable to assume, the re fo re , that moisture content is an important - 29 -5000 Figure 4.1 Force-compression curves fo r cheese of approximately the same age, composit ion and degree of p ro t eo l y s i s but varying in moisture content. Samples number 12, 6 and 7 with moisture of 36 .0 , 37.3 and 38.5% r e spec t i v e l y . See Tables 2 .1 , 3.1, 4 .1 . 4361.0 |12 • 14 3829.2 E 0) •6 • • 16 •^ ^^  - 32975 o u_ • 1 • 13 • 4 • 11 •10 .7 » ' 2765.7 2234.0 i i i •8 ii \.U • 1 — — . 1 1 • » 36.0 36.4 36.9 37.3 37.8 38.2 38.7 Mois ture , % Figure 5.1 Var ia t ion in the force at the y i e l d point fo r cheese samples number 1, 2, 4-14, 16 sub jec t i ve l y evaluated weak o f s l i g h t l y weak texture . See Table 2.1 for composition ( r 2 = 0.788). T a b l e 4 . 1 . Rheo log ica l , compositional and sensory eva luat ion o f the commercially produced cheese samples o f various ages using cremoris or l a c t i s cu l ture s t ra ins on both conventional and automated cheese processing equipment. A B C D E F e H I J *1 *2 NO. YIELD POINT DISTANCE X COMPRESSION DEFORMATION FIRMNESS MOISTURE -NH 2 CULTURE* AGE PROCESS** CATEGORY": TEXTURE^ (9) (mm) . AT (mm) ( g /mm) (*) (days) YIELD POINT 1 3139.17 24.46 30.55 6.11 514.15 36.8 46.0 1 199 1 1 2 2 3427.00 24.65 30.84 6.16 556.21 36.7 48.1 1 199 1 1 2 3 4763.30 33.44 41.80 8.36 570.03 36.1 38.0 2 116 1 1 1 4 2997.5 30.92 38.65 7.73 389.42 38.6 39.1 1 174 1 1 3 5 3874.00 25.90 32.40 6.48 598.63 36.2 41.0 2 160 1 1 6 3546.00 29.40 36.75 7.35 483.47 37.3 35.0 2 139 1 1 2 7 2863.00 30.30 37.90 7.58 378.24 38.5 40.0 1 185 1 1 3 8 2234.00 20.75 25.95 5.19 430.97 37.8 54.1 1 276 1 1 3 9 2906.00 38.64 48.30 9.66 302.50 38.7 47.0 1 227 1 1 3 10 2507.00 29.60 37.00 7.40 339.93 38.0 48.5 1 212 1 1 3 11 2797.00 27.60 34.50 6.90 405.89 37.6 50.3 1 221 1 1 3 12 3944.00 29.70 37.15 7.43 531.89 36.0 59.7 2 136 1 2 2 13 3212.00 27.65 34.60 6.92 465.54 37.8 61.0 1 300 1 2 3 14 4361.00 30.85 38.55 7.71 567.44 36.5 42.6 2 115 1 1 2 15 4680.00 29.80 37.25 7.45 631.22 36.6 35.6 2 36 1 I 1 16 3425.00 25.70 32.15 6.43 535.32 36.0 51.6 2 100 1 2 2 17 2752.00 17.30 21.65 4.33 639.00 35.3 94.2 2 424 2 1 18 4292.00 30.50 38.15 7.63 563.18 36.0 38.5 2 96 2 1 1 19 3409.00 32.05 40.05 8.01 426.55 37.0 60.1 2 150 2 2 20 3650.00 32.00 40.00 8.00 456.25 36.1 59.0 2 180 2 1 21 3650.00 32.00 40.00 8.00 456.25 35.9 57.8 2 180 2 2 1 22 3550.00 33.00 41.25 8.25 430.30 36.5 61.3 2 180 2 2 1 23 3450.00 32.00 40.00 8.00 431.25 36.0 58.6 2 180 2 2 1 24 4400.00 31.00 38.75 7.75 567.74 36.9 34.0 2 100 2 1 25 4500.00 31.50 39.40 7.88 571.06 36.0 35.0 2 100 2 1 1 26 4480.00 31.00 37.50 7.50 597.33 36.4 36.1 2 100 2 1 27 4580.00 31.00 38.75 7.75 590.00 36.2 35.8 2 100 2 1 | 28 4550.00 29.00 36.25 7.25 627.59 37.0 37.2 2 100 2 1 1 29 5300.00 28.00 35.00 7.00 757.14 36.0 38.0 2 100 2 J 30 3200.00 28.00 35.00 7.00 457.14 36.1 57.8 2 300 2 } 31 3053.00 28.00 35.00 7.00 435.72 35.5 59.0 2 300 2 | 32 3390.00 27.50 34.40 6.88 492.73 36.0 60.9 2 300 2 33 3250.00 26.50 33.15 6.63 490.19 35.9 59.7 2 300 2 3 34 3290.00 28.00 35.00 7.00 470.00 36.2 61.0 2 300 2 3 35 3350.00 28.50 35.65 7.13 469.85 36.3 58.5 2 300 2 3 • 36 2752.00 17.30 21.65 4.33 639.00 35.5 94.2 2 424 2 a 1 - S. cremoris; 2 - S. 1 act i s . b i . automated; 2 - conventional. c 1 - m i l d ; 2 - medium; 3 - aged; 4 - X-aged. d 1 - f i r m ; 2 - s l i g h t l y weak; 3 - weak. - 32 -f ac to r a f fec t ing texture and is not re lated to cu l ture type. Indeed, re l a t ionsh ips between various cheese-making parameters and grade estab l i shed c l e a r l y that MNFS and S/M were the s ing le most important parameters a f f e c t i ng the grade score of the cheese (Pearce and G i l l e s , 1979; Lawrence and G i l l e s , 1980). This information i s in agreement with the present study. The e f f ec t of moisture was found to be more c lose l y re la ted to the amount of moisture per unit of casein than with the absolute percentage of moisture in the cheese (Lawrence and G i l l e s , 1980) and i t i s in the mixture of moisture and casein that enzymatic react ions responsib le for r ipening la rge ly take p lace . Small increases in the MNFS leads to the a v a i l a b i l i t y of free moisture as seen in Table 2.1. This free moisture increases the a c t i v i t y of both micro-organisms and enzymes and yet has a detrimental e f f e c t on cheese tex ture . Cheeses with the lower S/M tended to mature quicker because of the reduced i nh ib i to r y e f f ec t of the sa l t on both s t a r t e r cu l ture bacter ia and normal milk f l o r a . During aging, however, the low S/M cheese deter iorated in qua l i t y and was downgraded due to the production of undesirable f lavour components. While small proport ions of NaCl have st imulatory e f f e c t s on l a c t i c cu l t u r e s , l a rger proport ions exh ib i t tox ic propert ies (Meister and Ledford , 1979). Walter et al_. (1958) reported that s t ra ins of S^ . l a c t i s were not s i g n i f i c a n t l y i nh ib i t ed in Cheddar cheese curd by 1.6 to 2.0% NaCl , whereas most S^ . cremoris s t ra ins were i nh ib i t ed s l i g h t l y at 1.4%, d e f i n i t e l y at 1.6% and almost completely at 2.0%. The t ra ined cheese graders had no d i f f i c u l t y in d i s t i ngu i sh ing cheeses of d i f f e r en t moisture contents. This i s reasonable s ince much of the water in Cheddar cheese i s "bound" s p e c i f i c a l l y to the case ins , t h e i r degradation - 33 -products and to the calcium lac ta te and sodium ch lor ide present . Thus, a small increase in MNFS leads to a r e l a t i v e l y large increase in f r ee l y ava i l ab le moisture which has been shown to be eas i l y detected by the t ra ined graders. Increases in sa l t content were a lso eas i l y detected because of the coarseness and other detrimental e f fec ts i t has on tex ture . Although microstructure examination was not the scope of th i s study, Emmons et al_. (1980) showed that increased moisture content decreased firmness and e l a s t i c i t y and that a r e l a t ionsh ip between mic ros t ruc ture , composition and texture ex i s t ed . Increased moisture, s p e c i f i c a l l y MNFS was responsible for a reduced firmness in the cheese texture confirming the resu l ts of the present study. Lawrence and G i l l e s (1980) found with some modern cheesemaking systems, that both the curd p a r t i c l e s ize and the curd matrix are not uniform. This var iab le s t ructure resu l ts in curd of d i f f e r i n g moisture content s ince a highly structured curd w i l l reta in s i g n i f i c a n t l y more moisture than a non-cheddared curd. Although increased moisture was d i r e c t l y re la ted to a decrease in y i e l d force (Figure 4 .1 ) , the resu l ts were not dramatic, while a de f i n i t e tendency was shown. This was due to the fact that the v a r i a b i l i t y in composition did not exceed the ranges for a qua l i t y cheese (Le l ievre and G i l l e s , 1982). EFFECT OF PROTEOLYSIS It has been considered that p ro teo lys i s occurr ing during r ipening could be due to enzymes from two major sources: namely rennet and the s t a r t e r cu l t u r e . Although most of the rennet is removed with the whey, a small amount does remain with the curd (Holmes et aj_., 1977). The act ion of th i s res idual c l o t t i n g enzyme combined with that of enzymes from s ta r t e r - 34 -bacter ia and normal milk f l o r a enzymes la rge ly determine the overa l l p ro t eo l y t i c pattern in r ipening cheese. The p r inc ipa l pathway of p ro t eo l y t i c breakdown during the r ipening process suggested by Lowrie and Lawrence (1972) i s one in which rennet causes a primary, l im i ted degradation of caseins to high molecular weight non-bit ter peptides that are fur ther hydrolyzed by enzymes from the bacter ia in the s t a r t e r cu l ture to smaller peptide fragments and free amino a c ids . The importance of rennet for cheese f lavour development and subsequent texture has been extens ive ly reviewed (Fox, 1981). Throughout th i s study, in order to minimize va r i a t ion in cheese f lavour and/or t ex tu re , e x t r a c e l l u l a r proteinases from Mucor mehei were used as the rennet source. The success of using th i s enzyme in commercially produced cheeses has been well documented. (V i s se r , 1981). As shown in Table 4 .1 , the degree of p r o t e o l y s i s , expressed as f ree amino groups, and moisture content have an e f f ec t on the y i e l d point of Cheddar cheese. As the cheese ages, p ro teo lys i s increases (Figure 6.1) and y i e l d point decreases. E lec t rophore t i c studies (Creamer and Olson, 1982) have re la ted the content of intact d s i - c a s e i n to y i e l d f o r ce , while de Jong (1975) developed a quant i ta t i ve e l ec t rophore t i c method to study p r o t e o l y s i s . They found that as the cheese ages, a s i - c a s e i n is degraded to a s i - I-case in while g-casein remains e s s e n t i a l l y i n t a c t . Although e lec t rophores is is a valuable ana l y t i ca l technique, f luorescamine which reacts s p e c i f i c a l l y with free amino groups at the appropriate pH serves as a more sens i t i ve assay method for p ro teo l ys i s of prote in in cheese by proteinases from the s t a r t e r cu l t u r e . The procedure which i s h ighly sens i t i ve and re la ted to the number of peptide bonds hydrolyzed c l e a r l y suggests that a l l other parameters Figure 6.1 Va r i a t ion in the f ree amino groups for cheeses o f d i f f e r e n t ages commercially produced using e i t he r S. cremoris or S^ . l a c t i s cu l tu re s t r a i n s . See Table 4 . 1 . ( r 2 = 0.862). - 36 -being equa l , cheeses manufactured using j>. cremoris s t ra ins have a l esser degree of p ro teo lys i s and subsequent f lavour development for approximately the same age than those produced with _S. l a c t i s cu l tures although i n i t i a l ac id production was the same (Figure 6 .1 ) . A poss ib le explanation i s that the buffered pH con t ro l l ed s t a r t e r media with added nitrogenous st imulants allowed se l ec t i ve growth of P r t " c e l l s (Richardson et a l_ . , 1983) to at l eas t 93% of t he i r P r t + counterpar ts . Pr t " var iants are de f i c i en t in c e l l wall-bound proteinase and have been found to reach only 10-25% of the maximum ce l l density atta ined by P r t + organisms (Thomas and M i l l s , 1981). Ce l l wall-bound proteinases allow the s t a r t e r c e l l s to hydrolyze the milk proteins to peptides which are then small enough to d i f fuse into the c e l l wall and fur ther degraded by the i n t r a c e l l u l a r peptidases to free amino a c i d s . These free amino acids are essent ia l for the growth of the s t a r t e r bac t e r i a . Therefore , the greater the c e l l density the higher is the concentrat ion of proteinase which increases exponent ia l ly with the growth of the cu l t u r e . Indeed i t i s f eas ib l e that the cremoris s t ra ins used in t h i s study did not reach maximum ce l l dens i t i es in the milk ( in sp i te of ac id production) compared to t h e i r S^ . l a c t i s counterparts thus providing a poss ib le explanation for the l imi ted p ro teo l y t i c a c t i v i t y due to lower amounts of c e l l wall-bound prote inases . Cheese made with S. cremoris and of approximately the same age as that made with 1act is s t a r t e r cu l ture was sub jec t i ve l y graded mild while the l a t t e r was graded medium. General ly cheeses made with j>. l a c t i s ripened to medium in approximately 6 months, whereas those using S. cremoris s t ra ins required an add i t iona l 3 months to reach the same degree of maturity a f te r which f lavour d i f fe rences were minimal. In case no 8 (Tables 3.1 and 4.1) the cheese was 276 days old - 37 -yet s t i l l c l a s s i f i e d as mild and case no 13, a 300 day old cheese was c l a s s i f i e d as medium. The pH of cheese is inf luenced by the bacter ia l cu l ture used. Moreover, Adda et ^1_. (1982) showed that the pH of the cheese con t ro l l ed the r ipening process and together with the water content af fected texture . As the pH drops during lactose fermentat ion, the ac id i t y of the curd leads to s o l u b i l i z a t i o n of the phosphates and calcium l inked to the prote in mice l les while at the same time lowering the a c t i v i t y of l i p o l y t i c and p ro teo l y t i c enzymes. The l a c t i c acid produced during r ipening serves as a substrate fo r natural milk f l o r a , not destroyed by cheese cook temperatures ( 3 7 . 8 ° C ) , to ra ise the pH to optimal leve ls for enzyme a c t i v i t y . Assuming the i n t e r r e l a t i onsh ip between pH and p ro teo l y t i c a c t i v i t y the i r e f fec t on cheese texture c l e a r l y showed that cheese made with cremoris had force y i e l d values t yp i ca l of mild cheese regardless of age (Table 4.1) and required addi t ional r ipening time compared to cheese made with _S. l a c t i s before rheograms were typ ica l of medium or aged cheese. (Figure 4 . 1 , sample no. 7 ) . Rate of i n i t i a l acid development in sp i te of the lower inoculum of 0.8% was the same for both l a c t i s and S_. cremoris cu l tures due to the large number of organisms inocu la ted , which is in agreement with others (Thunell et al_. , 1981; Lawrence et a]_., 1978). EFFECT OF HEAT TREATMENT Force-compression curves for cheese of approximately the same composi t ion, pH and age were af fected by the i n i t i a l heat treatment of the mi lk . Heat-treated milk (63°C/16 sec) had a higher y i e l d force (Figure 7.1) compared to pasteurized milk (72°C/16 sec ) . In cheese made from - 38 -5 0 0 0 100 C o m p r e s s i o n , % F igure 7.1 E f fec t of heat treatment for cheese milk on force-compression curves for commercially produced cheese of approximately the same age, composition and degree of p r o t e o l y s i s , a, 63 °C/16s; b, 72°C/16s; (Samples no. 3 and 5 r espec t i ve l y Table 4 .1) . - 39 -pasteur ized milk the serum pro te ins , p a r t i c u l a r l y /3-lactoglobul in , react through SH bonding with the casein thus destroying the capacity of the case in to form a f i rm curd r e su l t i ng in a weaker texture cheese (Figure 7.1) . In addi t ion to the i r r e v e r s i b l e prote in denaturat ion, the elevated temperatures have been impl icated in a l t e r i n g hea t- lab i l e enzyme substrates (V i sser , 1981). Thus the balance between act ive enzymes, substrates and r e su l t i ng p ro t eo l y t i c breakdown products are a l t e red or i n h i b i t e d . This may expla in why heat t reated milk has been t r a d i t i o n a l l y considered to produce more f l avour fu l cheese. D i f ferences in the force-compression curves between heat-treated (63°C/16 sec) and pasteur ized milk (72°C/16 sec) cheeses were obtained (Figure 7.1) . It was assumed that cheese texture and enzyme a c t i v i t y were due to the composition of the cheese and the cu l tu re type used respect i ve ly and not a f fec ted by the heat treatment. Indeed both S_. cremoris and S_. 1ac t i s were used fo r making pasteur ized milk cheese ye t t he i r p ro t eo l y t i c a c t i v i t i e s were d i f f e r en t (Figure 6 .1 ) . STEPWISE LINEAR DISCRIMINANT ANALYSIS (AGE) Data from ind iv idua l cheese ana lys is appear in Table 4 .1 , rows A through J . However, c l a s s i f i c a t i o n of the cheese samples into d i s t i n c t groups of m i ld , medium, o ld and X-aged as well as d i s t i n c t categor ies for texture (weak, s l i g h t l y weak and firm) was assessed by Grader sensory eva lua t ion . The data from the 36 commercially produced cheese samples (Table 4.1) were analyzed by stepwise l i n ea r d iscr iminant ana lys is (SLDA). The technique i s based on the concept of c l a s s i f i c a t i o n of objects on the basis of the patterns formed by a set of observations or r e s u l t s . In the - 40 -ana lys i s each cheese sample was considered as a point in a p-dimensional space (p being the number of var iab les appearing in Table 4 .1 , A through J ) . Each cheese sample analyzed, the re fo re , has a set of resu l ts which const i tu tes a pat te rn . Since groups of s im i l a r samples w i l l have s im i l a r pa t te rns , th i s information can be used to ident i f y groups of s im i l a r samples or i t can be used to character ize a group of s im i l a r samples in order to enable one to c l a s s i f y the group to which a new sample belongs. In the regression ana l y s i s , assessed by Grader evaluat ion was used for c l a s s i f i c a t i o n of the cheeses into age ca tegor ies . A l l other data were considered independent va r i ab l es . The program se lec ts the most s i g n i f i c a n t independent var iab le f i r s t followed by the second most s i g n i f i c a n t e t c . , in a stepwise f ash ion . The F-value of each var iab le is computed and the var iab le with the highest F-value is the one entered at the f i r s t s tep. This var iab le is then paired with each of the other ava i l ab le v a r i ab l e s , one at a time and the one giv ing the highest increase in d i sc r imina t ion is the next var iab le chosen. This is then entered in the next step along with the f i r s t var iable and the same procedure is continued unt i l the F-probabi 1 i t y of de le t ion becomes too high (Coomans et aj_., 1979). The s i g n i f i c a n t d i f f e rence among the four age groups was observed as shown in Table 5.1. By reduction of the o r ig ina l number of independent var iab les from 11 to 4, 100% of the sample cases were co r rec t l y i d e n t i f i e d (Table 6.1) . Because of the d i f f i c u l t y in v i s u a l i z i n g the mu l t i va r i a te data obtained in a p-dimensional space, the intent of SLDA is to reduce p to 2 so that data obtained can be represented in a two dimensional plot or space (X- and Y- a x i s ) . For th i s purpose two main canonical var iab les were ca l cu l a t ed . The f i r s t canonical var iab le which i s a l i nea r combination of - 41 -Table 5 . 1 . F-matrix among the four age groups (mi ld , medium, o l d , X-aged) fo r the cheese samples, a f te r d iscr iminant ana l y s i s . Age Age Mi ld Medium Old Medium 76.14** Old 67.54** 0.10 X-Aged 220.33** 86.62** 77.39** **P < 0 .01, F (4, 29; 0.01) = 4.04 - 42 -Table 6 . 1 . C l a s s i f i c a t i on-mat r i x fo r age generated by stepwise d isc r iminant ana l y s i s . 100% of the cases were c o r r e c t l y c l a s s i f i e d using F-to-enter o f 4 .0 . GROUP PERCENT CORRECT NUMBER OF CASES CLASSIFIED INTO GROUP MILD MEDIUM OLD X-AGED MILD 100.0 20 0 0 0 MEDIUM 100.0 0 8 0 0 OLD 100.0 0 0 6 0 X-AGED 100.0 0 0 0 2 TOTAL 100.0 . 20 8 6 2 - 43 -observat ions that best d iscr iminate among the samples i s p lo t ted on the X-axis . The second canonical var iab le which i s the next best l i nea r combination, orthogonal to the f i r s t one i s p lo t ted on the Y-axis . Therefore , each cheese sample i s represented by a pa i r of canonical v a r i ab l e s . A geometric i n te rp re ta t ion f o r age c l a s s i f i c a t i o n (Figure 8.1) shows the coordinates fo r a l l samples which are c l e a r l y separated into m i l d , medium, o ld and X-aged ca tegor ies . In add i t ion to the graphic representa t ion , SLDA enables c l a s s i f i c a t i o n of unknown samples by using d iscr iminant funct ions c h a r a c t e r i s t i c of each group of cheese. Per fect agreement for age c l a s s i f i c a t i o n between subject ive sensory eva luat ion and SDLA was obtained. The degree of p ro teo l ys i s (-NH 2), age of cheese (days) , cu l tu re type and firmness were se lec ted as the most important var iab les in th i s order fo r d i sc r im ina t ion among sample cases according to age. The degree of p r o t e o l y s i s , age of cheese and firmness have already been shown to a f f e c t age c l a s s i f i c a t i o n , ye t the s t a t i s t i c a l techique used suggests the important cont r ibut ion of cu l ture type to degree of age ( i . e . m i ld , medium, aged and X-aged). STEPWISE LINEAR DISCRIMINANT ANALYSIS (TEXTURE) Data from ind iv idua l cheese ana lys is appear in Table 4 .1 . Texture (K 2 ) in Table 4.1 was assessed by sensory Grader eva luat ion . The cheeses were d iv ided into three c l a s ses : weak, s l i g h t l y weak and f i rm and used for c l a s s i f i c a t i o n in SLDA. The s i g n i f i c a n t d i f fe rence among the three texture c l a s s i f i c a t i o n s i s observed as shown in Table 7.1. By reduct ion of the o r i g i na l 11 var iab les to 3, 94.4% of the cases are co r rec t l y c l a s s i f i e d (Table 8 .1 ) . A geometric i n te rp re ta t i on fo r texture c l a s s i f i c a t i o n (Figure 9.1) - 44 -12 C c1<£> 23CC21 M>: 31 35 M M M , 0  U " M 2 M , 3 . M g 2 8 M - M M 2524,27 29 M J 4 -18.7 -13.7 -8.7 -3.7 1.2 6.2 C a n o n i c a l var iab le 1 F igure 8.1 Canonical plot for age for the 36 commercially produced cheeses using the multivariate data in Table 4.1. A, X-aged; 0, old; C, medium; M, mild. Group means indicated by circles. - 45 -Table 7 .1 . F-matrix among the three texture groups ( f i rm, s l i g h t l y weak, weak) fo r the cheese samples af ter d iscr iminant ana l y s i s . TEXTURE Texture Firm SI. Weak SI. weak 41.22** Weak 95.08** 22.93** **P < 0.01, F (3, 31; 0.01) = 4.51 - 46 -Table 8.1 C l a s s i f i c a t i on-mat r i x fo r texture generated by stepwise d iscr iminant ana l y s i s . 94.4% of the cases were c o r r e c t l y c l a s s i f i e d using F-to-enter of 4 .0 . GROUP PERCENT CORRECT NUMBER OF CASES CLASSIFIED INTO GROUP FIRM SL. WEAK WEAK FIRM 90.0 20 2 0 SL. WEAK 100.0 0 7 0 WEAK 100.0 0 0 7 TOTAL 94.4 20 9 7 - 47 -5.0 N 2.5 0) _Q O w o > 0.0 o u c o c 0 . 9 5 5.0 --w, W4.7 F.9 -8 w„ ,-£{32,33 F27 C O f21 F f16,25 t rl7,3l -© S s 5 P3 .I5 1 1 1 J 1 1 —L 1 1 1 1 -8.2 -5.2 -2.2 0.7 C a n o n i c a l var iab le 1 3.7 6.7 F igure 9.1 Canonical plot for texture for the 36 commercially produced cheeses using the multivariate data in Table 4.1. F, firm; S, slightly weak; W, weak. Group means indicated by circles. - 48 -c l e a r l y separates a l l samples into three d i s t i n c t groups. Two m i s c l a s s i f i -ca t i ons , samples 3 and 15, occurred based on the data provided and would be bet ter c l a s s i f i e d as s l i g h t l y weak. It i s reasonable to assume that the sensory evaluat ion was incor rec t and SLDA provides a more object ive assessment in that i t is based en t i r e l y on the chemica l , physical and processing data . In f a c t , Le l i ev re and G i l l e s (1982) in t h e i r study of the re l a t ionsh ip between grade and composition of young commercial Cheddar cheese showed that sensory evaluat ion var ied from grader to grader. Results of the d iscr iminant ana lys is ind icated that moisture, process (conventional or automated) and y i e l d point were the most important var iab les for d i sc r im ina t ion among sample cases according to tex ture . The cheese making condi t ions p a r t i c u l a r l y with automated equipment tend to resu l t in cheese whose composition may change s i g n i f i c a n t l y throughout the course of a s ing le day's production (Lawrence and G i l l e s , 1980). This does not suggest that the cheese i s of poorer q u a l i t y ; however, i t may mature at a d i f f e r en t ra te . Although only 16 of the 36 cheese samples were manufactured on automated equipment, the s t a t i s t i c a l technique selected the process as the second most important var iab le a f f e c t i ng tex ture . F igure 10.1 shows the y i e l d force vs. moisture for samples 1 - 1 1 , 14, 15, 18, 24 - 29 grouped as mild by d iscr iminant ana l y s i s . It would appear that cheeses of s im i l a r composition and degree of p ro teo lys i s have y i e l d forces that are inf luenced by moisture a lone. Samples number 4, 7 - 11, 13 were assessed as being weak texture cheese by sensory evaluat ion and in a l l cases were analyzed as high moisture. Y ie ld force vs . moisture content for cheese i d e n t i f i e d as weak or s l i g h t l y weak by SLDA (samples number 1 - 16) is shown in Figure 11 .1 . The only d i f fe rence between the s t a t i s t i c a l ana lys i s of cheese texture (r^ = 0.817) and that of subject ive evaluat ion 2234.0 36.0 36.6 37.2 A f t . 37.8 38.4 Moisture , I Figure 10.1 Var ia t ion in the force at the y i e l d-po in t for cheese samples number 1-11, 14, 15, 18, 24-29 grouped as mi ld by SLDA and approximately same composition but varying moisture. See Table 3.1 for composition (r2 = 0.853). 4763 • 15 2234 36.0 36.9 378 38.7 Moisture , % F igure 11.1 Var ia t ion in the force at the y i e l d-po in t fo r cheese samples number 1-16 (r2 = 0.817). Numbers 4, 7-11, 13 o f approximately the same composition but varying moisture c l a s s i f i e d by SLDA as weak or s l i g h t l y weak texture. See Table 3.1 fo r composit ion. - 51 -( K = 0.788) was that samples no. 3 and 15 were c l a s s i f i e d as s l i g h t l y weak by the former technique. HPLC ANALYSIS The non-volat i le f lavour components conta in ing the s a l t s , amino acids and peptides and shown to contr ibute to Cheddar f lavour (McGugan et a l . , 1979) were extracted from samples number 2 and 12 (Table 4 .1 ) . These samples were se lected f i r s t l y because of t he i r s im i l a r composition and same manufacturing process but more importantly to demonstrate the e f f ec t of s t a r t e r cu l ture type on cheese t a s t e . Sample 2 made with S>. cremoris was 199 days old and assessed by sensory evaluat ion as mild whereas sample 12 made with J>. l a c t i s was 136 days old and graded as medium. Quant i ta t ive determination of p ro teo lys i s by assaying for free amino groups c l e a r l y showed less p ro teo lys i s in cheese made with j>. cremoris compared to that made with j>. l a c t i s (Table 4.1) although the former was 63 days older than the l a t t e r . The HPLC p r o f i l e using an Adsorbosphere C 8 column and o p t i -mized gradient e lu t ion as discussed in Part II ind icated less prote in degradation products in cheese made with S^ . cremoris (Figure 12.1) compared to that made with S^ . l a c t i s (Figure 13.1 ) . Indeed, based on the respect ive HPLC pattern obtained and using stepwise l i nea r d iscr iminant ana lys i s (Part I I ) , the cheeses were c l a s s i f i e d into mild and medium categor ies respect i ve ly regardless of age. - 52 -F igure 12.1 HPLC p r o f i l e of the water extract from a 199 day o ld cheese, made with jS. cremoris and sub jec t i ve l y graded m i l d . See Table 4.1 (sample no 2). - 53 -F igure 13.1 HPLC p r o f i l e of the water extract from a 136 day old cheese, made with ^ . l a c t i s and sub jec t i ve l y graded medium. See Table 4.1 (sample no 12) . - 54 -CONCLUSIONS Six defined cu l ture s t ra ins of bacter iophage-insens i t ive Streptococcus  cremoris propagated in whey-based s t a r t e r media were success fu l l y used over a period of 10 months to produce more than two m i l l i o n kg of Cheddar cheese on commercial continuous cheesemaking equipment. 1) The cu l tures grown in the pH cont ro l l ed whey-based media retained t he i r a c t i v i t y for up to s ix days. 2) Texture d i f fe rences in the cheese were s i g n i f i c a n t and were re lated to va r i a t ion in composition and degree of p r o t e o l y s i s . Heat t r ea t ing compared to pas teur iza t ion of the milk resul ted in d i f fe rences between the force-compression rheograms of the resu l tant cheeses. 3) Cheese made with S. cremoris s t ra ins required a longer r ipening time to develop the same f lavour i n t ens i t y compared to cheese made with j>. l a c t i s cu l t u r e s . This was re lated to p ro teo l y t i c a c t i v i t y and quant i f i ed by measurement of free amino groups. Moreover, HPLC p r o f i l e s of the non-volat i le water extracts from cheeses made with _S. cremoris were shown to have patterns c h a r a c t e r i s t i c of mild cheese. 4) Pattern recogni t ion techniques (SLDA) used to analyze the mul t i va r i a te data suggested texture was a f fec ted by the automated process , moisture content and y i e l d po in t . Casein p r o t e o l y s i s , age, cu l tu re type and firmness were the most d i sc r imina t ing var iab les a f f e c t i ng matur i ty . 5) App l i ca t ion of mu l t i va r i a te ana lys is demonstrates a technique for poss ib le use in evaluat ions of experiments or processes to detect sys temat ica l l y the s i gn i f i c ance of each of the experimental v a r i ab l e s . This approach a lso provides a method for evaluat ion of multidimensional data . In both age and texture assessment the number of s i g n i f i c a n t independent var iab les required to obtain correct c l a s s i f i c a t i o n was reduced from 11 to 4 and 11 to 3 - 55 -respect i ve ly without loss of in format ion. This technique indeed provides a powerful tool in ext rac t ing s i g n i f i c a n t var iab les and in der iv ing useful information from a mult i tude of data . - 56 -PART II - 57 -INTRODUCTION Flavour is the sensation produced by a material taken in the mouth, perceived p r i n c i p a l l y by the senses of taste and smell and a lso by the general pa in , t a c t i l e and temperature receptors in the mouth. Flavour a lso denotes the sum of the cha ra c t e r i s t i c s of the material which produce that sensation ( H a l l , 1968). Flavour therefore is a sensory phenomenon and is by far the most important fac tor which governs our apprec ia t ion of the foods we eat . The dairy industry i s aware of the importance of f lavour as an essent ia l fac tor in food se lec t ion and acceptance and therefore have attached much value to f lavour qua l i t y control (Badings and Neeter, 1980). The texture and f lavour propert ies of cheese are not obtained un t i l a f t e r a r ipening pe r iod , the length of which var ies with the type of cheese and cannot be maintained at t he i r best for an i n d e f i n i t e period of t ime. This means that what we have to observe is not constant with t ime. As a consequence of the heterogeneous nature of the product and the complexity of i t s cons t i tuen ts , the chemical basis of cheese f lavour and aroma has not been e luc ia ted despite a large number of s tud ies . Most studies have placed emphasis on the v o l a t i l e aroma, permi t t ing , in some cases, one to obtain an ins ight into the broad mechanism of f l a vo r and aroma development, but regret tab ly s t i l l leaving many questions unanswered (Adda et aj_., 1982). Cheese r ipening i s e s sen t i a l l y the con t ro l l ed slow decomposition of the const i tuents of milk in a phys i ca l l y complex matr ix. The prec ise nature of the react ions which produce f lavour compounds and the way in which the i r r e l a t i ve rates are cont ro l l ed is poorly understood. This has been due f i r s t l y to the lack of knowledge of the compounds which impart t yp i ca l f lavour to Cheddar cheese and secondly to the complexity of the cheese mic ro f lo ra as the potent ia l producers of f lavour compounds. - 58 -It becomes evident therefore that the most e lus ive and var iable parameter in cheese is the concept of f l avour i t s e l f . Not only have d i f f e r e n t research groups expressed d i f ferences of opinions on which compounds are important to cheese f l avour but considerable d i f ferences can occur among sensory evaluat ion pane l i s t s , which is how cheese is usua l ly assessed. The need fo r sensory panel t r a in ing and s t r i c t s t a t i s t i c a l control is therefore imperat ive. Sensory evaluat ion is also applied in research, but has obvious l im i t a t i ons when comparing samples over an extended time period or between research l abora to r i es . A major problem, there fore , encountered in cheese maturation is the lack of a r e l i a b l e method fo r est imating the extent of f lavour development. C l a s s i c a l l y , t ra ined graders and sensory panels have performed th i s duty, but because of the va r i ab l e , subject ive and time consuming nature of organolept ic methods, a s imple, more object ive and r e l i a b l e method fo r accurate ly assessing cheese f lavour is requ i red . A chemical index or indices of maturation would allow better comparisons of r ipen ing e f fec ts and more object ive universal assessment of cheese f lavour i n t ens i t y and qua l i t y . The analys is of cheese for f lavour enhancing compounds which might serve as indices has been extens ive ly conducted. Hence, t h i s study assesses the cont r ibut ion of the water-soluble cheese f r a c t i o n containing the s a l t s , amino acids and peptides which have been shown to be responsib le for the i n t ens i t y of cheese f l avour (McGugan et^  a l . , 1979). The non-volat i le water f r a c t i on was analysed by reversed-phase high performance l i q u i d chromatography (RP/HPLC). The use of reversed-phase high performance l i q u i d chromatography (RP/HPLC) as an ana l y t i ca l technique fo r prote in separat ion, i s o l a t i o n and cha rac te r i za t ion i s r ap id l y gaining in te res t (Hearn et al_. , 1982; Tweeten - 59 -and Euston, 1980) because of the potent ia l to permit rapid and highly se l e c t i ve separations of pept ides , po lypept ides , proteins and amino a c ids . However, a formidable task ex is ts to se lect the correct mobile phase as well as the operat ing cond i t i ons . Many l i q u i d chromatographic separations have been developed by random se lec t ion of various mobile phase solvent mixtures with l i t t l e theore t i ca l cons idera t ions . Systematic opt imizat ion of the mobile phase for s e l e c t i v i t y showed d i s t i n c t advantages in achieving good reso lut ion of a l l components of a mixture (Glajch et a l_ . , 1980). However, the systematic opt imizat ion of the mobile phase fo r s e l e c t i v i t y in l i q u i d chromatography i s a r e l a t i v e l y new development and is not yet widely p r ac t i c ed . Therefore , a fur ther aim of the present research was to use Snyder's (1974, 1978) empir ical c l a s s i f i c a t i o n of solvent propert ies for s e l e c t i v i t y and a new mapping simplex opt imizat ion technique (Nakai et a l . , 1984) to improve reso lu t ion of the HPLC p r o f i l e . These more e f f i c i e n t and e f f e c t i v e procedures have the potent ia l for both saving considerable development time and increas ing the information content of the f i n a l HPLC p r o f i l e . Moreover, because of the complexity of the HPLC p r o f i l e s i t i s somewhat d i f f i c u l t to determine or predict s i g n i f i c a n t peaks responsible for taste and/or c l a s s i f i c a t i o n of cheese samples. Therefore , s t a t i s t i c a l pattern recogni t ion techniques: p r inc ipa l component ana lys is and stepwise l i nea r d iscr iminant ana lys is were appl ied to the HPLC peaks in order to c l a s s i f y unknown cheeses into ca tegor ies , or to separate cheeses into ca tegor i es . S i gn i f i c an t peaks estab l i shed by pattern recogni t ion techniques were co l l e c t ed by using RP/HPLC. The amino ac id contents of the f r ac t ions were evaluated to provide fur ther information on the propert ies of pept ides , - 60 -thus providing a bet ter understanding of the peptides and/or amino acids impl icated in the development of the complexity of cheese f lavour and b i t t e r t as te . The present study recognizes taste in a purely phys io log ica l sense o f the term ( i . e . sweet, sour, s a l t , acid and b i t t e r ) and that f lavour i s a complex combination of t a s te , smel l , appearance, texture , e t c . and that the ana ly t i ca l technique proposed may be used for the assessment of Cheddar cheese t as te . - 61 -LITERATURE REVIEW CHEDDAR CHEESE FLAVOUR Early research on Cheddar cheese f lavour sought a s ing le compound or c l ass of compounds responsible fo r cha r a c t e r i s t i c f l a vour . When no such compound or c lass was found, Mulder (1952) proposed what was known as the Component Balance Theory. This theory suggested that Cheddar f lavour was made up of a balance of f lavours contr ibuted by a number of d i f f e r en t compounds. When the balance was upset by an excess or lack of one or more of the component compounds, atyp ica l f lavour was produced. Cheese f lavour r esu l t s from the act ion of microorganisms and enzymes on the carbohydrates, fa t and proteins of the milk and curd. The spectrum of compounds produced i s often so wide and complex that those involved in the f lavour remain unknown even a f t e r the app l i ca t ion of soph is t i ca ted chemical ana l y s i s . Because of the economic s i gn i f i c ance of Cheddar cheese and consumer acceptance, the dairy industry has placed great emphasis on f lavour qual i ty c o n t r o l . Despite the in tens ive research into Cheddar cheese f lavour the agents responsible and the i r modes of ac t ion are s t i l l l a rge l y unknown. Many aspects of cheese chemistry and f lavour have been r igorous ly i n v e s t i -gated and thoroughly reviewed (Aston and Du l l ey , 1982; Adda et aj_., 1982; Law, 1981; Badings and Neeter, 1980; Forss , 1979; McGugan et al_., 1979; Evans and Mabbit, 1974, K r i s t o f f e r s e n , 1973). However, attempts to re la te the cha r a c t e r i s t i c f lavour and f lavour qua l i t y of a given var iety to a s ing le or a group of c lose l y re la ted compounds have not been success fu l . AMINO ACIDS Amino ac ids , peptides and polypeptides produced during cheese r ipening 62 -are the r e su l t of the ac t ion of p ro t eo l y t i c enzymes o r i g ina t i ng from cheese mi lk , rennet and added bac te r ia l c u l t u r e s . The p r inc ipa l pathway of p ro teo l y t i c breakdown during the r ipening process i s thought to be the one by which rennet causes a primary, l i m i t e d , degradation of the caseins a f t e r which the s t a r t e r bac te r i a produce smal ler peptide fragments and free amino acids (V isser , 1981; Lowrie and Lawrence, 1972). The mechanism of prote in breakdown during r ipening i s well documented (Desmazeaud and Gr ipon, 1977). It begins with the act ion of rennet which cleaves the Pheio5 - MetiQ6 bond of *-case in , thus inducing c l o t t i n g . In add i t i on , rennet cleaves the Phe23 - Phe24 and Phe24 - Val25 bonds of a s i - c a s e i n during the ear ly stage of r i pen ing , with subsequent appearance of an a$\l f r a c t i o n . The act ion of the rennet goes on during the whole r ipening pe r iod , inducing mainly the release of large molecular weight pept ides , but no free amino acids are produced by the enzyme. Law (1981) summarized the sequence of enzymatic p ro te in breakdown by rennet, milk proteinase and s ta r t e r proteinase (Figure 14.2). Free amino acids were the f i r s t compounds invest igated for t he i r cont r ibu t ion to cheese f l a vour . Glutamic acid has been shown to be respons ib le for the "brothy" f lavour in cheese (Mulder, 1952; Harper, 1959) and Kosikowski (1951) found that as the total free amino acids increased , cheese f l avour increased . J a r r e t t et al_. (1982) l i k e others (Ney, 1971; Nieuwoudt, 1977; Pettersson and Sjostrom, 1975) recognized the importance of free amino acids to f lavour and proposed a method for measuring phosphotungstic ac id-soluble amino ni t rogen (PTA-soluble amino N). The amino N leve l in the PTA f i l t r a t e was shown to be a good estimate of the tota l f ree amino acid l e v e l s . They showed that although the technique did not measure d ibas i c amino acids and p ro l ine there was a l i n e a r r e l a t ionsh ip between cheese r ipen ing and free amino ac ids . - 63 -Casein Rennet Star ter Rennet V High mol. wt. peptides Low mol. wt. peptides Amino acids Star ter (milk) Star ter (milk) Flavour and aroma compounds ( v o l a t i l e f a t t y ac ids , amines, sulphur compounds) F igure 14.2 Schematic showing enzymatic hydro lys is of casein by rennet, milk proteinase and s ta r te r prote inase. - 64 -Pr io r to 1979 i t was genera l ly concurred that f ree amino acids made l i t t l e or no d i r ec t cont r ibut ion to Cheddar cheese f l a vour . However, McGugan et aj_. (1979) have recognized the importance of the non-volat i le water extractable f r a c t i o n s . This group f rac t ionated a mild and aged cheese and recombined the f rac t ions in various combinations to assess the f l avour of each f r a c t i o n . The water-soluble f r a c t i on conta in ing the s a l t s , amino acids and peptides contr ibuted most to the i n t ens i t y . No s i g n i f i c a n t d i f f e rence was found between the deodorised and undeodorised fat f r a c t i o n s , i nd i ca t i ng that the loss of vo l a t i l e s was not important, at least not to f l avour i n t ens i t y . They concluded that the vo l a t i l e s may contr ibute to the qua l i t y of f lavour while the water-soluble f r a c t i on provides the in tens i t y of f l a vou r . FREE FATTY ACIDS Free f a t t y acids are hydrolyzed from milk t r i g l y c e r i d e s by two major sources: (1) breakdown of the fat by l i p o l y s i s e i ther by mi lk , s t a r t e r bac te r i a and poss ib l y rennet l ipases and (2) metabolism of carbohydrates and amino acids by bac te r ia Adda et (1982). However, evidence ind icates that l i p o l y s i s is the major contr ibutor of f ree f a t t y ac ids . Forss (1979) showed that butyr ic acid in cheese was present at twice the amount of that in milk and suggested that i t should make a major c o n t r i -bution to cheese f lavour while some of the minor f ree f a t t y acids enhanced f l avour by synergism. Mabbit (1961), Patton (1963), and Forss and Patton (1966) also recognized the importance of f ree f a t t y acids as const i tuents to Cheddar cheese f l avour . Yet , l i k e others , B i l l s and Day (1964) and Law and Sharpe (1977) found no co r r e l a t i on between Cheddar cheese f lavour and - 65 -f ree f a t t y ac ids . The l i t e r a t u r e on the subject is therefore vast and c o n f l i c t i n g . The importance of f ree f a t t y acids to the f lavour of cheeses such as Romano and Parmesan is well accepted yet the i r ro l e in Cheddar f lavour i s much less apparent. Aston and Dul ley (1982) summarized techniques used to r e l a t e Cheddar f l avour to f ree f a t t y ac ids : (1) determination of t he i r l eve ls i n d i v i d u a l l y , c o l l e c t i v e l y or as ra t ios in order to co r re l a te these with f lavour development; (2) addi t ion of f a t t y ac ids , sometimes with other compounds to bland bases to determine whether Cheddar f lavour can be produced or improved; (3) s e l e c t i v e removal of the f ree f a t t y acids from extracts possessing Cheddar f lavour to detect any a l t e ra t ion in f l avour and (4) manufacture of Cheddar cheese whose fa t content has been decreased or replaced by a fat of vegetable o r i g i n . The l a t t e r technique has been used by a number of workers (Ltick and Downes, 1972; Foda et al_. , 1974; Harper, et aj_., 1979) and the cheese was found to have reduced leve ls of f ree f a t t y acids and subsequent lack of t yp i ca l Cheddar f l a vour . Anders and Jago (1970) manufactured cheese from polyunsaturated fat f o r t i f i e d milk and found the lack of Cheddar cheese f l avour due to the i n h i b i t o r y e f f ec t s of l i n o l e i c acid which i n h i b i t s the pyruvate dehydrogenase system in s t a r t e r S t rep tococc i . Increasing the fa t content above a cer ta in l im i t does not improve f l avour and may resu l t in l i p o l y s i s and ox ida t ion . - 66 -VOLATILE SULFUR COMPOUNDS V o l a t i l e su l f u r compounds (VSC's) have been considered important in cheese f lavour fo r some time but with the in t roduct ion of more sens i t i ve flame photometric detectors for gas chromatographic ana l y s i s , the very low l eve l s in cheese can now be detected. VSC's are unstable to condi t ions commonly encountered in f lavour chemical studies and th i s may account for the incomplete and contrad ic tory data on the ro les of su l fu r compounds in cheese f lavour (Manning, 1979; Lamparsky and Kl imes, 1981). K r i s to f f e r sen (1973) postulated that cheese f lavour qua l i t y resu l ted from a de f i n i t e re l a t ionsh ip between the r e l a t i ve concentrat ions of free fa t ty acids and hydrogen su lph ide . Cheese manufactured from milk with the sulphur groups in the -S-S- state developed f u l l e r and super ior f lavour in comparison with cheese made with the sulphur groups in the -SH s t a t e . The balance of -SH and -SS- groups modif ies enzyme a c t i v i t y in that hydrogen resu l t ing from oxidat ion-reduct ion react ions during the r ipening process cannot be deposited by reducing the -SS- groups. However, Law and Sharpe (1977) found no re l a t ion between free fa t ty acids and H2S to cheese f l a vou r . Methanethiol i s present in Cheddar in amounts s u f f i c i e n t to contr ibute to f lavour (Day et aT_., 1960). The absence of methanethiol from cheese and i t s removal from headspace v o l a t i l e s resu l t s in the absence of t yp i ca l cheese f lavour and aroma, respect i ve ly (Manning, 1974; Manning and P r i c e , 1977). Hydrogen sulphide contr ibutes to aroma (Manning and P r i c e , 1977) yet high concentrat ions resu l t in ' su lph ide ' o f f - f l a v o u r . Although sulphur compounds may be involved in cheese f lavour ( K r i s t o f f e r s o n , 1973) they cannot give balanced Cheddar f lavour and aroma - 67 -and may be responsible for undefined contr ibut ions to the overa l l f lavour p r o f i l e (Law, 1981). NON-SULFUR VOLATILE COMPOUNDS In mold cured cheese the production of methyl ketones may provide c h a r a c t e r i s t i c f l a vour . The mechanism of format ion, involvement of spores and environmental condit ions have been reviewed by Adda et al_. (1982). The cont r ibut ion of methyl ketones to Cheddar f lavour was invest igated by Walker and Keen (1974). Gamma and de l ta lactones have been quant i t a t i ve l y determined in Cheddar cheese by gas chromatography-mass spectrometry of a c e t o n i t r i l e extracts and used to provide an overa l l index for maturation and f lavour development (Wong et al_. , 1975). They suggested that the lactones not only contr ibuted a f lavour note of t he i r own but a lso when blended with other compounds produced an add i t i ve and smoother f l a vour . Concentrat ion of lactones was higher in rancid cheeses, suggesting that d i f f e r en t pathways fo r lactone formation were poss ib le when ranc id i t y occurred. As pointed out in the preceding review appreciable work has already been done to better understand Cheddar cheese f l avour . However, i t i s general ly agreed that more basic research into chemical and biochemical aspects of cheese r ipening is requ i red . In addi t ion to the complexity of the ana l y t i ca l data already ava i l ab le and the d i f fe rences of opinions on which compounds are important to Cheddar cheese f lavour considerable d i f fe rences occur among sensory panel r e s u l t s . Thus, with the data and ana l y t i ca l procedures present ly ava i l ab le a more object ive method of assess ing Cheddar cheese f lavour qua l i t y i s imperat ive. - 68 -ANALYTICAL TECHNIQUES P ro teo l ys i s and in some cases l i p o l y s i s appear to be key processes determining the rate of texture and f lavour development in most v a r i e t i e s of cheese (Law, 1978) and the extent of p ro teo l ys i s has been used as an i nd i c a t i on of the degree of maturity of r ipening cheese (Ney, 1971; Nieuwoudt, 1977). Various e l ec t rophore t i c methods present ly a v a i l a b l e , employing d i f f e r en t separat ion media (paper, starch g e l , polyacrylamide gel) and d i f f e r e n t procedures (SOS, high vo l tage , i s o e l e c t r o f o c u s i n g , bidimensional ) have been used to study prote in degradation during cheese r i p en i ng , but the most popular is the polyacrylamide gel e lec t rophores is (PAGE) method (Park et al_., 1978; Thomas and Pearce, 1981). A quant i ta t i ve method for est imat ion of non-soluble N substances was developed using PAGE (de Jong, 1975). The method i s r e l a t i v e l y simple and from the densitograms the measurement of the degree of unhydrolyzed, nat ive as- and 3-caseins and to a ce r ta in extent the i d e n t i f i c a t i o n of major breakdown products inc lud ing a s l -I-casein can be made. As pointed out by Olson and co-workers (1983) va r i a t ions of the method have been published and s tandard izat ion of the procedure i s required before i t can be used as a quant i ta t i ve assessment of cheese r ipen ing . Trieu-Cuot and Gripon (1982) used i s o e l e c t r i c focusing and 2-dimensional e lec t rophores i s to study the pH 4.6 inso lub le f r a c t i on during Camembert cheese r ipen ing . G e l - f i l t r a t i o n ( Sa l j i and Kroger, 1981) has been used as a separat ion technique fo r free amino acids and i s rout ine ly used to f r ac t iona te b i t t e r peptides from Cheddar cheese (Champion and Stanley , 1982; Edwards and Kosikowski, 1983). However, in sp i te of a var ie ty of the conventional chromatographic procedures ava i l ab le they - 69 -genera l ly lack speed, reso lv ing power and recovery. Therefore , high performance l i q u i d chromatography (HPLC) i s becoming inc reas ing ly important in quant i ta t i ve ana lys i s . The v e r s a t i l i t y of t h i s technique i s well recognized and technica l pub l i ca t ions are numerous. Bican and Blanc (1982) showed a great s i m i l a r i t y between the e lec t rophoret i c p r o f i l e of whey prote in and casein to t he i r respect ive HPLC ana lys is and suggested i t as an a l t e rna t i ve method to study casein degradation in a r ipening cheese. REVERSED-PHASE LIQUID CHROMATOGRAPHY Reversed-phase high performance l i q u i d chromatography u t i l i z i n g a nonpolar s tat ionary phase, usual ly a f u l l y porous mic ropar t i cu la te s i l i c a in the range of 5-10 um chemical ly bonded with alkyl chains and an aqueous polar mobile phase used to e lute the strongly retained compounds, permits the rapid and h ighly se lec t i ve separations of amino ac ids , pept ides , polypepides and proteins (Tweeten and Euston, 1980; Hearn et a l_ . , 1979). Although the number of studies using RP/HPLC are numerous (Heukeshoven and Dernick, 1982; Hancock and Sparrow, 1981; Hancock et a l . , 1981; Hearn, et al_., 1982; Gazdag and Szepes i , 1981; Wilson et al_., 1981; Bishop et a l . , 1980) and exce l l en t reviews are ava i l ab le (Regnier and Gooding, 1980), the ana lys is of proteins i s r e l a t i v e l y new. This has been due to the r e l a t i v e l y recent development of HPLC columns having support pa r t i c l e s with 300 or 500 A pores al lowing in te rac t i on of the proteins with the s ta t ionary phase. Diosady et al_. (1980) described the separat ion of whey prote in and a quant i ta t i ve est imation using RP/HPLC was obtained. Recently, Bican and Blanc (1982) studied the separat ion of whey proteins and suggested that the technique could provide HPLC p r o f i l e s which are prerequ is i tes when studying - 70 -prote in degradation products, during cheese r ipen ing . Pearce (1983) using a short a lky l chain Cg reverse phase column completely resolved the major whey proteins in 30 min using an ac id s a l i n e / a c e t o n i t r i l e grad ient . Champion and Stanley (1982) extracted water so luble Cheddar cheese f lavour compounds using the separat ion method of Harwalkar and E l l i o t (1970). A water/91% methanol l i nea r gradient at a rate of change of 1.14% methanol/min was used. The extract contained at least seventy-one compounds and from the HPLC e lu t ion pattern they suggested that hydrophobicity inf luences separat ion . High molecular weight compounds eluted e a r l i e r . There was however, no re l a t ionsh ip between molecular weight of the compound in the b i t t e r extract and the i r retent ion on the column. Pham and Nakai (1984) studied the importance of the non-volat i le f lavour components in Cheddar cheese thought to be responsib le for the in tens i t y of Cheddar cheese f lavour (McGugan et aj_., 1979). Using a reversed-phase column packed with Adsorbosphere Cg A l k y l s i l i c a and an i s o c r a t i c e lu t ion (0.1 M phosphate buffer pH 6 .0 ) , 13 c l e a r l y resolved peaks were obta ined. They suggested that RP/HPLC ana lys is of the water-extract could be used as an object ive evaluat ion of r ipening of Cheddar cheese. MOBILE PHASE A considerable body of experimental work has been addressed to the inves t iga t ion of mobile phase e f fec ts in the RP/HPLC of amino a c i d s , peptides and pro te ins . Taken in i s o l a t i o n , these studies provide a vast array of d i f f e r en t e lu t ion systems that makes the se l ec t ion of the mobile phase d i f f i c u l t . Due to the number of studies a v a i l a b l e , i t i s poss ib le to - 71 -make a rat ional i n te rp re ta t ion in se lec t ing e lu t ing mobile phase condit ions that provide good separat ion . However, in sp i te of the theore t i ca l background, optimized ana ly t i ca l methods are seldom used (Glajch et a l . , 1980). It i s well recognized that in RP/HPLC the more hydrophobic the compound the greater i s i t s i n te rac t ion with the hydrophobic bonded stat ionary phase at a given eluant composit ion. Snyder (1974, 1978) proposed a general scheme for c l a s s i f i c a t i o n of common solvents according to t he i r " p o l a r i t y " or chromatographic strength and to t he i r s e l e c t i v i t y or r e l a t i v e a b i l i t y to engage in hydrogen bonding. Various solvents could then be grouped into a s e l e c t i v i t y - t r i a n g l e concept and the strategy is to inves t iga te solvents from the apices of the solvent t r i a n g l e . The systematic opt imizat ion of the mobile phase for s e l e c t i v i t y to improve reso lu t ion in l i q u i d chromatography (Snyder and K i r k l and , 1979) i s a r e l a t i v e l y new development and not widely p rac t i s ed . A number of methods have been used for opt imizat ion of chromatographic systems (Morgan and Deming, 1975). The chromatographic response funct ion (CRF) i s used to measure the performance of both gas and l i q u i d chromatography. The CRF has the general form: k CRF = ^ In (Pi) i=l Where Pj i s a measure of separat ion between adjacent peaks in a chromatogram for k peak p a i r s , where k i s one less than the to ta l number of peaks. The la rgest CRF value general ly ind icates the best separat ion of the mixture. A major problem encountered during opt imizat ion i s a way of measuring - 72 -the quality of the analysis. Four parameters are generally used to evaluate the quality of the liquid chromatographic system. These parameters are kl, the capacity factor; a , the separation factor; N, the number of theoretical plates; and R, resolution. Mathematically these are: v 0 a = K _ Z o ^1 _ 2(V ? - VT)  K Wx + w 2 The k1 is a measure of a compound's retention in terms of the column volume. The separation factor a", is the ratio of the net retention time for any two components. N, is the number of theoretical plates in a column and can be increased by increasing column length (Pomeranz and Meloan, 1980). Glajch et al_. (1980) reviewed optimization techniques for chromatographic systems. They proposed a new method of data analysis called overlapping resolution mapping (ORM) which combines the Snyder solvent selectivity-triangle concept with a statistical simplex design technique for optimizing the mobile phase composition in RP/HPLC. Advantages claimed were that all pertinent information regarding a peak or pair of peaks is used in the optimization. In addition, crossovers and fused peaks can be handled thus providing good resolution. - 73 -MULTIVARIATE ANALYSIS A major problem encountered in cheese r ipening i s the lack of a r e l i a b l e method fo r est imating the extent of f lavour development. C l a s s i -c a l l y , t ra ined graders and sensory panels have performed th i s duty, but because of the var iab le and subject ive nature of organolept ic methods, a simple more ob jec t i ve method for accurate ly assessing cheese f lavour i s requ i red . Almost a l l ana ly t i ca l techniques proposed to assess cheese f lavour development are compared to Graders' r esu l t s and/or a t ra ined sensory panel (Aston et a]_., 1983; Grieve and Du l l ey , 1983; McGugan et a l . , 1979; Harwalkar and E l l i o t , 1970). Data obtained from sensory evaluat ion are usual ly subjected to merely rout ine s t a t i s t i c a l ana l ys i s . Seldom are factors being used as desc r ip to r s , examined f i r s t to determine whether they are t ru l y c r i t i c a l to the evaluat ion (Derde and Massart, 1982). The d i f f i c u l t y in i n te rp re t ing the resu l t s of ana l y t i ca l techniques which produce simultaneously a large number of parameters has led to increas ing i n t e r es t in pattern r ecogn i t i on . Exce l l ent reviews have been presented (Kowalski, 1980; Frank and Kowalski , 1982) on the t o p i c . These authors have introduced an i n t e r d i s c i p l i n a r y sc ience, Chemometrics, which has been defined as the app l i ca t ion of mathematical and s t a t i s t i c a l methods to chemical measurements. Pattern recogn i t ion techniques are a method fo r analyzing mul t i va r ia te data in order to c l a s s i f y unknown objects into ca tegor i es , or to separate objects into categor ies (Varmuza, 1980) while giv ing data reduct ion with minimal loss of in format ion. Mu l t i v a r i a t e data ana lys i s o f fe rs various methods fo r e f f i c i e n t s i m p l i f i c a t i o n and in te rp re ta t i on of many d i f f e r en t var iab les - 74 -simultaneously . Pattern recogni t ion fo r mul t i va r ia te ana lys is has been used in chemistry , microbiology and medicine for i d e n t i f i c a t i o n of chemical compounds, c l a s s i f i c a t i o n of bacter ia species and diagnosis of diseases r espec t i ve l y . The mul t i va r i a te data ana lys is technique i s f ind ing increased app l i ca t ion in food science and technology. Moreover, because of the in te res t in computer aided ana lys is of mu l t i va r i a te food research data an i n t e r d i s c i p l i n a r y IUFoST-symposium was held in Os lo , Norway (1982). The proceedings to that symposium were prepared in an exce l lent review on the subject (Martens and Russwurm, 1983). Powers and Keith (1968) recognized the importance of gas chromatographic d i f ferences in the v o l a t i l e s of foods and how these d i f fe rences are re la ted to q u a l i t y . They observed that attempts to co r re l a te gas chromatographic data with organolept ic f lavour did not use the en t i r e chromatogram but rather one or r e l a t i v e l y few peaks and suggested that v isual inspect ion or even simple mathematical evaluat ion resul ted in poor co r re l a t ions between the f lavour and chemical compound. Using stepwise d iscr iminant ana lys is (SLDA, a pattern recogni t ion technique) four lo ts of roasted cof fee with d i f f e r en t organolept ic q u a l i t i e s were examined by gas chromatography (GC). By ca l cu l a t i ng a l l poss ib le ra t ios among peak heights and subject ing these ra t ios to SLDA the cof fee was co r rec t l y c l a s s i f i e d into the four f lavour ca tegor ies . Smeyers-Verbeke et aj_. (1977) appl ied d iscr iminant ana lys is to the c l a s s i f i c a t i o n of milk samples according to t he i r o r i g i n . The ana lys is was based on GC data for fa t ty acids in milk fat samples from cows, sheep and goats. Between the pure samples, SLDA allowed complete d i s c r im ina t i on . Extension of the inves t iga t ion to mixtures of these samples showed a high degree of correct c l a s s i f i c a t i o n s between the mixtures and the pure m i lks . - 75 -They concluded that the method should allow a higher proport ion of correct c l a s s i f i c a t i o n s than i s poss ib le by v isual inspect ion of the chromatograms. The data for s ixteen d i f f e r en t chemical components of wine were analyzed by pattern recogni t ion techniques and the samples c l a s s i f i e d according to vintage year and wine region (Kwan and Kowalski , 1978). Aishima (1979) appl ied SLDA and pr inc ipa l component ana lys i s (PCA) to GC p r o f i l e s of soy sauce v o l a t i l e s in order to compare soy sauce qua l i t y assessed by sensory evaluat ion and GC data. The soy sauce brands were roughly separated into four groups by PCA whereas SLDA co r r ec t l y c l a s s i f i e d each sample into the proper group of the eight brands. A highly s i g n i f i c a n t re la t ionsh ip between the p r inc ipa l components and sensory score were obtained ( r 2 = 0.915). Aishima (1979, 1982, 1983), using mul t i va r i a te ana lys is co r rec t l y c l a s s i f i e d soy sauce samples based on t he i r GC p r o f i l e s into the correct brands and found a c lose re l a t ionsh ip with sensory eva lua t ion . Eight p r inc ipa l components were extracted from the 39 GC peaks as s i g n i f i c a n t factors responsib le for the soy sauce aroma. Multielemental data of wine determined by atomic emission spectrometry was analyzed by pattern recogni t ion techniques to extract key elements and combination of elements which were c h a r a c t i s t i c of geographic o r ig ins and to inves t igate subt le d i f fe rences in elemental concentrat ions due to in t rareg iona l and vintage var ia t ions (Kwan et a l_ . , 1979). The quant i ta t i ve determination of v o l a t i l e compounds of grape brandies were analyzed by GC and using SLDA the cognacs from the respect ive brandies of d i f f e r en t regions were d e f i n i t e l y d i s t ingu ished (Schreier and Reiner, 1979). Kwan and Kowalski , (1980b) using gas chromatography-mass spectrometry data c l a s s i f i e d Pinot Noir wines according to t he i r geographic o r i g i n s . These same workers in a subsequent study (Kwan and Kowalski , - 76 -1980a) corre la ted organic and elemental compositions with sensory evaluat ions and by using PCA were able to reduce the large number of chemical components to only a few key components re lated to overa l l qua l i t y of the wine. PCA appl ied to scores obtained from the sensory evaluat ion of wine provided information on the consistency of the pane l i s t and areas requ i r ing improvement (Kwan and Kowalski , 1980a). Recently Woo and Lindsay (1983) developed a method for detect ing and pred ic t ing hydro ly t i c ranc id i t y o f f f lavours in butter by co r r e l a t i ng sensory data with free fa t ty acid concentrat ions by using stepwise regression and pattern recogni t ion techniques. They suggested that the procedure developed should allow the dairy industry to p rac t i ce more va l id qua l i t y control programs for butter f l a vours . Pham and Nakai (1984) analyzed water extracts from Cheddar cheese samples by RP/HPLC. SLDA appl ied to the HPLC p r o f i l e s c l a s s i f i e d the cheeses into m i l d , medium, old and extra-old based so le l y on the water-soluble compounds. Discr iminant funct ions were ca lcu la ted and used to c l a s s i f y unknown samples of cheese into t he i r correct age ca tegor ies . - 77 -MATERIALS AND METHODS 1. SAMPLE PREPARATION Cheddar cheese samples from d i f f e r en t manufacturers and represent ing d i f f e r en t ages and qua l i t y were purchased from r e t a i l o u t l e t s . The ext rac t ion method as out l ined by McGugan et ^1_. (1979) was fol lowed (Figure 15.2) with some minor modi f icat ions as described by Pham and Nakai (1984). Four grams of the commercially produced cheese sample was centr i fuged at 30,000 X g fo r 25 min at 25 °C . A f te r decanting the f a t , the residue was extracted to remove residual fat by mixing f i r s t with 2 mL methyl a l c o h o l , then adding 0.2 mL water and mixing. A f te r cent r i fug ing the mixture (1000 X g; 30 min) the solvents were decanted and the residue was extracted twice more, as above, with 1.0 mL methyl a l c o h o l , 1.0 mL methylene ch lo r ide and 0.6 mL water for each ex t r a c t i on . The water-soluble f r a c t i on was prepared by adding 2.0 mL water to the combined solvent ex t r a c t s . Two solvent layers and a p rec ip i t a t e were separated by cen t r i fug ing (1000 X g; 30 min) and the resu l t ing p rec ip i t a t e added to the previous res idue. The methyl alcohol f r a c t i on was extracted with two 0.25 mL port ions of methylene ch lor ide to ensure removal of fa t-so lub le ma te r i a l . The methyl alcohol-water f r a c t i on was evaporated by p lac ing the tubes in a S i l l i Therm™ heating module (Pearce Chemical Company, Box 117, Rockford, IL) set at a temperature of 45°C under a steady stream of n i t rogen . Water (2.0 mL) was added and evaporation cont inued. The procedure was repeated a fur ther s ix times to remove v o l a t i l e s from the water-soluble f r a c t i o n . This f r a c t i on was combined with the water-soluble extract obtained by ex t rac t ing the res idue . The combined water-soluble extracts were fur ther evaporated to approximately 1.0 mL and then made up - 78 -MILD or AGED CHEESE Centr i fuge UNDEODORIZED FAT RESIDUE Weigh Steam d i s t i l Vacuum d i s t i l Mix with methyl a l c o h o l , methyl ene ch lo r ide and water, cen t r i fuge : repeat two t imes. DEODORIZED FAT EXTRACT add 0.2 mL water cent r i fuge (1000 X g) METHYLENE CHLORIDE LAYER METHYL ALCOHOL WATER LAYER PRECIPITATE extract with 2 X 0.25 mL methylene ch lo r ide i add to residue RESIDUE METHYLENE CHLORIDE LAYER METHYL ALCOHOL LAYER evaporate record weight of fa t DISCARD FAT evaporate, add 7 X 2.0 mL water evaporate a f te r each addi t ion mix with 2.5 mL e thano l , evaporate. RESIDUE add 6 X 20 mL water evaporate RESIDUE Freeze 29°C Freeze Dry WATER EXTRACT WATER EXTRACT MILD OR AGED CHEESE cent r i fuge evaporate f reeze 29°C freeze dry DISCARD RESIDUE UNDEODORIZED FAT RESIDUE Extract with 2.5 mL water RESIDUE Freeze 29°C Freeze Dry WATER.EXTRACT RESIDUE F igure 15 .2 . Flow diagram of preparat ion of cheese f r a c t i o n s . - 79 -to a f i na l volume of 2.0 mL. The samples were f i l t e r e d through 0.45 p m M i l l i p o r e f i l t e r and stored f rozen . 2. NITROGEN DETERMINATIONS A Technicon Auto Analyser (Technicon Instrument Corpora t ion , Tarrytown, New York) was used to analyse nitrogen (N) in the cheese, washed res idue, methylene ch lor ide layer and water extract obtained in the ext rac t ion procedure out l ined in sect ion 1. Non-protein nitrogen (NPN) was determined by cent r i fug ing ca 4g cheese (30,000 X g; 25 min; 25°C) to remove f a t . To 0.1 g of the defatted cheese, t r i c h l o r o a c e t i c ac id (TCA) was added to a f i na l concentrat ion of 12%, centr i fuged (10,000 X g; 10 min) and the supernatant analyzed fo r NPN as above. 3. COMPOSITIONAL ANALYSIS A l l cheese samples were analyzed in dup l i ca te fo r f a t , moisture, to ta l so l i ds and sa l t according to o f f i c i a l A0AC methods (A0AC, 1980). Samples for ana lys is were taken from the ins ide area of the piece of cheese and shredded f i n e l y using a 'home s t y l e ' g ra te r . (a) Fat Butter fat was determined by the Roese-Gottl ieb method. (b) Moisture Described in Part 1-2 of Mater ia ls and Methods. (c) Sa l t Described in Part 1-2 of Mater ia ls and Methods. - 80 -(d) PH pH was measured at ambient temperature (20-22°C) using a hydrogen s p e c i f i c ion sens i t i ve e lectrode purchased from F isher S c i e n t i f i c (Fa i r Lawn, NJ). Compositional parameters inc lud ing moisture in the non-fatty substance (MNFS), sa l t- in-moisture (S/M), and fat in dry matter (FDM) were ca l cu la ted (Lawrence and G i l l e s , 1982). 4. SENSORY EVALUATION A l l cheese samples were o f f i c i a l l y graded by Federal Graders fo r f lavour and physical cha r a c t e r i s t i c s as soon as poss ib le a f te r purchase according to the guide l ines es tab l i shed for the examination of dairy products (Agr i cu l ture Canada, Ottawa, Onta r io ) . F i f teen members from the Department of Food Sc ience, Univers i ty of B r i t i s h Columbia, who l i ked Cheddar cheese were screened fo r t he i r a b i l i t y to d iscr iminate between m i l d , medium and aged Cheddar cheese samples. Ten pane l i s ts were chosen from the f i f t e e n for t he i r s e n s i t i v i t y to Cheddar cheese taste according to the fo l lowing c r i t e r i a : t yp i ca l aged Cheddar t a s t e , b i t t e rnes s , r a n c i d i t y , ac id tendency and s t a l e . T r iang le tes ts were used in order to i den t i f y the odd sample c l a s s i f i e d according to age within each group ( i . e . group 1: m i l d , medium, medium; group 2: medium, medium, aged). Rep l i ca t ion of the evaluat ion on the same sample was conducted to provide an est imation of experimental e r ror and the order of presentat ion was randomized to avoid any pos i t iona l e f f e c t s . Concurrent with the sensory ana l y s i s , two Federal Graders evaluated dup l i ca te samples according to t h e i r standard methods. The graders were asked to c l a s s i f y the samples according to age and a lso to report overa l l grade and any taste or texture - 81 -defec t . Corre la t ions between the t ra ined sensory panel and Federal Graders were ca l cu l a t ed . 5. HPLC ANALYSIS A Spectra-Physics 8100 HPLC and 8400 var iab le wavelength detector (Spectra-Physics, Santa C l a r a , CA) operated at a wavelength of 220 nm were used for the ana l y s i s . One modi f i ca t ion to the equipment was a gas de-bubbler (Terochem Labora tor ies , Edmonton, Alberta ) and i n s t a l l e d between the ternary pump and the column. As described by Hearn et aj_. (1982 ) and es tab l i shed by Pham and Nakai (1984) a reversed-phase column (250 X 4.6 mm I.D.) packed with Adsorbosphere C3 (5 urn) (Applied Sciences Labora tor ies , State Co l l ege , PA) was used for a l l chromatographic runs. The volume of the sample loop was 50 U L . A ternary gradient system was used to e lute the non-volat i le tas te components from the column. The i n i t i a l solvent volume r a t i o was 44 .6 :0 .0 :55 .4 fo r t r i f l u r o a c e t i c acid (0.1%), a c e t o n i t r i l e and water. Over 56.6 min the r a t i o was changed to 0 .0 :36 .6 :63 .4 . A l l chromatographic runs were performed at ambient temperature (20-22°C) at a flow rate of 0.97 mL/min. Doubly d i s t i l l e d deionized water f i l t e r e d through a M i l l i -Q system ( M i l l i p o r e , Bedford, MA) was used to prepare the t r i f l u r o a c e t i c a c i d . A c e t o n i t r i l e and a l l other sample so lut ions were f i l t e r e d through a 0.45 pm M i l l i p o r e f i l t e r . P r io r to use, a l l so lut ions and water were evacuated for about 15 min and fur ther degassed for an addi t iona l 10 min with hel ium. During gradient e lu t ion a f ine steady stream of helium was allowed to cont inua l l y degas the so lu t i ons . A Spectra-Physics SP 4100 computing in tegra tor was used to ca l cu l a te - 82 -peak areas. Under BASIC c o n t r o l , the SP 4100 in tegra tor contro ls the pumping system and the var iab le wavelength detector . In addi t ion to computing in tegra t ion func t ions , SP 4100 i s capable of graphical presentat ion in X-Y axes. A f u l l alphanumeric keyboard and a LED d isp lay f a c i l i t a t e s entry , review and ed i t ing of a l l f i l e and run data as well as operating status and system d iagnos t i c s . 6. OPTIMIZATION TECHNIQUES (a) Super Modif ied Simplex A Super Modif ied Simplex (SMS) proposed by Routh et a l . (1977) and wr i t ten for a Monroe 1880 c a l cu l a t o r (Monroe, The Ca lcu la to r Company, Orange, NJ) was used to f ind optimum condi t ions for HPLC r e so lu t i on . The method is a modi f i ca t ion of a simplex opt imizat ion (Nelder and Mead, 1965) incorporat ing a quadrat ic curve f i t t i n g to determine the vertex immediately fo l lowing the r e f l e c t i on vertex to replace the worst vertex in the preceding simplex. To measure the qua l i t y of the HPLC peak separat ion as well as c r i t e r i a to compare one mobile phase with another the method of Glajch et aj_. (1980) was used where P-j i s a measure of the peak separat ion of the i t h pa i r of peaks in a system with k tota l pa i rs of i n t e r e s t . Peak separa t ion , P-j i s simply de f ined : where f i s the depth of the va l ley below a s t ra igh t l i n e connecting two adjacent peak maxima and g i s the height of the s t ra igh t l i ne above the basel ine at the v a l l e y , as i l l u s t r a t e d in Figure 16.2. The SP^ was used as the response for the Super Modif ied Simplex. - 83 -F igure 16.2 Schematic o u t l i n i n g method used to c a l c u l a t e peak r e s o l u t i o n , adapted from Glajch et a K (1980). - 84 -(b) Mapping Super Simplex and Simultaneous Factor S h i f t . In order to speed up the i t e r a t i v e opt imizat ion procedure and g raph i -c a l l y i l l u s t r a t e the experimental response surface the Mapping Super-Simplex introduced by Nakai et al_. (1984) and writ ten for an Amdahl 470 V/8 computer was used. For each f a c to r , the level values used in the opt imiza -t ion are d iv ided into four groups based on t he i r locat ions on a scale within l a rge , medium and small l i m i t s . The large and small l im i t s were determined from ind iv idua l p lots of response value ( P )^ vs. each fac tor level ( i n i t i a l and f ina l concentrat ion of t r i f l u o r a c e t i c ac id and ace ton i -t r i l e , and t ime) . The medium l im i t is an average of both large and small l i m i t s . These l im i t s were used for grouping the data . Data points for one fac tor which belongs to the same groups of other fac tors were jo ined together thus g iv ing an estimate of the response sur face . The maps for a l l fac tors provided target level values where the optimum mobile phase condit ions were loca ted . A Simultaneous Factor Sh i f t Program (Nakai et aj_., 1984) wr i t ten for a Monroe 1880 programmable ca l cu l a to r was used. Target values (estimated best mobile phase condi t ions ) were determined from the graphs. The program is designed to s h i f t a l l f ac tor l eve ls obtained from the mapped graphs simultaneously one f i f t h the distance between the present best value and the target va lue. The new experimental condit ions ( ver t i ces ) resu l t ing from the Simultaneous Factor Sh i f t Program were invest igated and t he i r response values ca lcu la ted as described prev ious ly . 7. STATISTICAL ANALYSIS S t a t i s t i c a l pattern recogni t ion techniques (p r inc ipa l component ana lys i s (PCA), stepwise l i nea r d iscr iminant ana lys is (SLDA) and regression - 85 -on p r i n c i p a l components) were used to i n t e r p r e t the HPLC p r o f i l e s . Programs employed were BMD:4M, BMD:7M and BMD:4R r e s p e c t i v e l y , (BMDP S t a t i s t i c a l Software I n c . , Un ivers i ty of C a l i f o r n i a ) . Computer analyses were performed on an Amdahl 470 V/8 computer. Pattern recogn i t ion i s a method for analyzing m u l t i v a r i a t e data . For each sample analyzed, a set of r e s u l t s c o n s t i t u t i n g a pattern i s obta ined. S i m i l a r samples w i l l therefore have s i m i l a r pat terns . These patterns can be used to i d e n t i f y groups of s i m i l a r objects or they can be used to charac ter i ze a group of s i m i l a r objects in order to enable one to c l a s s i f y a new object in the group i t belongs to based on r e g u l a r i t i e s in the data . The c h a r a c t e r i s t i c feature of m u l t i v a r i a t e ana lys i s i s the cons iderat ion of a set of n ob jec ts , on which are observed the values of p v a r i a b l e s . Each cheese sample was considered as a point in a p-dimensional space (p being the number of peaks from the HPLC p r o f i l e ) , a) P r i n c i p a l Component Analys is P r i n c i p a l component a n a l y s i s , a d i r e c t pattern recogni t ion technique, was appl ied to the HPLC data obtained for each cheese sample. A l l peak areas were entered into the ana lys i s to s e l e c t the most important var iab les (peak) for d i s c r i m i n a t i o n among the cheese samples. Each var iab le (peak) has a variance (a measure of d i spers ion of values around the mean) and usual ly the v a r i a b l e s are assoc iated with each other i . e . covariance between p a i r s of v a r i a b l e s . The data set as a whole has a. tota l var iance which i s the sum of the ind iv idua l var iances . PCA transforms the data to descr ibe the same amount of v a r i a b i l i t y ( tota l var iance) with the same number of axes, where each v a r i a b l e measured i s an axis of v a r i a b i l i t y . The f i r s t axis accounts for as much of the tota l var iance as p o s s i b l e . The second axis accounts fo r as much of the remaining var iance as p o s s i b l e , - 86 -while being uncorrelated with the f i r s t ax i s . This continues un t i l a l l var iab les have been accounted for (Daul try , 1976). This resu l ts in a few large axes accounting for most of the tota l var iance , and a la rger number of small axes accounting for very small amounts of the to ta l var iance. These small axes are normally discounted from fur ther cons ide ra t ion . The o r ig ina l set of cor re la ted var iab les (HPLC peak areas) were transformed into a smaller set of uncorrelated p r inc ipa l components. The interdependence of the var iab les was invest igated by the co r r e l a t i on c o e f f i c i e n t matrix (pos i t i ve one, perfect pos i t i ve c o r r e l a t i o n ; zero , no c o r r e l a t i o n ; negative one, perfect negative c o r r e l a t i o n ) . E igen-values greater than one were considered s i g n i f i c a n t in descr ib ing tota l variance and used for the basis for the se l ec t ion of the components. Factor loadings from the s i g n i f i c a n t p r inc ipa l components were p lot ted to give a two dimensional representat ion of the data, b) Stepwise L inear Discr iminant Ana lys is The object ive of the ana lys is is to weigh and combine l i n e a r l y the d i sc r imina t ing var iables so that the groups are forced to be as d i s t i n c t as poss ib le and the number of var iab les are reduced with minimum loss of d i f f e r e n t i a t i o n between the groups. Some l i nea r combinations provide better separat ions than others and therefore the weight attached to each var iab le must be optimized so that o r i en ta t ion w i l l be used, which w i l l provide maximum d i f f e r e n t i a t i o n between the groups. The optimal weight c o e f f i c i e n t s are obtained by maximization of the r a t i o of between-groups variance of the samples to t h e i r within-group variances and the co r re l a t i on between the var iab les must be as small as poss ib le (Coomans et al_. , 1979). The cheese samples were c l a s s i f i e d into correct taste categor ies by the - 87 -Federal Graders and t ra ined sensory panel and used for the development of the d iscr iminant and c l a s s i f i c a t i o n funct ions ( learning s e t ) . Because of the d i f f i c u l t y in v i s u a l i z i n g the mul t i va r ia te data (p, peaks) obtained from the HPLC pattern for each cheese sample, SLDA was used to reduce the p-variables in a p-dimensional space to two. The c l a s s i f i c a t i o n of the groups i s represented graph ica l l y in a two-dimensional space (X and Y-axis) by a plot of the f i r s t canonical var iab le against the second canonical v a r i ab l e . The f i r s t canonical var iab le p lot ted on the X-axis i s a l i nea r combination of peaks that best d iscr iminates among the groups. The second canonical v a r i ab l e , which i s the next best l i nea r combination orthogonal to the f i r s t one, i s p lot ted on the Y-axis . Therefore , each cheese sample i s character ized by the co-ordinates for the canonical v a r i ab l e s . In addi t ion to canonical var iab les used for graphic representat ion of the data , SLDA y i e l d s a new set of var iab les which are l i nea r combinations of the o r ig ina l parameters. These d iscr iminant funct ions enable c l a s s i f i c a t i o n of unknown samples. The d iscr iminant funct ions can be represented by the equat ion: Z = a^x^ + a 2 x 2 * , , ^ i x i » where a^ . . . . a^  are c o e f f i c i e n t s of the d iscr iminant funct ion for each group i . e . young, m i l d , medium, o ld and X-aged. xi-..Xj are the peak areas 1 to i for the unknown sample. Z i s ca l cu la ted in order to maximize the r a t i o of between-group va r i a t ion to within-group v a r i a t i o n . Z can be ca l cu la ted by subs t i t u t i ng x j . . . x ^ by t h e i r respect ive peak areas fo r the unknown cheese sample obtained from the HPLC pa t te rn . The d iscr iminant funct ions provides a maximum d i f f e r e n t i a t i o n between groups of cheese. The unknown sample i s then assigned to the group with the l a rges t d iscr iminant score . - 88 -8. AMINO ACID ANALYSIS The water-soluble extracts (sect ion 1) from a young, m i l d , medium, o ld and X-aged cheese were freeze dr ied in a Labconco Model 75018 freeze dryer (Labconco, Kansas C i t y , MO). 2.5 mg dry prote in sample (ca lcu la ted by N determinat ion, sect ion 2) was added to 1.0 mL d iges t ion mixture of 3-(2-aminoethyl) indole and p-toluene su l f on i c ac id (Simpson et a l_ . , 1976) in P ierce vacuum hydro lys is tubes (P ie rce , Rockford, IL) . Fol lowing evacuation and nitrogen f lush ing (three times) the contents of the hydro lys is tube were hydrolyzed at 110°C for 24 h. A f te r hyd ro l y s i s , 2 mL IN NaOH was added to the digested mixture then t rans fe r red to a 4 mL volumetric f l a s k , pH adjusted to 2.1 and made to volume with H2O. The sample was f i l t e r e d through a 0.22 urn M i l l i po r e f i l t e r (M i l l i po re Co rp . , Bedford, MA). A Beckman System 6300 high performance amino ac id analyzer (Beckman Instruments Inc . , Palo A l t o . , CA) was used for ana l y s i s . The volume of the sample loop was 100 pL de l i v e r i ng 50 uL to the column. A Hewlett Packard 3390A Reporting Integrator (Hewlett Packard, Avondale, PA) was used to ca l cu la te peak areas of the p a r t i c u l a r amino acid and compared to that of a standard. 9. ANALYSIS OF HPLC FRACTIONS Peaks se lected to be most e f f e c t i v e for d i sc r imina t ion by SLDA were f rac t ionated by in j ec t ing an aged sample into the HPLC using the same gradient e l u t i o n . Af ter repeated in j ec t ions of the sample the ind iv idua l f r ac t ions were concentrated and in jec ted to determine homogeneity. Amino ac id analyses , as out l ined in Sect ion 8, were ca r r i ed out on the s p e c i f i c f r ac t i ons c o l l e c t e d . - 89 -RESULTS AND DISCUSSION More than 60 samples of cheese were used in the study and purchased from r e t a i l out le ts representing d i f f e r en t manufacturers across Canada. A group of 'young' cheeses, one week o l d , made on continuous cheddaring equipment were obtained from a loca l da i r y . SAMPLE PREPARATION The method of McGugan et £l_. (1979) used to extract the non-vo la t i le water extractable f lavour components conta in ing the s a l t s , amino acids and pept ides . The i r method was selected because of the cont r ibut ion of the non-volat i le components to Cheddar cheese f l a vour . Sensory evaluat ion of the non-volat i le water extracts from a mild and aged cheese were co r rec t l y i d e n t i f i e d . Harwalkar and E l l i o t (1970) a lso suggested that the nonvo la t i l e components contr ibuted to Cheddar f l a vour . Although they were in terested in the b i t t e r and astr ingent f r a c t i o n s , t h e i r method of f r a c t i ona t i on was e s sen t i a l l y the same as that of McGugan et a]_. (1979). The methylene ch l o r i de : methyl a l coho l : water ext rac t ion method of the l a t t e r workers was bas i ca l l y the only d i f fe rence compared to chloroform: methyl a l coho l : water ext rac t ion procedure of the former. NITROGEN DETERMINATIONS Pham and Nakai (1984) showed these non-volat i le water f r ac t ions to have a pos i t i ve react ion with the ninhydrin reagent and suggested they were prote in degradation products from cheese r ipen ing . Total nitrogen was determined on freeze dr ied samples of a mild and aged cheese. As w e l l , - 90 -tota l nitrogen was determined for f rac t ions co l l ec ted and out l ined in sect ion 2. The resu l t s (Table 9.2) ind icated that N was not detected in the methylene ch lor ide fat f r a c t i on suggesting that the r epe t i t i v e methyl alcohol water extract ions removed any residual water so luble N components from the f a t . McGugan et al_. (1979) showed that the weight recovery of the water-soluble extract from an aged cheese was two times that of a m i ld . The f ind ings of the present study confirmed these resu l ts that as the cheese ages, prote in degradation products accumulate in the water-soluble f r a c t i o n . Table 9.2 shows that aged cheese has a higher percentage of so lub le N compared to mild cheese (21.0 and 13.6% r e spec t i v e l y ) . Soluble N has been used as a method to study p ro teo l ys i s during cheese r ipening and was found to increase with age (Park et a K » 1978). Various ext rac t ion procedures have been compared (Kuchroo and Fox, 1982; Kuchroo and Fox, 1983) however despite the extensive use of p ro teo lys i s products as r ipening indices l i t t l e work has been done to re late these to f lavour development. COMPOSITIONAL ANALYSIS A l l cheeses were analyzed for moisture, f a t , sa l t and pH. Lawrence and G i l l e s (1982) showed that the qua l i t y of cheese was inf luenced most by moisture in the non-fat-substance (MNFS), fat- in-dry matter (FDM), sa l t- in-moisture (S/M) leve ls and pH. Therefore , only cheeses with the optimum s p e c i f i c a t i o n s : MNFS, 52-56%; FDM, 52-55%; S/M, 4.0-6.0%; and pH 4.95-5.10 were selected for the i n i t i a l study. Cheeses that did not f a l l in to these s p e c i f i c a t i o n l im i t s were general ly downgraded for f lavour and/or texture by the Federal graders. - 91 -Table 9 . 2 . Nitrogen (N) d i s t r i b u t i o n of the various f rac t ions from mild and aged cheese. % Nitrogen % Prote in* (%N X 6.25) Sampl e Mi ld Aged Mi ld Aged Cheese 3.87 3.75 24.19 23.44 Washed residue 3.31 2.92 20.68 18.25 Water soluble f r a c t i on 0.53 0.79 3.31 4.94 Methylene ch lo r ide f r a c t i on 0.00 0.00 0.00 0.00 1 wet bas is - 92 -SENSORY EVALUATION Lawrence and G i l l e s (1980) demonstrated how small increases in MNFS leads to f ree moisture in the cheese. Without exception the t ra ined cheese graders had no d i f f i c u t l y in d i s t i ngu i sh ing textura l d i f f e rences in cheeses of d i f f e r en t moisture contents . No s i g n i f i c a n t d i f f e rence (P > 0.05) was found fo r cheese f lavour assessment fo r the more than 60 samples of cheese between the subject ive sensory eva luat ion by the t ra ined panel to that of the Federal graders. Sensory ana lys is for the 15 pane l i s t s to determine the odd mild sample from two medium was s i g n i f i c a n t (P < 0.05). Randomizing the same samples resu l ted in no s i g n i f i c a n t d i f f e rence (P > 0.05) . C lea r l y the dynamic nature of cheese f lavour i s d i f f i c u l t to assess p a r t i c u l a r l y in the mild to medium ca tegor ies . In fac t Le l i ev re and G i l l e s (1982) showed that sensory eva luat ion of p a r t i c u l a r l y young cheese var ied from grader to grader. McBride and Hall (1979) recognized that the t r a d i t i o n a l procedures for assessing cheese f lavour qua l i t y at an ear l y age are not a r e l i a b l e guide to i t s a c cep t ab i l i t y at matur i ty . S i gn i f i c an t d i f f e rences (P < 0.01) between medium and aged cheeses were found in t h i s study suggesting that the f lavour components respons ib le are more pronounced and poss ib l y eas ier to d i s t i n g u i s h . This i s substant iated by the fact that cheese f lavour and texture are obtained only a f ter a r ipen ing period (Adda et al_., 1982) and the length of time i s c h a r a c t e r i s t i c fo r the pa r t i cu l a r cheese. Notwithstanding, the spectrum of compounds i s often so wide and complex that those involved in the f lavour remain unknown (Law, 1981). HPLC ANALYSIS An Adsorbosphere C3 reversed-phase column was chosen s ince i t provided - 93 -the best reso lu t ion and greatest number of peaks using i so c r a t i c separat ion of water extractable f lavour components (Pham and Nakai , 1984). Moreover, i t i s genera l l y recognized that RP/HPLC permits the h ighly se l ec t i ve separat ion of amino ac ids , pept ides, polypeptides and prote ins (Hearn et al_., 1982). F igures 17.2-21.2 represent HPLC p r o f i l e s for young, m i l d , medium, o ld and X-aged cheese respec t i ve l y . From v isual examination, the p ro f i l e s are complex with more than 48 peaks. Peaks with s im i l a r re tent ion times were considered to be the same. Repeated i n j e c t i on (10 times) of the water extract from a X-aged sample, chosen because of the greatest number of peaks, and in tegra t ion of the peak areas at the respect ive retent ion times confirmed the r e p r o d u c i b i l i t y of the technique. The ca l cu l a t ion of average and standard dev iat ion for HPLC peak reso lu t ion resulted in reproducible separat ion (27.5 ± 0.23) . The va r i a t ion in column performance was minimized by repeated washing with f i r s t , organic solvent followed by water. This procedure ensured complete e lu t ion of any s t rongly bound hydrophobic compounds not eluted during the i n i t i a l run. Regeneration of the column to the same s t a r t i ng working pressure (ca 600 ps i ) using the s t a r t i ng mobile phase suggested complete regenerat ion. Indeed, i t is well recognized that in RP/HPLC, as the po l a r i t y of the substance increases , the weaker is i t s i n te rac t ion with the hydrophobic bounded s ta t ionary phase. Peaks o, q, f f , mm and nn increased with age while peak f remained r e l a t i v e l y constant in comparison from v isual examination. S i gn i f i c an t va r i a t ion in other peaks was d i f f i c u l t to in te rpre t v i s u a l l y because of the complexity of the chromatograms. Pham and Nakai (1984) using i so c r a t i c e lu t ion f rac t iona ted 13 peaks from the water soluble f r ac t ions of m i l d , medium, old and X-aged cheese and suggested changes in compositon with - 94 -O F igure 17.2 RP/HPLC separat ion of the water-soluble f r a c t i o n from a young cheese. - 95 -0 inject 10 20 30 Time (mins) F igure 18.2 RP/HPLC separat ion of the water-soluble f r a c t i o n from mi ld cheese. - 96 --Figure 19.2 RP/HPLC separat ion of the water-soluble f r a c t i o n from a medium cheese. - 97 -' F igure 20.2 RP/HPLC separat ion of the water-soluble f r a c t i o n from an o ld cheese. - 98 -F igure 21.2 RP/HPLC separat ion of the water-soluble f r a c t i o n from a X-aged cheese. - 99 -r ipen ing . Genera l l y , the older the cheese the greater the concentrat ion of more hydrophobic compounds (Figures 20.2 and 21.2 , peaks mm and nn). Champion and Stanley (1982) f rac t ionated b i t t e r peptides from Cheddar cheese and found that f rac t ions with higher hydrophobicity were b i t t e r . Although i t can be assumed that these strongly retained f rac t ions represent b i t t e r peptides in the aged cheeses, t he i r concentrat ion was not enough to e l i c i t b i t terness and subsequent detect ion by sensory eva lua t ion . The HPLC p r o f i l e for a t yp i ca l b i t t e r X-aged cheese sub jec t i ve ly evaluated is shown in Figure 22.2 . The pattern i s e s sen t i a l l y ident i ca l to the non-bitter X-aged sample (Figure 21 .2 ) . However, at the higher retent ion t ime, there was an increased number of peaks with a s i g n i f i c a n t peak (rr ) appearing at about 45 min. These strongly bound hydrophobic compounds, e luted only at the high concentrat ions of a c e t o n i t r i l e , are assumed to be responsible for the b i t te rness detected in the cheese. Furthermore, the most charac-t e r i s t i c d i f fe rence between the HPLC p r o f i l e s of b i t t e r cheese and non-b i t t e r were these strongly retained peaks suggesting that RP/HPLC with the present mobile phase condit ions permit separat ion of b i t t e r peptides from the water soluble ex t rac t . OPTIMIZATION a) Super Modif ied Simplex Di f ferences in e lu t ion cha ra c t e r i s t i c s were obtained with d i f f e r en t solvent systems. Mobile phases containing i sopropano l , or propanol resu l ted in components e lu t ing qu ick ly with poor separa t ion . This was due to the higher solvent strength and s o l u b i l i z a t i o n of the pept ides . The use of a c e t o n i t r i l e as an organic modif ier in sp i te of i t s low v i s cos i t y and UV-absorption combined with 0.1% t r i f 1uo roace t i c acid (TFA, pH 2.1) s i g n i f i c a n t l y increased the e f f i c i e n c y of separat ion . When TFA was used, - 100 -F igure 22.2 RP/HPLC separat ion of the water-soluble f r a c t i o n from a b i t t e r cheese. - 101 -the carryover of hydrophobic peptides was great ly reduced. Acharya et a l . (1983) in t he i r study on the t r yp t i c peptides of hemoglobin suggested that TFA serves as an ion-pai r ing reagent and inf luences the re tent ion times of h i s t id ine-conta in ing pept ides. Therefore , the basic groups in the presence of TFA makes the peptides r e l a t i v e l y more hydrophobic. Hearn et al_. (1979) invest igated the e f f e c t of pH and ion-pai r formation on the re tent ion of pept ides. At the low pH, hydrophobic anionic reagents r e su l t in increased re tent ion of the peptide samples whereas hydrophobic ca t i on i c reagents caused decreased re ten t ion . Thus retent ion was explained on the basis of e i the r ion-pa i r ing or ion-exchange in te rac t ions of the reagent with the protonated pept ide. The capacity r a t i o s , k, of polypeptides are general ly very sens i t i ve to the concentrat ion of the organic mod i f i e r . However, only a small d i f ference in s e l e c t i v i t y can be achieved by changing the organic solvents (Gazdag and Szepes i , 1981). Therefore , methanol and a c e t o n i t r i l e are general ly accepted solvents in the HPLC ana lys is of po lypept ides . The pH s t a b i l i t y of the Adsorbosphere Cs column (pH 2-8) and the p rec ip i t a t i on of the water-soluble extracts at the higher concentrat ion of a c e t o n i t r i l e were used in developing const ra in ts for the opt imiza t ion . Levels chosen were: a c e t o n i t r i l e , 0-40%; TFA, 0-100%; flow r a t e , 0.5-2.0 mL/min; t ime, 30-60 min. General ly peptides were e luted with in one hour at flow rates of approximately 1.0 mL/min. Temperature was not a factor in that Wilson et al_. (1981) studied the inf luence of various experimental parameters on the behaviour of peptides on RP/HPLC and showed peptide re tent ion to be temperature-independent between 20 and 55 °C . A l l runs for the present study were ca r r i ed out at room temperature (21-23 °C ) . - 102 -An ult imate goal of most s c i e n t i f i c research and inves t iga t ion is to f ind the optimum condit ions for achieving the object ives (Nakai, 1981). Therefore , the super modified simplex opt imizat ion introduced by Routh et a l . (1977) with the experimental const ra in ts prev ious ly discussed was employed. The basic p r i n c i p l e of the simplex opt imizat ion is to move away from the experimental condit ions (Vertex) which has y i e lded the worst resu l t in each simplex cons i s t ing of (n + 1) experiments, where n i s the number of f a c t o r s . Algorithms to f ind the next experimental condit ions which move the experiment in the opposite d i r e c t i on from the worst resu l t in the previous simplex can be programmed into a computer (Nakai et , 1984). The advantages of computer ass i s ted opt imizat ion techniques are many: improved product i v i t y by fewer experimental runs; mu l t i f ac to rs inc luding mutual in te rac t ions can be invest igated without missing the true optimum, and poss ib ly most important the optimum experimental factors can be found without re l y ing on human in te r ven t ion . The experimental condi t ions (Vertex) were generated by computer and a f te r each experiment the peak reso lu t ion of the HPLC p r o f i l e was ca l cu la ted as descr ibed to provide the response. Each response was subsequently used in the algorithm to automat ica l ly d i r ec t the condi t ions to the optimum. Figure 23.2 shows the response values of the pattern search. C lea r l y the i n i t i a l mobile phases, experiments 1 to 5 (S tar t ing Simplex) resul ted in poor peak r e so lu t i on . As the opt imizat ion cont inued, reso lut ion improved with the best condi t ion obtained for experiment no. 9. It i s a general trend in i t e r a t i v e opt imizat ion procedures that the speed of approaching the optimum is quickest at the beginning and progress ive ly slows down therea f te r (Nakai et a l . , 1984). This s i tua t i on is true for mathematical models, however, in - 103 -30 20 -10 -0 I I I I l i I I i l I l I I I I I I 0 2 4 6 8 10 12 14 16 Experiment No. F igure 23.2 Opt imizat ion of the mobile phase to increase peak reso lu t ion of the water-soluble extract from a X-aged cheese sample. An Adsorbosphere Cs column was used for a l l chromatographic runs. \ - 104 -real l i f e s i tua t ions converging at the optimum may not happen as shown in F igure 23.2 . It i s postulated that poor response values obtained a f t e r experiment 9 were due to the fact that some important fac tor may have been overlooked in the opt imiza t ion . To ver i f y that the mobile phase condi t ions of experiment no. 9 did in fact e l i c i t the true optimum, repeated in j ec t i ons (5X) using the optimized condi t ions were ca r r i ed out. Results obtained confirmed th i s as the true optimum with an average value of 106 and standard dev iat ion of 1.58. It is i n te res t ing to note that experiment no. 12 resulted in a response s im i l a r to that of no. 9. Comparison of the mobile phase condit ions for experiments no. 9 and 12 showed that the only s i g n i f i c a n t d i f fe rence was 100% t r i f l u r o a c e t i c acid (TFA) fo r the l a t t e r and 44.6% for the former. It i s reasonable to speculate that the increased concentrat ion of TFA exceeded the amount required to make the peptides hydrophobic through ion-pa i r ing and concentrat ions greater than 44.6% had no e f fec t on improving r e so lu t i on . b) Mapping Super Simplex (MSS) This technique allows the graphic representat ion in two dimensions of the response surface of the simplex opt imiza t ion . The mobile phase condi t ions or factors were: flow ra te , a c e t o n i t r i l e concent ra t ion , time and TFA concentrat ion and t h e i r const ra in ts se lected as prev ious ly d i scussed . The response value for each experiment in the MSS i s p lot ted against the ind iv idua l fac tor as shown in Figures 24.2-27.2. The data points for one fac tor but a lso belonging to the same group of other factors are re ferred to as matched data po in t s . Therefore j o in ing these points together with l ines estab l i shes the response surface and the d i r e c t i on or loca t ion of the optimum level i s ind icated by where the l i nes converge on the graph. Sometimes the l i nes converge dec i s i v e l y as shown in F igure 26.2 - 105 -F igure 24.2 Mapping r e su l t s of experiments to maximize peak r e s o l u t i o n . Factor 1, flow rate with cons t ra in ts 0.5-2.0 m L / m i n . - • - , group 1 ; - A - , group 2;-D- , group 3 ; - « - , group 4. - 106 -28 ID */> C o a or Factor 2 F igure 25.2 Mapping r e su l t s of experiments to maximize peak r e s o l u t i o n . Factor 2, a c e t o n i t r i l e concentrat ion with cons t ra in ts 10 - 40%. group 1; - A - , group 2; group 3; group 4. - 107 -F a c t o r 3 F igure 26.2 Mapping r e su l t s of experiments to maximize peak r e s o l u t i o n . Factor 3, time with cons t ra in ts 30 - 60 min. - • - , group 1; - • - , group 3; group 4. - 108 -Mapping r e su l t s of experiments to maximize peak r e s o l u t i o n . Factor 4, TFA concentrat ion 0 - l 0 0 % > group 1 ; - A - , group 2; group 3; group 4. - 109 -where the optimum value for time i s c l e a r l y Vertex 9 represent ing 56.6 min. In F igures 24.2 and 25.2, the optimum flow rate and a c e t o n i t r i l e concentrat ion respec t i ve l y are not as c l ea r and may be located between Vertex 9 and 12. TFA concentrat ion (F igure 27.2) was the l eas t conc lus ive but the trend curves are toward the d i r e c t i on of Vertex 9. c) Simultaneous Factor S h i f t Results of the simultaneous f ac to r s h i f t experiments which invest igates condi t ions around the optimum l e v e l s , in a l l cases produced response values l ess than that obtained for Vertex 9. The simultaneous s h i f t fac tor program can be considered a ' f i n e tun ing ' to ensure that the true optimum i s not missed. Indeed in the present study, the optimum mobile phase condi t ions were found to be Vertex 9 and these cond i t ions resu l ted in the best HPLC p r o f i l e s . PRINCIPAL COMPONENT ANALYSIS (PCA) The s ixty HPLC chromatograms, each one represent ing a s p e c i f i c cheese sample for which each peak was converted to percent area, were analyzed by PCA. The in tent of PCA was to se lec t from each HPLC p r o f i l e patterns (peaks) c h a r a c t e r i s t i c of the cheeses. Thus on the bas is of these patterns the cheeses could be c l a s s i f i e d in to groups. The HPLC data, s p e c i f i c a l l y peak areas, were transformed to descr ibe the same amount of v a r i a b i l i t y in order to extract s i g n i f i c a n t factors for cheese f lavour contained in the 48 peaks. Eleven factors had eigenvalues l a rger than 1.0 as shown in Table 10.2. In other words, the eigenvalue describes the magnitude of variance of each pr inc ipa l component which accounts for the total var iance. The la rger the e igenvalue, the greater the cont r ibut ion of the pr inc ipa l component fo r c l a s s i f i c a t i o n . From Table 10.2 the cumulative proport ion of - 110 -Table 10.2 Eigenvalue (VP), sum of VP, and cumulative proport ion in tota l variance [%) in a pr inc ipa l component ana l ys i s . P r inc ipa l Eigen Sum of VP Cumulative Component Value Proport ion 1 17.39 17.39 36.2 2 5.54 22.93 47.8 3 5.27 28.20 54.6 4 3.04 31.24 60.9 5 2.47 33.71 66.0 6 2.38 36.09 71.0 7 2.09 38.18 75.4 8 1.80 39.98 79.1 9 1.60 41.58 82.5 10 1.32 42.90 85.2 11 1.19 44.09 87.7 12 0.94 45.03 89.7 13 0.82 45.85 91.4 - I l l -to ta l variance up to the eleventh p r inc ipa l component was 87.7%. There-f o r e , to ta l variance contained in the 48 peaks could be contracted in 11 p r inc ipa l components with 12.3% loss of var iance . Factor loading for the major f i v e p r inc ipa l components and communality are shown in Appendix I. Peaks with large fac tor loadings ind icate c lose assoc ia t ion with the p r inc ipa l components. As shown, peaks 1, 19, 28, 36, 39 and 41 had the la rgest fac tor loadings for the f i r s t p r inc ipa l component whereas peaks 3, 8, 9 and 33 were largest for the second p r inc ipa l component. Visual examination of these most d i sc r iminat ing peaks on the HPLC chromatograms were d i f f i c u l t to i n t e rp r e t . Moreover, the intent of p r inc ipa l component ana lys is as a pattern recogni t ion technique is to g raph ica l l y i l l u s t r a t e the data. However, when th i s was attempted, d i s t i n c t age c l a s s i f i c a t i o n into groups based on the p r inc ipa l components was not obtained. Figure 28.2 , a p lot of Factor loading 1 vs. Factor loading 2 did not separate the cheese samples. Addit ional p lots of Factor 1 vs . Factor 3 and Factor 2 vs . Factor 3 although not shown a lso resul ted in poor separat ion of the cheeses. A poss ib le explanation i s that the p r inc ipa l components are c l a s s i f y i n g the cheeses based on something other than age. Aishima (1979) suggested that the fac tor loadings for each p r inc ipa l component lacked d i sc r im ina t ing information which may have occurred in the present study for age c l a s s i f i c a t i o n . An attempt was made to examine the re l a t ionsh ip between sensory score of the cheese and pr inc ipa l components from the HPLC p r o f i l e s . P lots of the f i r s t and second pr inc ipa l components (Appendix I) vs . sensory score could not separate the cheese samples based on age c l a s s i f i c a t i o n . It i s reasonable to assume that based on the HPLC data prov ided, p r inc ipa l component ana lys is i s extract ing d i sc r imina t ing f a c t o r s , but factors which are not s i g n i f i c a n t for age c l a s s i f i c a t i o n . - 112 -F a c t o r 2 H g u r e 28.2 P r i n c i p a l c o g e n t p , . . * f a c t o r ^ mild - o ; young -- 113 -REGRESSION ON PRINCIPAL COMPONENTS The transformed HPLC data were analyzed by mul t ip le regress ion ana l y s i s , in order to determine which of the p r inc ipa l components was important for age d i s c r im ina t i on . Appendix II shows the co r r e l a t i on between pr inc ipa l components (PC) and the dependent v a r i ab l e . The f i r s t PC with a co r r e l a t i on of -0.79978 best describes the r e l a t i o n s h i p . The index of the PC's being entered show that PCs no 1, 10, 3, 15, 29 and 21 resul ted in a r 2 value of 0.8130. In the regression equation the degrees of freedom for the F-value are p and N - p - 1 where p i s the number of components in the regression (p r inc ipa l components) and N is the number of cases (cheese samples). This i s an overa l l tes t of s i gn i f i c ance for the regression equat ion. The F-value component to enter is a tes t of the s i gn i f i c ance of the c o e f f i c i e n t of th i s component only and the degrees of freedom for th i s F-value are 1 and N - p - 1. C l e a r l y , the degrees of freedom a f te r the f i r s t three PC i s (1, 37) and F = 7.42. Therefore , a f te r the t h i r d PC the F-values are no longer s i g n i f i c a n t . Thus, the PC's 1, 10 and 3 best descr ibe the age of the cheese. The eigenvalues for the fac tor loadings for PC one through ten are shown in Appendix III. The HPLC peaks with high eigenvalue were considered s i g n i f i c a n t in descr ib ing the tota l variance and used for the basis for the se l ec t ion in descr ib ing the c l a s s i f i c a t i o n . However, from the plots i t was evident that c l a s s i f i c a t i o n was not based on age. STEPWISE LINEAR DISCRIMINANT ANALYSIS Using the optimized gradient e lu t i on 48 peaks were separated from the water-extract . The percent area for each peak was ca lcu la ted and the f lavour score , grade and age assessed as prev iously d i scussed . The peak - 114 -areas and age c l a s s i f i c a t i o n were used in the d iscr iminant analys is (BMDP:7M program). F-values of 4.000 and 3.996 to-enter and to-remove r espec t i ve l y were used. The p a r t i c u l a r var iab le (s ) (peak) that adds the most or least to the separat ion of the groups i s entered into the d iscr iminant func t ion . The va r i ab le (peak) with the highest F-to-ehter i s the var iab le that d isc r iminates best between groups and was found to be the 32nd peak l abe l l ed ( f f ) in F igures 17.2 to 21.2. The next va r i ab le having the highest F-to-enter was entered and the procedure continued in a stepwise fash ion . From the 48 peaks obtained the s t a t i s t i c a l procedure was able to reduce t h i s number to 8 s i g n i f i c a n t peaks r e su l t i ng in 100% cor rec t c l a s s i f i c a t i o n . The s i g n i f i c a n t peaks in the fo l lowing order of importance were: f f , p, nn, v, q , o, k, e, and f . Peak f f was l a te r removed from the d iscr iminant ana lys is because of a F-value of 3.687. Figure 29.2 shows a canonical p lot of the group means. The f i r s t canonical va r iab le i s the l i nea r combination of var iab les entered that best d iscr iminates among the groups; and as shown, good c l a s s i f i c a t i o n was obta ined. Figure 30.2, a canonical p lo t of each ind iv idua l cheese sample was able to separate the cheese into d i s t i n c t groups. However, the mild and medium category were not c l e a r l y separated. This was not su rpr i s ing in that with the sensory eva luat ion s i g n i f i c a n t d i f f e rences between these two groups was not found (P > 0.05) . The p lot (Figure 30.2) shows that cheese aging i s a complex dynamic process evident by the almost l i nea r progression of the group c l a s s i f i c a t i o n s with X-aged separated from the others . This continuous funct ion along the X-axis was also found by Pham and Nakai (1984). Based on the data prov ided, SLDA ca l cu la tes d iscr iminant funct ions which are the l i nea r combinations of the eight s i g n i f i c a n t peaks and used - 115 -(7)X-Aged o u n g ( 4 > i l d © O l d i i 1 1 — L --2.7 0.0 2.7 5.4 8.1 C a n o n i c a l var iab le 1 Canonical p lot of the group means for the 60 cheese samples. - 116 --8.1 C a n o n i c a l var iab le 1 F iqure 30.2 Canonical p lot of the 60 cheese samples. A, X-aged; 0 9 o l d ; C, medium; M, m i l d ; Y, young. Overlap indTcated by as ter i sk. - 117 -for c l a s s i f i c a t i o n of the groups. The d iscr iminant funct ions were ca l cu la ted as fo l lows fo r the f i v e groups: X-Aged: 1.16(e) - 4.34(f ) + 3.55(k) + 0.16(o) + 5.55(p) + 1.50(q) + 0.20(v) - 1.08(nn) - 62.53 O ld : -0.94(e) - 2.82(f) + 3.98(k) + 0.44(o) - 5.99(p) + 1.71(q) + 5.89(v) - 2.99(nn) - 75.42 Medium: -1.02(e) - 0.13(f) + 0.17(k) + 0.25(o) - 2.93(p) - 0.73(q) + 2.45(v) - 1.53(nn) - 19.56 M i l d : -0.21(e) - 0.92(f) + 0.93(k) + 0.14(o) - 1.81(p) + 0.58(q) + 1.84(v) - 1.04(nn) - 9.25 Young: 0.52(e) - 0.51(f) + 0.12(k) + 0.0(o) + 0.57(p) + 0.07(q) -0.34(v) - 0.17(nn) - 2.81 Four cheese samples which were assessed as o ld by the sensory panel were handled as i f there were unknown samples. By subs t i tu t ing the peak areas (e, f , k, o, p, q , v, and nn) into each of the c l a s s i f i c a t i o n funct ions as shown above y i e l d s a d iscr iminant score . The la rgest d iscr iminant score i s therefore the group to which the unknown samples are assigned. A graphic representat ion i s shown in Figure 31.2 with the cor rec t c l a s s i f i c a t i o n of the unknown samples into the o ld group. Indeed as suggested by Pham and Nakai (1984) an improvement in RP/HPLC reso lu t ion as found in the present study r e su l t s in better d iscr iminant ana l ys i s . The HPLC p r o f i l e (peak areas) of the water-soluble extract of a downgraded acid and f r u i t y cheese sample was subjected to SLDA. The technique c l e a r l y c l a s s i f i e d the samples into the cor rec t age group but was unable to d iscr iminant between the f i r s t grade and the two downgraded cheese samples. The r e su l t was not su rp r i s i ng in that both these defects are not associated with the water-extractable port ion of the cheese. - 118 -5.0 CN 0> A _Q O 'Z 2.5 D > c M Young M Mi ld A A X - A g e d Q A — O u o 0.0 c o U M CC (3) c c x c c c o O A X 0 ° O l d -2.5 c o o o Unknown . i i - 1 • ' i i 1 1 • " . — 6.7 -4.5 -2.2 0.0 2.2 4.5 C a n o n i c a l var iab le 1 6.7 F igure 31.2 Canonical p lot of the cheese samples.showing loca t ion of the unknown samples, X. Overlap ind icated by aster i - 119 -F ru i t i ness was found to be mainly due to the esters ethyl butyrate and ethyl hexanoate (Aston and Dul ley , 1982; Forss , 1979) while a c i d i t y i s determined by pH and l a c t i c a c id . However, when the s t a t i s t i c a l technique was appl ied to b i t t e r cheeses with a t yp i ca l HPLC p r o f i l e as shown in F igure 22.2 two d i s t i n c t c l a s s i f i c a t i o n s were obtained (Figure 32.2 ) . Peak q was the va r i ab le that was found to d isc r iminate best between groups and a new set of d iscr iminant funct ions were ca l cu l a ted . Although good separat ion was obtained between f i r s t grade and downgraded b i t t e r cheese samples the s t a t i s t i c a l techique resu l ted in only 88.1% correct c l a s s i f c a t i o n for the s ix groups. Aishima (1979) in h is study of the aroma qua l i t y of soy sauce using d iscr iminant ana lys is suggested that complete d i sc r im ina t ion of a l l groups on the bas is of two axes on the canonical p lot i s d i f f i c u l t . AMINO ACID ANALYSIS Figure 33.2 shows a t yp i ca l amino acid p r o f i l e fo r the water-soluble extract from a X-aged cheese sample. P r o f i l e s from young, m i l d , medium and o ld a l l had s im i l a r pat terns . The r esu l t s of the amino acid ana lys is (Table 11.2) showed no s i g n i f i c a n t d i f f e rence (P > 0.05) among the cheese samples inc luding the b i t t e r and acid sample. Gooda et _a_K (1983) using the ext rac t ion procedure of Harwalkar and E l l i o t (1970) fol lowed the changes in f ree amino acids and peptides during the r ipening of Cheddar cheese from milk t reated with lactase and found the accumulation of f ree amino acids was proport ional to the r ipening pe r iod . The amount of water-soluble peptides and amino acids increased during p ro teo l ys i s - 120 -12 o o > _ 0 o u c o J-6 12 --0 ° c © * o 0 ° C w C C M M S T M M i i J 1 . i . ._ 1 27.0 -22.5 -18.0 -13.5 -9.0 -4 .5 0.0 4.5 C a n o n i c a l var iab le 1 F igure 32.2 C a n o n i c a l p l o t o f t h e c h e e s e samples show ing l o c a t i o n o f t h e b i t t e r s a m p l e s . A , X - a g e d ; 0, o l d ; C , medium; M, m i l d ; Y, y o u n g ; B, b i t t e r . O v e r l a p i n d i c a t e d by a s t e r i s k . - 121 -R et e nt ion Time F igure 33.2 Amino acid p r o f i l e of water-extract from an X-aged cheese sample. - 122 -Percentage of Total N Amino Acid Young Mi ld Medium Old X-Aged Old (b i t t e r ) X-Aged (acid) ASP 7.0 6.7 5.7 7.1 7.3 7.0 6.9 THR 3.5 3.3 3.1 3.2 2.6 3.0 3.2 SER 4.7 4.9 4.3 4.4 4.3 5.2 4.4 GLU 21.6 23.6 24.6 22.7 22.3 23.3 22.7 PRO 9.4 9.0 8.7 8.8 8.3 8.2 9.2 GLY 1.7 1.5 1.9 1.7 1.9 1.8 1.6 ALA 2.0 2.0 3.3 2.0 1.9 1.9 2.0 CYS 0.0 0.0 0.0 0.0 0.0 0.0 0.0 VAL 5.6 5.8 6.3 5.5 5.5 5.7 5.6 MET 2.0 2.4 2.3 2.1 1.9 2.2 2.0 ILE 4.9 5.0 5.6 4.8 4.7 5.1 4.8 LEU 9.4 9.0 9.6 9.2 10.3 9.4 9.4 TYR 3.2 2.7 0.9 3.1 3.1 2.8 3.5 PHE 6.3 6.6 5.5 6.5 5.9 5.7 6.5 HIS 4.5 5.2 6.0 4.5 6.0 4.5 4.5 LYS 9.4 9.4 10.9 9.6 9.9 9.8 9.2 ARG 4.6 4.0 1.4 4.7 4.2 4.4 4.6 Table 11.2 Amino ac id ana lys is of the to ta l water-soluble extract from various cheeses. - 123 -(Table 9.2) and could be used as an ind i ca t ion of degree of maturity of r ipen ing cheese as has been done by others (Ney, 1971; Nieuwoudt, 1977). Law and Sharpe (1977) considered amino acids to be intermediate products in the production of ce r ta in aroma compounds. Therefore, there appears to be considerable confusion in the l i t e r a t u r e on the ro le of f ree amino acids during r ipening and the i r subsequent cont r ibut ion to Cheddar cheese f l a vou r . Although the amino acid composition of the water-extractable f r a c t i o n was the same for the d i f f e ren t ages of cheese (Table 14.2) the respect ive HPLC p r o f i l e s were cha r a c t e r i s t i c fo r each age category suggesting the importance of the pept ides. ANALYSIS OF THE HPLC FRACTIONS Fract ions f, o, f f and r r from a X-aged b i t t e r sample (Figure 22.2) were co l l e c ted af ter repeated i n j e c t i o n s : Homogeneity of each f r a c t i o n a f te r HPLC analys is indicated freedom from other i n t e r f e r i ng peaks. Further work is required on the amino acid composition of these f r a c t i ons thereby r e l a t i ng peptide hydrophobici ty and subsequent retent ion on the HPLC column. Moreover, the importance of the s t a t i s t i c a l pattern recogn i t ion techniques in se l ec t ing s i g n i f i c a n t peaks cha rac t e r i s t i c of Cheddar cheese taste and in pa r t i cu l a r b i t terness should be assessed. - 124 -CONCLUSIONS Cheese aging i s complex and the agents responsible for Cheddar cheese f lavour are s t i l l l a rge ly unknown. However, the non-volat i le water extractable f r ac t i on has been shown to be important for f lavour i n t e n s i t y . As the cheese ages prote in degradation products accumulate in the water-soluble f r a c t i o n . The ext rac t ion procedure removed soluble N components from the methylene ch lo r ide fa t f r a c t i o n . No s i g n i f i c a n t d i f fe rence was found for cheese f lavour assessment between the t ra ined sensory panel and the Federal graders. Ana lys is of the water extracts from more than 60 Cheddar cheese samples by RP/HPLC on a C3 Adsorbosphere column using an optimized ternary gradient system y i e lded more than 45 peaks. The systematic opt imizat ion of the mobile phase for s e l e c t i v i t y resul ted in well resolved HPLC p r o f i l e s . The e lu t ion patterns d i f f e red for the various age categories of cheese. Graphic representat ion of the data using mapping super simplex c l e a r l y located the optimum experimental condi t ion and was con-firmed by simultaneous fac to r s h i f t . S t a t i s t i c a l pattern recogni t ion techniques were used to in te rpre t the peaks in the HPLC p r o f i l e s and co r rec t l y c l a s s i f i e d the cheese samples into d i s t i n c t groups. Based on the ca lcu la ted d iscr iminant funct ions unknown samples were co r rec t l y c l a s s i f i e d according to t he i r d iscr iminant score . B i t t e r cheese samples were c l e a r l y i d e n t i f i e d by t he i r t yp i ca l HPLC p r o f i l e showing strongly adsorbed hydrophobic peptides e lu t ing only at the high concentrat ion of organic so lvent . Amino ac id ana lys is of the water extract for the d i f f e r en t ca te -gories of cheeses showed no re l a t ionsh ip between the to ta l amino acids and the i n tens i t y of Cheddar cheese f l avour . - 125 -Further work i s current ly underway to determine the d i s t r i b u t i o n of amino acids from peaks d iscr iminated as being important to Cheddar tas te with pa r t i cu l a r emphasis on b i t t e r pept ides . Moreover, the mul t i va r i a te data technique analyzes the en t i r e HPLC p r o f i l e , thus has s i g n i f i c a n t advantages over v isual i n te rp re ta t ion which may, by i t s subject ive nature, overlook important va r i ab l e s . This study proposes a chemical index to measure cheese maturat ion. However, incorporat ion of addi t iona l chemical and physical data into the d iscr iminant analys is would provide complete object ive assessment of overa l l Cheddar cheese qua l i t y thus al lowing in te rp re ta t ion of many d i f f e r en t var iables s imultaneously. - 126 -Appendix 1 Factor loading and communality matr ix. - 127 -Cumulat ive Peak P r inc ipa l Components propor t ion K of total var iance 0. 362430 0.477758 0.545785 0.609057 0 660511 0.710010 0.753719 0.791335 0.824826 0.852286 0.877159 0.896808 0.913942 0.928255 0.942308 0.952797 0.962063 0.968440 0.974073 0 978997 0.982891 0.986122 0.9B8729 0.991084 0.992781 0 994129 0.995265 0.996384 O.997237 0.998070 0.998685 O.999033 O.99933 I O. 999591 0.99978 1 0.999896 0.999955 1 OOOOOO 1 OOOOOO 1 OOOOOO 1 OOOOOO 1.OOOOOO 1.OOOOOO 1.OOOOOO 1 OOOOOO 1.OOOOOO 1.OOOOOO 1 OOOOOO No. 1 2 3 4 5 1 0. 796 -o. 04 4 0. 077 -0.441 -0.016 2 O. 28 1 0 123 0. 702 -0.343 0 132 3 0 152 0 833 O. 243 -0.124 -0.207 4 -0 179 0 278 -0. 264 -0.14 1 -0.514 5 0 170 0 535 O. 366 0. 539 0.033 6 0 214 0. 254 0. 728 0.222 0.063 7 0 076 0 483 0. 299 0.151 -0.199 8 0 089 0 842 -0. 038 0.096 0.046 9 0 048 O 736 0 496 -0 188 -0 073 10 0 221 0 654 -0 133 0.086 -0.094 11 0 088 0 653 0 530 -0.132 0.067 12 0 143 0 344 0 776 0.250 0.061 13 0 191 0 564 0. 355 O.OBB 0.090 14 0 361 -0 027 0 139 0.740 0.029 15 o 504 0 156 0 628 0. 186 0.095 16 0 014 0 518 0 522 -0.003 -0.064 17 0 307 0 323 o 149 0. 171 0.727 18 0 089 -0 113 0 812 0. 144 0.424 19 0 740 0 388 0 224 0.238 0.093 20 0 528 -0 067 0 032 0 146 0.556 21 0 339 0 137 0 482 0.383 0.430 22 0 016 0 232 0 155 0.118 0.062 23 0 064 0 187 0 123 0.090 0. 128 24 0 072 0 180 0 263 -0.028 -0.052 25 0 336 -0 078 0 096 0.6B1 0.117 26 0 04 1 -0 056 0 152 -0.115 0.853 27 0 582 0 148 0 403 0.203 0.4 14 28 o 827 0 295 0 210 0.222 0.068 29 0 477 0 481 0 04 1 -0 028 0.417 30 0 444 0 418 0 249 0.471 0,257 31 0 503 -o 165 o 273 0. 136 0. 354 32 0 487 o 325 0 729 0. 139 0. 153 33 o 004 0 794 0 155 0 097 0. 278 34 0 572 0 637 0 221 0 040 0.009 35 0 678 -0 140 0 163 0 408 O. 1 13 36 0 900 -0 01 1 o 002 0. 319 0.077 37 o 7 lO 0 183 0 338 O. 203 0.424 38 -0 054 o 053 o 028 0.055 0.310 39 0 828 0 296 0 096 0. 156 0. 1 16 40 0 596 0 156 0 .213 0.528 O.OB6 4 1 0 755 0 017 0 123 0.087 0. 1 12 42 0 361 o 458 0 . 1 16 0.362 0.043 43 0 .003 0 . 145 -0 .013 0. 132 0.055 44 0 .674 o . 294 0 . 272 0.006 0.018 45 -0 001 0 .089 -o .043 O. 145 0.049 46 0 .260 0 .038 o 107 0.053 0 095 47 0 . 327 -0 .011 0 .090 0 578 0.098 48 0 .074 o .049 o .046 -0 015 -O.102 - 128 -Appendix II Table showing co r r e l a t i on between p r inc ipa l components and dependent v a r i ab l e , regress ion c o e f f i c i e n t s of p r inc ipa l components and c o e f f i c i e n t s of var iab les obtained from regress ion on p r i n c ipa l components. CORRELATION BETWEEN PRINCIPAL COMPONENTS AND DEPENDENT VARIABLE -0 7997B -0 13799 0 2 17 14 0 03238 -0. 12266 -0 004G2 -o 05299 O 03828 0 02153 -0 24620 0 08300 -0 0580« -0 0354 7 0 00707 -0 1588 I -O 04982 -o. 1 1507 O 09496 O. 128 16 0 0687 1 -0 14 154 0 13993 -0 04092 -0 00929 -0. 02526 0 12266 0. 05347 -o 02 159 -0. 14257 -0 02104 -o 0O224 O 07 190 0 01222 0 .10696 -O. 06297 0 00822 -0 03201 0 01627 -0 00523 REGRESSION COEFFICIENTS OF PRINCIPAL COMPONENTS CONSTANT COMPONENTS (MEAN OF V) 2.86273 -O. 25097 -0 07230 0 14921 0 .02395 -0. 09400 -0 00374 -0 04459 0 03459 0. 02000 -0 24620 0 09598 -0 06435 -0 04341 0 00928 -0 22342 -0 07473 -0. 18390 0. 16459 0. 23514 0 13776 -O 29875 0 30165 -0 09825 -0 025 15 -O 06935 0 .36995 O 175 14 -O. 08220 -O. 59235 -O. ,09154 -O 01026 0 35189 0 06686 0 63352 -O 41 128 0 .06189 -O 30109 0 16703 -O 05469 COEFFICIENTS OF VARIABLES OBTAINED FROM REGRESSION ON PRINCIPAL COMPONENTS INOEX OF RESIDUAL F-VALUES COMPONENTS SUM OF REGRESSION COMPONENT ENTERING SQUARES MODEL TO ENTER R2 CONSTANT VARIABLES 1 A 2 B 3 C 4 D 5 E 6 F 1 26 .68069 86 .98 86 98 0 6396 4 . 2694 -0 0024 -O. 0020 -0 0018 0 0012 -0 0O47 -0 0048 10 22 .19292 56 07 9 . 7 1 0 7003 4 . 4650 -o .0039 0 0022 -0 .0021 -0 0018 -0 0072 -0 .0087 3 18 .70187 46 .36 8 77 0 7474 4 4494 -O 0036 0 OOIB 0 OOI 1 -0 0021 -O 0013 -0 .0103 15 16 .83458 39 .08 5 io 0 7726 4 .4364 -0 .0040 0 0036 -0 0017 -0 0005 0 OOIB -0 .0105 29 15 .32969 34 47 4 .42 0 7930 4 .8315 -0 0072 O 0019 O OOIB -0 .0133 0 0020 -0 0153 2 1 13 .84647 31 .88 4 .71 0 8 130 4 6B10 -0 .0029 0 0035 O .0023 -0 .0030 O .0166 -0 0208 22 12 .39673 30 .55 5 03 0 8326 4 7456 -O .0020 -O 0004 O 0083 -0 .0022 0 002B -0 .0158 2 10 .98685 30 . 13 5 .39 0 8516 4 .7577 -0 .001 1 -O OOOG 0 .0071 -O .0036 0 .0027 -0 0164 19 9 .77070 29 .97 5 . IO 0 8680 4 .6963 O 0009 0 OOI 1 O .0069 -0 0002 0 0090 -0 0136 5 8 .65679 30 .21 5 . 15 0 B831 4 .6967 0 .0002 -0 0007 0 0057 -0 OOOB 0 0090 -0 .0139 26 7 54291 31 26 5 .76 0. 8981 4 6427 0 0020 -0 OOOI 0 0004 0 0053 -0 0156 -o 0104 17 6 .56262 32 56 5 .68 0. 9114 4 .6931 0 .0024 -0 002 1 -O 0004 0 0006 -0 0051 -0 0120 34 5 .7 1551 34 .02 5 .48 0 9228 4 .8916 -0 0066 -0 0028 -O 0007 0 .0023 O .0007 -0 .0097 18 5 .04 780 35 . 15 4 . 76 0 9318 4 9301 -0 0O86 -O 0044 -O OOI 1 0 00B3 o .0006 -0 0129 11 4 .46135 36 39 4 .60 O. 9397 4 .9129 -0 0081 -O 004 1 -O. 0017 0 0086 0 0029 -0 0125 32 4 .07859 36 45 3 . 19 0 9449 4 9424 -0 0086 -0 0060 -0 0036 0 .0046 -0 0006 o 0008 20 3 .72901 36 .60 3 .09 0 9496 4 .9388 -0 .0073 -0 0055 -0 0055 0 OOB2 -0 0006 o .0019 35 3 .43540 36 54 2 73 O. 9536 4 8348 -0 .0047 -0 0002 -0 0005 0 0092 0 .0010 -0 0068 12 3 . 18626 36 .28 2 .42 0 9570 4 .8774 -O 004 1 -0 OOI 1 -0 0CO2 0 O064 0 OOOI -0 0083 27 2 .97455 35 84 2 . 14 0 9598 4 8083 -0 0019 O OOOO -0 OOIO 0 0096 0 0043 -0 0066 7 2 .76669 35 .57 2 18 0 9626 4 8036 -0 OOIO 0 00O9 -o 0008 o .0091 0 0026 -0 0069 16 2 .5829 1 35 21 1 99 O 9651 4 8374 -0 0015 0 0017 -0 OOOI 0 0078 o 0037 -0 .0089 23 2 .45893 34 17 1 36 0 9668 4 8391 -O 0023 0 0029 -o 0016 o 0085 0 0044 -0 0070 8 2 .35042 33 04 1 . 20 0 9683 4 8564 -0 0027 o 0031 -0 0012 0 O072 0 0058 -0 0061 13 2 . 25727 31 80 1 .03 0. 9695 4 8629 -0 0026 0 0021 -o 0013 o 0075 o 0078 -o. 0051 ro Table showing cor re l a t ion between p r inc ipa l components and dependent v a r i a b l e , regress ion c o e f f i c i e n t s of p r inc ipa l components and coe f f i c i en t s of var iab les obtained from regress ion on p r i n c i pa l components. - 130 -Appendix III Eigenvalues fo r p r i n c ipa l components 1 through 10 for each peak obtained from the HPLC p r o f i l e . Component Peaks Eigenvalues for P r i n c i p a l Components 1 2 3 4 S 6 7 8 9 10 t A o. 1 169 -0. 1603 O. 0246 0. 1176 0. OB53 O. 1350 -O. 2431 -o. 1574 0. 3945 0 0694 2 B o. 1 104 0. 0244 -0. 0407 -0. 2202 0. 2734 O. 0921 -O. 2607 0 0675 0. 0225 -0 2354 3 C 0 0968 0 2272 0. 2945 0 0584 0. 1766 -o. 0646 -o. 0545 0 1659 0. 0923 0 0174 4 0 -0. 0411 0 1603 -0. 0191 0. 0852 0. 0576 -0 23BO 0. 1111 -O. 3087 -0. 1759 o 1043 5 E 0. 1183 0. 0056 0. 2461 -0. 0530 -O. 0022 o. 0767 0 2343 0 2635 -0. 0467 0 0617 6 F 0. 1742 0. 0755 -0. 0974 0. .0604 0 0294 o. 1954 0 0692 0 2551 O. 0688 0. 1461 7 G o. 0976 0. 2114 o. 0388 -0. . 1746 0. .0235 o. 1024 0. 2672 -O. 2235 -O. 0305 -0 14 15 a H 0. 0919 0. 2247 o. 2481 0 0089 0. 0057 -o. 2470 o 0381 0. 0090 -O. 0212 0 0503 9 I 0. 1408 0. 2872 -0 0505 0 0451 0. 1823 o. 0912 -0. 0476 -0 0570 -0. 0206 -o. 0434 10 vl o. 1O50 0. 2138 -0. 0615 0. 2219 -0. 048 2 -0. 1700 0. 0963 -O. 2279 0. 0069 0. 0987 11 K o. 1461 O. 2414 0 .0281 -0. . IB20 0 1812 0. 0157 o. 0204 -0. .0622 -0. 0369 0 0097 12 L 0. 1398 0 1442 -0. .2880 0 2628 0 0747 0. 0615 0. 0105 0 .0886 -O. 0247 -0 0029 13 M 0. 1429 0. 14 19 0. . 1578 -0 . 1628 o. .0163 0 1954 0. .0516 -0 . 1567 -O. 2263 -0 . 1156 14 N o. 1203 -0. 101 1 -0. 0125 -0. 0522 -0. 1968 -o. 2318 0. 2777 0. .2245 0. 0605 -0 . 1599 IS 0 0. 1958 -0. 0112 -0. 0024 -0. 0028 0 0504 0. 2124 o 1005 0 1 106 O. 08 3S o .1117 16 P o. 1501 0 2549 -o. 1357 0. .0846 0 0725 0. 0256 -0 0098 o 0739 O. .0321 -0 . 1564 17 0 0. . 1419 0 .0303 -o. 1287 -0 0028 0. 0829 -0. . 1887 0 0832 -o . 1072 -0 1988 0 4098 IB R 0. 1374 -0. 0803 -0. 1584 -o .3129 o . 1247 o. 1646 o 0179 0 .2339 0 0072 o . 1099 19 S 0. 1958 -0 0787 o. 0580 0. . 1778 o. 0775 -o. 0760 -o. 0047 -o 0348 0 1 188 0 .0623 20 T 0. 1375 -0 . 1769 -0. 1351 -0 0602 o. 0204 -0. 2027 o. 0291 -0 .0662 0. 0729 o 2533 21 U 0. . 1875 -0. 0621 -o. . 1784 -0. . 1559 b 0856 -o. 1372 0. 1044 o .0835 -0. . 1292 -0 0527 22 v 0. 0895 0. . 1627 -0. 0998 0. 1 1 19 -0. 1820 0 0352 0. 04 36 0 .3174 0. .0967 -0. .0670 23 w 0. 1068 0 0700 -o. . 1250 -0. .2125 -0 1245 o. OI7I 0. 0659 -0 . 1691 o. 4527 -0 .0322 24 X 0. 1 120 0. 1469 -0. .2222 -0 . 1778 -0. 0225 -o. 1017 -o. .0157 -0 . 2686 0. .3050 -0 191 I 25 V 0. 1466 -0. 0516 -0. . 1637 -0. 0448 -0. 28 IO -o. 0745 0. 2812 -0 0353 -0. 0582 -0 2007 26 z 0. 0941 -0. 0329 -0. 2312 -0 2364 -0 1098 -o. 0731 -0. 2866 0 .0296 -0. 2653 0 2752 27 AA 0. 1920 -0. 17 12 -o. 0159 -0 .0729 0. 1770 -0. .0550 -0. 067 1 o .0128 -0. 1257 -0 O055 28 BB 0. .2111 -0 .0923 o. 0258 0 .2128 0. 0482 -0 097 1 -0. 0643 -o 0246 0. .0792 -0 .0989 29 CC o. 1923 0 0122 0 0162 -0 0367 -0 0610 -o 0936 -0 2132 -0 .077 • 0 .0047 -0 0097 30 00 0. 1970 0. O039 o 0302 0 0182 -o. 0931 -o 2280 -o. O077 o . 1731 -o. OI51 -o 1628 31 EE o. 1363 -0. 2047 -o. 1101 -0. 0559 -0. 0169 -o. 0332 -0. 1597 o .0689 -0 1351 -0 . 1252 32 FF 0. 2175 o. 0407 -o. 1049 0 0896 0. . 1681 o. . I09B o. 0631 o . 1214 o 0732 0 .0929 33 GG 0. 1318 0. 207 1 0 . 1319 -0 1517 -0 0225 -0 .2229 -0 1 173 0 .0186 0 .02 18 0 .0514 34 HH o. 1702 0 0556 o. 3179 0 .0421 -o OI36 o 0775 -o 17 11 -0 0212 -0 . 1555 -0 0702 35 I I o. 1361 -0 2544 0. 0358 -0. 0665 0. 0217 -o 0335 0 1 170 -o 0682 -0. 0493 -0 . 1947 36 JJ 0. 1676 -0 2397 0. 1073 0. . 1492 -0 .0455 -o 1296 0 0216 -0 0555 o. 0558 -0 .0917 37 KK 0. 2159 -0. 1299 -o. 0317 -0 0329 0. . 1216 -0. 1031 -0. 0186 -o .0770 -o. 1 195 0 0960 38 L L 0. 0587 0 0889 -o. 0083 -0 2346 -0 2902 O 2821 0 097 3 -o . 1207 -0 0787 0 . 1651 39 MM 0 1905 -o 1245 0. 2139 0 .0480 -0 0132 O 0286 0 04 38 -o . 1380 o .094 2 0 . 1334 40 NN 0. 1493 -o. 1720 o. 1682 -o O024 o 0281 o. 2001 0 3045 -o 1 123 -o. 0596 0 .02 73 41 00 o. 1386 -0 2165 o. 1549 -0 OOOI 0 0047 o 1881 0 0279 -o . 2002 0 .0651 0 . 1165 42 PP 0. 1553 0 0818 0. 2036 0 .0620 -0 2851 0 0197 0 O067 0 .0843 -0 0525 0 .0705 43 00 0. 0525 0 0612 o 0991 -0 0882 -0 3813 -o oaoa -0 2482 0 . 1188 -o .0103 0 . 1 122 44 RR 0. 1872 0 0053 0 0544 0 . 1483 -0. .0929 0 0123 -0 2437 -0 .0104 -0 .0380 -0 .0782 45 SS o 0264 0 0447 o 0877 -0. 2197 -o. 2239 -0. 0695 -o 0801 o. 0342 o. 3194 0. 1984 46 TT 0. O04 2 -o 0187 o. 02 14 -0. 06G6 -0. 1984 o. 15 IO -o. 2406 -Oi1757 -o. 1953 -0. 2949 47 uu o. 1386 -o O051 -o. 2374 0. 2625 -o. 2075 o. 004 2 0 O029 -0. 0738 -0. 1 133 -0. 1 184 48 Vv o. OC96 o 089 1 -0. 1388 0 2587 -o. 1981 o 3140 -0 0700 -o 0749 0 0. .1325. Eigenvalues fo r p r i nc ipa l components 1 through 10 f o r each peak obtained from the HPLC D r o f i l e . CO »—> I - 132 -LITERATURE CITED Acharya, A . S . , A. Di Donato, B.N. Manjula, V.A. F i s h e t t i , and J . M . Manning. 1983. Influence of t r i f l u o r o a c e t i c acid on re tent ion times of h i s t id ine-conta in ing t r yp t i c peptides in reverse phase HPLC. Int. J . Peptide Protein Res. 22:78. Adda, J . , J . C . Gr ipon, and L. Vassa l . 1982. The chemistry of f lavour and texture generation in cheese. Food Chem. 9:115. Aishima, T. 1983. Comparison of head space and d i s t i l l a t i o n techniques for soy sauce aroma in r e l a t i on to ana lys is by s i l i c a c a p i l l a r y gas chromatography and sensory eva lua t ion . Report of the NODA Ins t i tu te for S c i e n t i f i c Research No. 27. 399 NODA, Noda-Shi, Chiba-ken 278, Japan. Aishima, T. 1982. D iscr iminat ion and c l u s t e r ana lys i s of soy sauce GLC p r o f i l e s . J . Food S c i . 47:1562. Aishima, T. 1979. Object ive evaluat ion of soy sauce by s t a t i s t i c a l ana lys i s of GC p r o f i l e s . Ag r i c . B i o l . Chem. 43:1935. Anders, R. F. and G. R. Jago. 1970. The concentrat ion of o l e i c acid in Cheddar cheese r e l a t i v e to loss of s t a r t e r v i a b i l i t y . Aust. J . Dairy Technol . 25:68. AOAC. 1980. O f f i c i a l Methods of Ana lys i s . 13th ed. Assoc ia t io nof O f f i c i a l Ana l y t i ca l Chemists, Washington, DC. Aston, J . W. and J . R. Dul ley . 1982. Cheddar cheese f l avour . Aust. J . Dairy Technol . 37:59. Aston, J . W., I. G. Durward and J . R. Dul ley . 1983. P ro teo l ys i s and f lavour development in Cheddar cheese. Aust. J . Dairy Technol . 38:55. BMDP7M. 1981. In "BMDP S t a t i s t i c a l Software", p. 519-536. Univers i ty of C a l i f o r n i a Press, Los Angeles, CA. Babel , F. J . 1977. An t i b i o s i s by l a c t i c cu l ture bac t e r i a . J . Dairy S c i . 60:815. Badings, H. T. and R. Neeter. 1980. Recent advances in the study of aroma compounds of milk and dairy products. Neth. Milk Dairy J . 34 :9 . Baron, V. M. and G. W. Scott B l a i r . 1953. Rheology of cheese and curd . In "Food S t u f f s : The i r P l a s t i c i t y , F l u i d i t y and Cons is tency , " by G. W. Scott B l a i r , North-Holland Publ ish ing Company, Amsterdam, The Netherlands, pp. 124. B i can , P. and B. B lanc. 1982. Milk prote in ana lys is - a high-performance chromatography study. Mi lchwissenschaft 37:592. - 133 -B i l l s , D. D. and E. A. Day. 1964. Determination of the major f ree fa t ty acids of Cheddar cheese. J . Dairy S c i . 47:733. Bishop, C. A . , D. R. K. Harding, L. J . Meyer, W. S. Hancock and M. T. W. Hearn. 1980. High-performance l i q u i d chromatography of amino ac ids , peptides and proteins XXI. The app l i ca t ion of preparat ive reversed-phase high performance l i q u i d chromatography fo r the p u r i f i c a t i o n of a synthet ic under ivat ized pept ide. J . Chromatogr. 192:222. Bourne, M.C. 1978. Texture p r o f i l e ana l ys i s . Food Techno l . 32(7):62. Ca r t e r , E.J.V. and P. Sherman. 1978. Evaluat ion of the firmness of Le i ces te r cheese by compression tes ts with the Instron Universal Test ing Machine. J . Texture S tudies . 9:311. Champion, H.M. and D.W. Stanley. 1982. HPLC separat ion of b i t t e r peptides from Cheddar cheese. Can. Inst . Food S c i . Technol . J . 15:283. Chen, A . H . , J.W. La rk in , C. J . Clark and W.E. Irwin. 1979. Textural ana lys i s of cheese. J . Dairy S c i . 62:901. Coomans, D., D.L. Massart and L. Kaufman. 1979. Opt imizat ion by s t a t i s t i c a l l i nea r d iscr iminant ana lys is in ana l y t i ca l chemistry. Anal . Chim. Acta 112:97. Creamer, L.K. and N.F. Olson. 1982. Rheological eva luat ion of maturing Cheddar cheese. J . Food S c i . 47:631. C u l i o l i , J . and P. Sherman. 1976. Evaluat ion of Gouda cheese firmness by compression t e s t s . J . Texture S tudies . 10:165. Czulak, J . , D.J. Bant, S.C. Blyth and J . Crace. 1979. A new cheese s ta r te r system. Dairy Ind. Int . 44(2) :17. D a n i e l l , S.D. W.E. Sandine. 1981. Development and commercial use of a mul t ip le s t r a i n s t a r t e r . J . Dairy S c i . 64:407. Dau l t ry , S. 1976. " P r inc ipa l Components Ana l y s i s . " by Geo. Abstracts L t d . , Un ivers i ty of East Ang l i a , Norwich, England. Day, E.A., R. Bassette , and M. Keeney. 1960. I den t i f i c a t i on of v o l a t i l e carbonyl compounds from Cheddar cheese. J . Dairy S c i . 43:463. Derde, M.P. and D.L. Massart. 1982. Ext rac t ion of information from large data sets by Pattern Recogni t ion. Fresenius Z e i t s c h r i f t fu r Ana ly t ische Chemie 313(6):484. de Jong, L. 1975. A quant i ta t i ve e lec t rophoret i c method of studying cheese r i pen ing . Neth. Milk Dairy J . 29:162. Desmazeaud, M. and J . C . Gripon. 1977. General mechanism of prote in breakdown during cheese pathway fo r saturated ac ids . Biochem. J . 111:395. - 134 -Dick inson , E. and I.e. Goulding. 1980. Y i e ld behaviour of crumbly Engl ish cheeses in compression. J . Texture Studies 11:51. Diosady, L . L . , I. Bergen and V.R. Harwalkar. 1980. High-performance l i q u i d chromatography of whey p ro te ins . Mi lchwissenschaft . 35:671. Edwards, J . and F.V. Kosikowski. 1983. B i t t e r compounds from Cheddar cheese. J . Dairy S c i . 66:727. Emmons, D.B. 1978. Recent developments in the process and mechanization of Cheddar cheese product ion. XX International Dairy Congress, Pa r i s , pp. 1-8. S c i e n t i f i c and Technical Sess ions. Emmons, D.B., M. Kalab, E. Larmond and R.J. Lowrie. 1980. Milk gel s t ruc tu re . X. Texture and microstructure in Cheddar cheese made from whole milk and from homogenized low-fat mi lk . J . Texture S tud ies . 11:15. Evans, E.W., L.A. Mabbit. 1974. The f lavour of Cheddar cheese. in B iennia l Reviews. Nat. Inst. Res. in Da i ry ing , Reading, England, pp. 59-71. Foda, E.A. , E.G. Hammond, G.W. Reinbold and D. K. Hotchkiss. 1974. Role of f a t in f lavour of Cheddar cheese. J . Dairy S c i . 57:1137. Forss , D.A. 1979. Review of the progress of dairy sc ience: Mechanisms o f formation of aroma compounds in milk and milk products. J . Dairy Res. 46:691. Forss , D.A. and S. Patton. 1966. Flavour of Cheddar cheese. J . Dairy S c i . 49:89. Fox, P.F. 1981. Proteinases in dairy technology. Neth. Milk Dairy J . 35:233. Frank, I.E. and B.R. Kowalski. 1982. Chemometrics. Ana l . Chem. 54(5):232. Gazdag, M. and G. Szepes i . 1981. Separation of large polypeptides by high-performance l i q u i d chromatography. J . Chromatogr. 218:603. G la j ch , J . L . , J . J . K i rk l and , K.M. Squire , and J . M . Minor. 1980. Opt imizat ion of solvent strength and s e l e c t i v i t y for reversed-phase l i q u i d chromatography using in te rac t i ve mixture-design s t a t i s t i c a l technique. J . Chromatogr. 199:57. Gooda, E., W. Bednarski and S. Poznanski. 1983. The prote in degradation in Cheddar cheese manufactured from milk treated with l ac tase . Mil chwissenschaft 38:83. Green, M.L., A. Turvey and D.G. Hobbs. 1981. Development of s t ructure and texture in Cheddar cheese. J . Dairy Res. 48:343. - 135 -Gr ieve , P.A. and J .R . Dul ley . 1983. Use of Streptococcus 1act is Lac mutants for acce le ra t ing Cheddar cheese r i pen ing . 2. Their e f f e c t on the rate of p ro teo l ys i s and f lavour development. Aust. J . Dairy Technol . 38:49. H a l l , R.L. 1968. Food f l avours : benef i t s and problems. Food Technol . 22:1388. Hancock, W.S., J . D . Capra, W.A. Bradley and J . T . Sparrow. 1981. The use of reversed-phase HPLC with rad ia l compression for the ana lys is of peptide and prote in mixtures. J . Chromatogr. 206:59. Hancock, W.S., and J . T . Sparrow. 1981. Use of mixed-mode, high-performance l i q u i d chromatography for the separation of peptide and prote in mixtures. J . Chromatogr. 206:71. Harper, W.J. 1959. Chemistry of cheese f l avours . J . Dairy S c i . 42:207. Harper, W.J., J . Y . Wang and T. K r i s t o f f e r s e n . 1979. Free fat ty acids and amino acids in fa t modif ied cheese. Mi lchwissenschaft 34:525. Harwalkar, V.R. and J . A . E l l i o t . 1970. I so la t ion of b i t t e r and as t r ingent f r ac t i ons from Cheddar cheese. J . Dairy S c i . 54:8. Heap, H.A. and R.C. Lawrence. 1976. The se lec t ion of s t a r t e r s t ra ins fo r cheesemaking. N.Z. J . Dairy S c i . 11:16. Hearn, M.T.W., B. Grego, and W.S. Hancock. 1979. High-performance l i q u i d chromatography of amino ac ids , peptides and pro te ins . J . Chromatogr. 185:429. Hearn, M.T.W., F.E. Regnier and C.T. Wehr. 1982. HPLC of peptides and p ro te ins . Amer. Lab. October, 1982:18. Heukeshoven, J . and R. Dernick. 1982. Reversed-phase high-performance l i q u i d chromatography of v i rus proteins and other large hydrophobic proteins in formic acid containing so lvents . J . Chromatogr. 252:241. Holmes, D.G. , J.W. Duersch and C A . Ernstrom. 1977. D i s t r i b u t i o n of milk c l o t t i n g enzymes between curd and whey and t he i r surv iva l during Cheddar cheese making. J . Dairy S c i . 60:862. Hong, G .L . , L.H. Davis and G.H. Richardson. 1977. Production of bulk l a c t i c cu l tu re concentrates for r e f r i ge ra t ed storage and d i s t r i b u t i o n . J . Dairy S c i . (Abstract ) 60(Suppl. 1 ) :38. H u l l , R.R. 1977. Methods for monitoring bacteriophage in cheese f a c t o r i e s . Aust. J . Dairy Technol . 32:65. Imoto, E.M., C-H. Lee and C. Rha. 1979. E f f ec t of compression r a t i o on the mechanical propert ies of cheese. J . Food S c i . 44:343. J a r r e t t , W.O., J.W. Aston and J .R . Dul ley . 1982. A simple method for est imat ing free amino acids in Cheddar cheese. Aust. J . Dairy Technol . 37:55. - 136 -Keogh, B.P. 1972. A re-assessment of the s t a r t e r r o t a t i o n system. Aust. J. Dairy Technol. 27:86. Kosikowski, F.V. 1951. The l i b e r a t i o n of free amino acids i n raw and pasteurized m i l k Chedder cheese during r i p e n i n g . J . Dairy S c i . 34:235. Kowalski, B.R. 1980. Chemometrics. Anal. Chem. Rev. 52:112R. K r i s t o f f e r s e n , T. 1973. Biogenesis of cheese f l a v o u r . J . A g r i c . Food Chem. 21:573. Kuchroo, C.N. and Fox, P.F. 1982. Soluble n i t r o g e n i n Cheddar cheese: Comparison of e x t r a c t i o n procedures. Milchwissenschaft 38:331. Kuchroo, C.N. and P.F. Fox. 1983. F r a c t i o n a t i o n of the water-soluble n i t r o g e n from Cheddar cheese: Chromatographic methods. Milchwissenschaft 37:76. Kupke, I and D. Sauer. 1970. Spektrophotometrische Mikrobestimmung von c h l o r i d a l s h e x a c h l o r o f e r r a t III-komplex i n Blutserum and anderen K o p e r f l u s s i g k e i t e n . J . Anal. Chem. 250:101. Kwan, K.K.H., S. Nakal and B.J. Skura. 1983. Comparison of four methods f o r determining protease a c t i v i t y i n milk. J . Food S c i . 48:1418. Kwan, W-0. and B.R. Kowalski. 1978. C l a s s i f i c a t i o n of wines by applying pat t e r n r e c o g n i t i o n to chemical composition data. J . Food S c i . 43:1320. Kwan, W-0. and B.R. Kowalski. 1980a. Data a n a l y s i s of sensory scores. Evaluations of p a n e l i s t s and wine score cards. J . Food S c i . 45:213. Kwan, W-0. and B.R. Kowalski. 1980b. Patt e r n r e c o g n i t i o n a n a l y s i s of gas chromatographic data. Geographic c l a s s i f i c a t i o n of wines of V i t i s  v i n i f e r a cv. Pinot Noir from France and the United States. J . A g r i c . Food Chem. 28:356. Kwan, W-0., B.R. Kowalski and R. K. Skogerboe. 1979. P a t t e r n r e c o g n i t i o n a n a l y s i s of elemental data. Wines of V i t i s v i n i f e r a cv. Pinot Noir from France and the United S t a t e s . J . A g r i c . Food Chem. 27:1321. Lamparsky, D. and I . Klimes. 1981. 3rd Weurraan Symposium, Proc. of the I n t e r n a t i o n a l Conference, Munich, A p r i l 28-30, Ed. P. S c h r e i e r . Walter de Gruyter, B e r l i n , pp. 557-577. Law, B.A. 1978. The accelerated r i p e n i n g of cheese by the use of non-conventional s t a r t e r s and enzymes - a pr e l i m i n a r y assessment. XX I n t e r n a t i o n a l D airy Congress, P a r i s , Commission F, pp. 40-50. Law, B.A. 1981. The formation of aroma and f l a v o u r compounds i n fermented d a i r y products. Dairy S c i . A b s t r a c t s . 43:143. Fermentation Group Symposium of the Society f o r General Microbiology i n D u b l i n , Sept. 1979. - 137 -Law, B.A., M.J. Castafion and M.J. Sharpe. 1976. The c o n t r i b u t i o n of s t a r t e r s t r e p t o c o c c i to f l a v o u r development i n Cheddar cheese. J . Dairy Res. 43:301. Law, B.A. , and M.E. Sharpe. 1977. The i n f l u e n c e of the m i c r o f l o r a of Cheddar cheese on f l a v o u r development. Dairy Inds. I n t . 42:10. Lawrence, R.C. and J . G i l l e s . 1980. The assessment of the p o t e n t i a l q u a l i t y of young Cheddar cheese. N. Z. J . Dairy S c i . Technol. 15:1. Lawrence, R.C. and J . G i l l e s . 1982. Factors that determine the pH of young Cheddar cheese. N.Z. J . Dairy S c i . Technol. 17:1. Lawrence, R.C, H.A. Heap, G. Limsowtin and A.W. J a r v i s . 1978. Cheddar cheese s t a r t e r s : current knowledge and p r a c t i c e s of phage c h a r a c t e r i s t i c s and s t r a i n s e l e c t i o n . J. Dairy S c i . 61:1181. Lawrence, R.C, T.D. Thomas and B.E. Terzaghi. 1976. Reviews of the progress of d a i r y s c i e n c e : Cheese s t a r t e r s . J . Dairy Res. 43:141. Lee, C-H. , E.M. Imoto and C.K. Rha. 1978. E v a l u a t i o n of cheese te x t u r e . J. Food S c i . 43:1600. L e l i e v r e , J . 1983. The i n f l u e n c e of cheese composition on the f i n a n c i a l r e t u r n from the manufacture of Cheddar cheese i n New Zealand. N.Z. J. Dairy S c i . Technol. 18:63. L e l i e v r e , J . and J . G i l l e s . 1982. The r e l a t i o n s h i p between the grade (product value) and composition of young commercial Cheddar cheese. N.Z. J . Dairy S c i . Technol. 17:69. Limsowtin, G.K.Y., H.A. Heap and R.C. Lawrence. 1980. A new approach to the p r e p a r a t i o n of bulk s t a r t e r i n commercial cheese p l a n t s . N.Z. J . Dairy S c i . and Technol. 15:219. Limsowtin, G.K.Y., H.A. Heap and R.C. Lawrence. 1977. A m u l t i p l e s t a r t e r concept f o r cheese making. N.Z. J . Dairy S c i . Technol. 12:101. Limsowtin, G.K.Y. and B.E. Terzaghi. 1976. Phage r e s i s t a n t mutants: t h e i r s e l e c t i o n and use i n cheese f a c t o r i e s . N.Z. J . Dairy S c i . Technol. 11:251. Lowrie, R.J. and R.C. Lawrence. 1972. Cheddar cheese f l a v o u r . IV. Anew hypothesis to account for the development of b i t t e r n e s s . N.Z. J . Dairy S c i . Technol. 7:51. Lowrie, R. J . , R.C. Lawrence and M.F. Peberdy. 1974. Cheddar cheese f l a v o u r . V. Influence of bacteriophage and cooking temperature on cheese made under c o n t r o l l e d b a c t e r i o l o g i c a l c o n d i t i o n s . N.Z. J. Dairy S c i . Technol. 9:116. Luck, H. and T.E.H. Downes. 1972. Flavour development i n cheese made from skimmilk and vegetable f a t . S. A f r . J . Dairy Technol. 4:221. - 138 -Mabbit, L.A. 1961. Reviews of the progress of d a i r y s c i e n c e . Section B. B a c t e r i o l o g y , the f l a v o u r of Cheddar cheese. J . Dairy Res. 28:303. McBride, R. L. and C. H a l l . 1979. Cheese grading versus consumer a c c e p t a b i l i t y : an i n e v i t a b l e discrepancy. Aust. J . Dairy S c i . Technol. 34:66. McDowall, F.H. and L.A. Whelan. 1933. The e f f e c t of common s a l t on the growth of l a c t i c s t r e p t o c o c c i i n milk. J . Dairy Res. 5:42. McGugan, W.A., D.B. Emmons, and E. Larmond. 1979. Influence of v o l a t i l e and n o n v o l a t i l e f r a c t i o n s on i n t e n s i t y of Cheddar f l a v o u r . J . Dairy S c i . 62:398. Manning, D.J. 1974. Sulphur compounds i n r e l a t i o n to Cheddar cheese f l a v o u r . J . Dairy Res. 41:81. Manning, D.J. 1979. Chemical production of e s s e n t i a l Cheddar flavour compounds. J . Dairy Res. 46:531. Manning, D.J. and J.C. P r i c e . 1977. Cheddar cheese aroma - the e f f e c t of s e l e c t i v e l y removing s p e c i f i c c l a s s e s of compounds from cheese headspace. J . Dairy Res. 44:357. Martens, H. and H. Russwurm, J r . 1983. Food Research and Data A n a l y s i s . Applied Science P u b l i s h e r s , London and New York. M e i s t e r , K.A. and R.A. Ledford. 1979. Optimum c u l t u r a l conditions f o r i n d u c t i o n of temperate bacteriophages i n l a c t i c s t r e p t o c o c c i . J . Food P r o t e c t i o n . 42:396. Mermelstein, N.H. 1982. Advanced bulk s t a r t e r medium improves fermentation processes. Food Technol. 36:69, August. Morgan, S.L. and S.N. Deming. 1975. Optimization s t r a t e g i e s for the development of g a s - l i q u i d chromatographic methods. J . Chromatogr. 112: 267. Mulder, H. 1952. Taste and f l a v o u r forming substances i n cheese. Neth. M i l k Dairy J . 6:157. Nakai, S. 1981. Comparison of o p t i m i z a t i o n techniques f o r a p p l i c a t i o n to food product and process development. J . Food S c i . 47:144. Nakai, S., K. Koide and K. Eugster. 1984. A new mapping super-simplex o p t i m i z a t i o n f o r food product and process development. J . Food S c i . In press. Nakai, S., H.K. Wilson and E.O. H e r r e i d . 1964. Assaying s t e r i l e concentrated m i l k f o r native p r o t e o l y t i c enzymes. J . Dairy S c i . 47:754. - 139 -Nelder, J.A. and R. Mead. 1965. A simplex method f o r f u n c t i o n m i n i m i z a t i o n . Computer J . 7:308. Ney, K.H. 1971. Determination of the degree of r i p e n i n g of cheese us i n g simple c h a r a c t e r i s t i c f a c t o r s . Milchwissenschaft 26:269. Nieuwoudt, J.A. 1977. Ripening of mature Cheddar cheese and po s s i b l e means of determining degree of r i p e n i n g . South A f r . J . Dairy Technol. 9:143. Olson, N. F., R.C. Lindsay and R. Grappin. 1983. Chemical i n d i c e s of cheese maturation. Background document f o r d i s c u s s i o n at the 1983 IDF Annual Session i n Oslo, Norway. Park, S.S. , K.S. Kim, Y.J. Kim and Y.K. Kim. 1978. Studies on cheese r i p e n i n g , v i . Changes i n water soluble nitrogenous compounds during the r i p e n i n g of Cheddar cheese. Korean J . A n i . S c i . 20:200. Patt o n , S. 1963. V o l a t i l e acids and the aroma of Cheddar cheese. J . Dairy S c i . 46:856. Pearce, R.J. 1983. A n a l y s i s of whey proteins by high performance l i q u i d chromatography. Aust. J . Dairy Technol. 38:114. Pearce, K.N. and J . G i l l e s . 1979. Composition and grade of Cheddar cheese manufactured over three seasons. N.Z. J . Dairy S c i . Technol. 14:63. Pettersson, H-E. and G. Sjbstrom. 1975. Accelerated cheese r i p e n i n g : a method f o r i n c r e a s i n g the number of l a c t i c s t a r t e r b a c t e r i a i n cheese without de t r i m e n t a l e f f e c t to the cheese-making process and i t s e f f e c t on cheese r i p e n i n g . J . Dairy Res. 42:313. Pham, A.M. and S. Nakai. 1984. A p p l i c a t i o n of stepwise d i s c r i m i n a n t a n a l y s i s to HPLC p r o f i l e s of water e x t r a c t for judging r i p e n i n g of Cheddar cheese. J. Dairy S c i . In press. Pomeranz, Y. and C.E. Meloan. 1980. "Food a n a l y s i s : Theory and P r a c t i c e . " The AVI P u b l i s h i n g Company, Inc. Westport, CT. Powers, J . J . and E.S. K e i t h . 1968. Stepwise d i s c r i m i n a n t a n a l y s i s of gas chromatographic data as an aid i n c l a s s i f y i n g the f l a v o u r q u a l i t y of foods. J . Food S c i . 33:207. Regnier, F.E. and K.M. Gooding. 1980. High-performance l i q u i d chromatography of p r o t e i n . Anal. Biochem. 103:1. Richardson, G.H. , CA. Ernstrom and G.L. Hong. 1981. The use of defined s t a r t e r s t r a i n s and c u l t u r e n e u t r a l i z a t i o n i n U.S. cheese p l a n t s . C u l t . Dairy Prod. J. 16(2):11. Richardson, G.H. , CA. Ernstrom, J.M. Kim and C. Daly. 1983. Proteinase negative v a r i a n t s of Stretococcus cremoris for cheese s t a r t e r s . J . Dairy S c i . 66:2278. - 140 -R i c h a r d s o n , G . H . , G . L . Hong and C A . E r n s t r o m . 1 9 8 0 . D e f i n e d s i n g l e s t r a i n s o f l a c t i c s t r e p t o c o c c i i n b u l k c u l t u r e f o r C h e d d a r a n d M o n t e r e y c h e e s e m a n u f a c t u r e . J . D a i r y S c i . 6 3 : 1 9 8 1 . R o u t h , M . W . , P . A . S w a r t z and M . B . D e n t o n . 1 9 7 7 . P e r f o r m a n c e o f t h e s u p e r m o d i f i e d s i m p l e x . A n a l . Chem . 4 9 : 1 4 2 2 . S a l j i , J . P . a n d M. K r o g e r . 1 9 8 1 . P r o t e o l y s i s a n d l i p o l y s i s i n r i p e n i n g C h e d d a r c h e e s e made w i t h c o n v e n t i o n a l b u l k s t a r t e r and w i t h f r o z e n c o n c e n t r a t e d d i r e c t - t o - t h e - v a t s t a r t e r c u l t u r e . J . F o o d S c i . 4 6 : 1 3 4 5 . S c h r e i e r , P . a n d L . R e i n e r . 1 9 7 9 . C h a r a c t e r i z a t i o n a n d d i f f e r e n t i a t i o n o f g r a p e b r a n d i e s b y m u l t i p l e d i s c r i m i n a n t a n a l y s i s . J . S c i . F o o d A g r i c . 3 0 : 3 1 9 . S h a m a , F . and P . S h e r m a n . 1 9 7 3 . E v a l u a t i o n o f some t e x t u r a l p r o p e r t i e s o f f o o d s w i t h t h e I n s t r o n U n i v e r s a l T e s t i n g M a c h i n e . J . T e x t u r e S t u d . 4 : 3 4 4 . S i m p s o n , R. J . , M .R . N e u b e r g e r and T . Y . L i u . 1 9 7 6 . C o m p l e t e a m i n o a c i d a n a l y s i s o f p r o t e i n f r o m a s i n g l e h y d r o l y s a t e . J . B i o l . C h e m . 2 5 1 : 1 9 3 6 . S m e y e r s - V e r b e k e , J . , D . L . M a s s a r t a n d D . C o o m a n s . 1 9 7 7 . A p p l i c a t i o n o f l i n e a r d i s c r i m i n a n t a n a l y s i s t o t h e d i f f e r e n t i a t i o n o f p u r e m i l k f r o m d i f f e r e n t s p e c i e s and m i x t u r e s . J . A s s o c . O f f . A n a l . C h e m i s t s 6 0 : 1 3 8 2 . S n y d e r , L . R . 1 9 7 4 . C l a s s i f i c a t i o n o f t h e s o l v e n t p r o p e r t i e s o f common l i q u i d s . J . C h r o m a t o g r . 9 2 : 2 2 3 . S n y d e r , L . R . 1 9 7 8 . C l a s s i f i c a t i o n o f t h e s o l v e n t p r o p e r t i e s o f common l i q u i d s . J . C h r o m a t o g r . S c i . 1 6 : 2 2 3 . S n y d e r , L . R . a n d J . J . K i r k l a n d . 1 9 7 9 . " I n t r o d u c t i o n t o M o d e r n L i q u i d C h r o m a t o g r a p h y , " W i l e y , N Y . , 2nd e d . , C h . 2 . S z c z e s n i a k , A . S . 1 9 7 5 . G e n e r a l F o o d s t e x t u r e p r o f i l e r e v i s i t e d t e n y e a r s p e r s p e c t i v e . J . T e x t u r e S t u d . 6 : 5 . T h o m a s , T . D . a n d O . E . M i l l s . 1 9 8 1 . P r o t e o l y t i c e n z y m e s o f s t a r t e r b a c t e r i a . N e t h . M i l k D a i r y J . 3 5 : 2 5 5 . T h o m a s , T . D . a n d K . N . P e a r c e . 1 9 8 1 . I n f l u e n c e o f s a l t o n l a c t o s e f e r m e n t a t i o n a n d p r o t e o l y s i s i n C h e d d a r c h e e s e . N . Z . J . D a i r y S c i . T e c h n o l . 1 6 : 2 5 3 . T h u n e l l , R . K . , W . E . S a n d i n e and F . W . B o d y f e l t . 1 9 8 1 . P h a g e - i n s e n s i t i v e m u l t i p l e - s t r a i n s t a r t e r a p p r o a c h t o C h e d d a r c h e e s e m a k i n g . J . D a i r y S c i . 6 4 : 2 2 7 0 . - 141 -T r i e u - C u o t , P . a n d J - C . G r i p o n . 1 9 8 2 . A s t u d y o f p r o t e o l y s i s d u r i n g C a m e m b e r t c h e e s e r i p e n i n g u s i n g i s o e l e c t r i c f o c u s i n g a n d t w o - d i m e n s i o n a l e l e c t r o p h o r e s i s . J . D a i r y R e s . 4 9 : 5 0 1 . T w e e t e n , T . N . a n d C . B . E u s t o n . 1 9 8 0 . A p p l i c a t i o n o f h i g h p e r f o r m a n c e l i q u i d c h r o m a t o g r a p h y i n t h e f o o d i n d u s t r y . F o o d T e c h n o l . 1 2 : 2 9 . V a r m u z a , K. 1 9 8 0 . P a t t e r n r e c o g n i t i o n i n a n a l y t i c a l c h e m i s t r y . A n a l . C h i m . A c t a . 1 2 2 : 2 2 7 . V i s s e r , S . 1 9 8 1 . P r o t e o l y t i c e n z y m e s and t h e i r a c t i o n o n m i l k p r o t e i n s . A r e v i e w . N e t h . M i l k D a i r y J . 3 5 : 6 5 . V o i s e y , P .W. 1 9 7 7 . I n t e r p r e t a t i o n o f f o r c e - d e f o r m a t i o n c u r v e s f r o m t h e s h e a r - c o m p r e s s i o n c e l l . J . T e x t u r e S t u d . 8 : 1 9 . W a l k e r , N . J . a n d A . R . K e e n . 1 9 7 4 . F o r m a t i o n o f m e t h y l k e t o n e s d u r i n g r i p e n i n g o f C h e d d a r c h e e s e . J . D a i r y R e s . 4 1 : 7 3 . W a l t e r , H . E . , A . M . S a d l e r , C . D . M i t c h e l l a n d R . E . H a r g r o v e . 1 9 5 8 . E f f e c t o f s a l t o n a c i d d e v e l o p m e n t i n C h e d d a r c h e e s e . J . D a i r y S c i . ( A b s t r a c t ) 4 1 : 7 1 8 . W i l s o n , K . J . , A . H o n e g g e r , R . P . S t o t z e l , R . P . a n d G . J . H u g h e s . 1 9 8 1 . T h e b e h a v i o u r o f p e p t i d e s o n r e v e r s e - p h a s e s u p p o r t s d u r i n g h i g h - p r e s s u r e l i q u i d c h r o m a t o g r a p h y . B i o c h e m . J . 1 9 9 : 3 1 . W o n g , N . P . , R. E l l i s a n d D . E . L a C r o i x . 1 9 7 5 . Q u a n t i t a t i v e d e t e r m i n a t i o n o f l a c t o n e s i n C h e d d a r c h e e s e . J . D a i r y S c i . 5 8 : 1 4 3 7 . W o o , A . H . a n d R . C . L i n d s a y . 1 9 8 3 . S t a t i s t i c a l c o r r e l a t i o n o f q u a n t i t a t i v e f l a v o u r i n t e n s i t y a s s e s s m e n t s a n d i n d i v i d u a l f r e e f a t t y a c i d m e a s u r e m e n t s f o r r o u t i n e d e t e c t i o n and p r e d i c t i o n o f h y d r o l y t i c r a n c i d i t y o f f - f l a v o u r s i n b u t t e r . J . F o o d S c i . 4 8 : 1 7 6 1 . 

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:
https://iiif.library.ubc.ca/presentation/dsp.831.1-0096248/manifest

Comment

Related Items