Development of Saponin-Rich Baked Goods DISSERTATION Presented in Partial Fulfillment of the Requirements for the Degree Doctor

Development of saponin-rich baked goods

DISSERTATION

Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy in the Graduate School of The Ohio State University

Luca Serventi

Graduate Program in Food Science and Nutrition

The Ohio State University

2011

Dissertation Committee:

Dr. Yael Vodovotz, Ph.D., advisor

Dr. Joshua Bomser, Ph.D.

Dr. Mark Failla, Ph.D.

Dr. Steven Schwartz, Ph.D.

Luca Serventi

2011 Abstract

Cardiovascular disease is a major cause of death in the western nations that affects as many as 81.1 million citizens in the US each year (American Heart Association

2010). Recent studies showed that soy and chickpea saponins may play a role in cardiovascular disease prevention (Southon et al. 1988; Harwood et al. 1993; Matsuura

2001; Kerwin 2004; Kim et al. 2004; Yamsaengsung et al. 2010).

In this study, isolation protocol was optimized for high yield saponins extraction from food ingredients by semi-preparatory HPLC. Accurate analytical techniques were developed for identification and quantification of soy and chickpea saponins in food ingredients and biological samples by HPLC-PDA and LC-MS. Saponin extracts (in the form of isolates and mixtures) and breads developed with saponin-rich ingredients (soy flour, soy milk powder, chickpea protein isolate) were processed by in vitro digestion in order to assess saponin effect on cholesterol micellarization and their bioaccessibility in the in vitro model. Saponin-rich baked goods were developed for a broad variety of systems: pocket-type flat doughs reformulated with soy blend; soy bread added of soy saponin extract; wheat breads added of chickpea saponin isolate and soy bread reformulated with the addition of chickpea protein isolate.

Upon screening of several food ingredients by HPLC-PDA, soy and chickpea flours were chosen for saponin isolation. The isolation protocol resulted in high yield.

Furthermore, control of temperature, pH and the extraction time were optimized,

ii minimizing the conversion of DDMP to type B saponins. LC-MS detected 17 saponins in soy ingredients (A, B, E and DDMP type) while only Bb and βg in the chickpea counterparts. Conversion of DDMP to B type was observed in the protein isolates.

No significant effect of saponins on cholesterol micellarization was observed during in vitro digestion, thus suggesting an alternate path for their proposed hypocholesterolemic activity. LC-MS analysis confirmed stability of saponins during bread making with the exception of type E saponins (~30% recovery). Some loss occurred upon in vitro digestion: recovery in digesta was 60% for type A, 100% for type

B, 100% for type E and 90% for DDMP type suggesting minimal degradation and partial conversion of the DDMP type during digestion. Higher recoveries were observed for soy bread as compared to chickpea containing breads, thus suggesting a matrix effect on saponin stability and bioaccessibility. Micellarization rate was higher for DDMP type (80 vs. 60%) and low uptake of saponins by Caco-2 cells was observed with predominance of

B type (1-3% test medium).

Saponin addition affected texture of baked goods differently, depending on the product and the system chosen. Soy blend addition to pocket-type flat doughs resulted in soft, yet tough and rubbery texture. Increased “freezable” water (from 7.0 to 16 g water/100 g sample) was observed thus depicting poor plasticization of the gluten-starch network. When soy saponins were added as an extract to soy bread, a dramatic reduction of hardness during 7 days ambient storage was observed, thus suggesting anti-staling properties of these compounds. Chickpea saponins (1% addition) in the form of solvent extract resulted in harder texture of the wheat breads, significant loaf volume increase and lower “freezable” water content. The isolate contained also water soluble fiber that

iii may have contributed to such changes. Reformulation of soy bread with chickpea protein isolate resulted in harder and denser breads at 2/3 substitution.

In conclusion, a method for high yield isolation of soy saponins by semi- preparatory HPLC was optimized and a method for soy saponin identification and quantification by LC-MS was developed. Soy saponins were shown to be stable under bread making conditions and to degrade partially during in vitro digestion.

Micellarization was necessary for the uptake of most soy saponins although the uptake measured by Caco-2 cells was very low (1-3%). Saponins addition to baked goods in the form of extracts negatively affected their physicochemical properties. Nonetheless, incorporation of saponin-rich ingredients (i.e. soy blend) improved textural qualities of flat products.

Dedicated to my parents Emma and Mario

La gînta la ga’ da magnér

Acknowledgments

A four year-long experience in Columbus is about to be completed, thanks to all those professionals and friends who supported me during this time.

First of all, I would like to say thank you to my advisor, Dr. Yael Vodovotz, for giving me the unique opportunity to enter the Ph.D. program at The Ohio State

University and for guiding me through a path of challenges and responsibility. I would like to thank Dr. Failla for his precious discussions that helped me to further develop scientific skills and Dr. Chitchumroonchokchai for helping with the bioaccessibility experiments. Thanks to Dr. Mark Berhow, from USDA, for providing standards of saponins and sharing with me his expertise in the field. I would also like to thank Dr.

Schwartz for giving me the opportunity to work in his laboratory, where I learned how to be a good chemist from Greg Hostetler, Ken Riedl, Rachel Kopec and Stella Wang. I thank Aran Zouela for helping with the isolation of saponins and Dr. Chunhua Yuan for helping with the NMR experiments. Furthermore, I would like to thank Dr. Joshua

Bomser for being part of my committee and for contributing to my dissertation with his expertise and professionalism. Also, thank you to Dr. Ken Lee for supporting my professional development during the studies. I acknowledge BARD and CIFT for sponsoring my research and Adi Nudel, from The Hebrew University of Jerusalem, for sharing with me the challenges of the saponins project.

Thank you to my lab mates Alex, Amber, Jenn, Karina, Pauline, Ruth and Rachel for their help and for sharing smiles in the office and Sunny for practicing our line-up of jokes: “We are winning”. Thank you to the “Coccobello” team (Amrish, Ben, Bryan,

Jenna, Joe, Kim and Natalie) for sharing hard work, success and friendship. Also, thank you to my friends Amit, Fer, Franco, Greg, Joseph, Lisya, and Thomas for their support in everyday life. Finally, thanks to my former advisor, Dr. Elena Vittadini, for introducing me to research as a world-wide experience and thank you to my parents, for being always there.

vii

Vita

July 5, 1982 ...... Born - Fidenza (PR), Italy

September 2004 ...... Laurea Triennale in Food Science and

Technology, Università degli studi di Parma,

Italy

October 2006 ...... Laurea Specialistica in Food Science and

Technology, Università degli studi di Parma,

Italy

September 2007 to today ...... Graduate Research Associate, Department of Food Science and Technology, The Ohio State University

Publications Scazzina F, Del Rio D, Serventi L, Carini E, Vittadini E. 2008. Effect of formulation on physico-chemical properties and water status of nutritionally enhanced tortillas. Food Biophysics 3 (2): 235–240.

Serventi L, Sachleben J, Vodovotz Y. Effect of soy and water mobility of microwavable frozen doughs. In: The 9th International Conference on the Applications of Magnetic Resonance in Food Science. Challenges in a changing world. Reykjavik, Iceland, September 15-17, 2008.

Serventi L, Carini E, Curti E, Vittadini E. 2009. Effect of formulation on physicochemical properties and water status of nutritionally enhanced tortillas. Journal of the Science of Food and Agriculture 89 (1): 73-79

Field of Study Major Field: Food Science and Nutrition

viii

Table of Contents

Abstract ...... ii Acknowledgments ...... vi Vita ...... vii List of Tables ...... xi List of Figures ...... xii Chapter 1: Introduction ...... 1 Chapter 2: Statement of the Problem ...... 5 Chapter 3: Literature Review ...... 8 3.1 Saponins ...... 8 3.1.1 Structure ...... 8 3.1.2 Bioactivity of saponins ...... 10 3.2 Analysis ...... 15 3.2.1 Extraction and separation ...... 15 3.2.2 Quantification ...... 19 3.3 Saponin-rich baked products ...... 21 3.3.1 Formulation ...... 21 3.3.2 Textural properties and state of water ...... 23 3.3.3 Functional ingredients ...... 27 3.3.4 Development of functional foods...... 29 Chapter 4: Research Methods ...... 32 4.1 Specific Aim 1 Objective ...... 32 4.1.1 Saponins Extraction ...... 33 4.1.2 LC-MS Analysis ...... 34 4.2 Specific Aim 2 Objective ...... 36 4.2.1 Saponins effect on cholesterol micellarization studied by in vitro digestion ...... 37 4.2.2. Saponins bioaccessibility upon in vitro digestion ...... 37 4.3 Specific Aim 3 Objective ...... 40 4.3.1 Development of pocket-type soy doughs ...... 40 4.3.2 Development of saponin-rich breads ...... 42

4.3.3 Development of breads containing saponin-rich ingredients ...... 44 4.3.4 Texture Profile Analysis (TPA) ...... 45 4.3.5 Thermal Gravimetric Analysis (TGA) ...... 45 4.3.6 Differential Scanning Calorimetry (DSC) ...... 45 4.3.7 Nuclear Magnetic Resonance (NMR) ...... 46 4.3.8 Color Analysis ...... 47 4.3.9 Loaf Volume Measurement ...... 48 4.3.10 LC-MS ...... 48 Chapter 5: Results and Discussion ...... 49 5.1 Specific Aim 1 ...... 49 5.1.1 Saponins extraction ...... 49 5.1.2 LC-MS Analysis ...... 54 5.2 Specific Aim 2 ...... 64 5.2.1 Saponins effect on cholesterol micellarization studied by in vitro digestion ...... 64 5.2.2 Saponins bioaccessibility upon in vitro digestion ...... 66 5.3 Specific Aim 3 ...... 78 5.3.1 Development of pocket-type doughs ...... 78 5.3.2 Development of saponin-rich breads ...... 85 5.3.3 Development of breads containing saponin-rich ingredients ...... 93 Chapter 6: Conclusions ...... 97 Chapter 7: Future Studies ...... 101 References ...... 102

List of Tables Table Page

1 Structure of soy saponins (Güçlü-Üstündağ and Mazza 2007) ...... 10

2 Pocket-type dough formulations developed in this study ...... 41

3 Control bread (STD) and soy saponin-soy bread (SAP) formulations ...... 42

4 Control bread (0%) and chickpea saponin wheat breads (1 and 3%) formulations ...... 43

5 Bread formulations containing no (SOY), 1/3 (1/3) and 2/3 (2/3) chickpea proteins ...... 44

6 Saponin content of soy and chickpea ingredients indicated as μg saponins/g ingredient . 63

7 Saponin content of soy-chickpea ingredients and estimated saponin content of soy- chickpea breads...... 67

8 Physicochemical characterization of baked pocket-type products (Table 2; 4.3.1). Different letters refer to statistically different samples (α 0.05) ...... 83

9 Saponin profile and composition of the chickpea isolate fortified wheat breads (Table 4; section 4.3.2) ...... 88

10 Physicochemical characterization of the chickpea saponin fortified wheat breads (Table 4; section 4.3.2). Different letters refer to statistically different samples (α 0.05) ...... 90

11 Physicochemical characterization of the breads formulated with different amounts of soy blend and chickpea protein isolate (Table 5; section 4.3.3). Different letters refer to statistically different samples (α 0.05) ...... 94

List of Figures Figure Page

1 Structures of sapogenins: triterpenoids (a) and steroid (b) (Francis et al. 2002) ...... 9

2 Structure of group B saponins I (Bb), II (Bc), III (Bb‟), IV, V (Ba) and group DDMP (2,3 dihydro-2,5 dihydroxy-6 methyl-4 H pyranone) saponins βg, βa, γg, γa, αg (Hu et al. 2002)...... 13

3 HPLC chromatogram of soy saponins in soy protein isolates at 205 nm indicating formononetin (isoflavone) and saponins V (Ba), I (Bb), II (Bc), αg, βg and βa (Hu et al. 2002) ...... 16

4 Proposed mechanism for DDMP saponin decomposition at acidic and alkalin pH. R is the saponin B aglycone. IA and IB are the intermediate products catalyzed by acid and base respectively (modified version of figures from Heng et al. 2006) ...... 18

5 RP-HPLC elution profiles of extracts from two cultivars of soybeans (A and B) recorded by the Evaporative Light Scattering Detection (ELSD; Decroos et al. 2005) ...... 21

6 Trends of T1 and T2 as a function of the correlation time (Cornejo and Chinachoti 2003) ...... 26

7 Experimental design for aim 1 ...... 32

8 Experimental design for aim 2 ...... 36

9 Experimental design for aim 3 ...... 40

10 HPLC-PDA chromatogram at 205 nm of soy flour (A); chickpea flour (B) at 50 mg/ml, 50 µl injections; Ab and Bb standards (C), 1 mg/ml, 50 µl injections ...... 50

xii

11 HPLC-PDA chromatograms at 205 nm of protein extracts from soy (A) and chickpea (B) at 50 mg/ml. 50 µl injections ...... 51

12 HPLC-PDA chromatograms at 205 nm of soy milk powder at 50 mg/ml, 50 µl injection ...... 51

13 PDA chromatogram of chickpea flour in semi-preparatory HPLC at 205, 260, 292 nm at 100 mg/ml, 1 ml injection ...... 53

14 PDA chromatogram of soy flour in semi-preparatory HPLC at 205, 260, and 292 nm at 100 mg/ml, 1 ml injection ...... 53

15 HPLC-PDA chromatograms at 205 nm of saponin extracts: soy type A (A), soy type B (B), soy DDMP (C) and chickpea βg (D) at 1 mg/ml, 10 µl injections ...... 54

16 LC-MS spectrum of saponins in soy flour (50 mg/ml, 1 µl injection) ...... 57

17 LC-MS spectrum of saponins in chickpea flour (50 mg/ml, 1 µl injection) ...... 58

18 LC-MS spectrum of saponins in soy protein isolate (50 mg/ml, 1 µl injection) ...... 59

19 LC-MS spectrum of saponins in chickpea protein isolate (50 mg/ml, 1 µl injection) ...... 60

20 LC-MS spectrum of saponins in soy milk powder (50 mg/ml, 1 µl injection) ...... 61

21 Chromatogram at 205 nm (A) and mass spectrum (B) of equimolar solution of saponins Ab, Bb and βg, injection volume 5 μl ...... 62

22 Micellarization of cholesterol during simulated digestion of test meal (yogurt) with and without various saponins ...... 65

23 Recovery (%) of saponin groups in breads from ingredients. Data are mean ± SD, n = 3. One-way ANOVA was performed to compare recovery of each saponin type across the three breads. Presence of a different letter above the error bars

xiii

within the saponin type indicates that recovery differed significantly (α<0.05) in the different matrices...... 69

24 Representative LC-MS spectra in soy bread, chyme and filtered aqueous fraction generated during simulted digestion and Caco-2 cells incubated with diluted aqueous fraction for 4 h...... 71

25 Recovery (%) of saponin groups in chyme after in vitro digestion of breads. Data are mean ± SD, n = 5. One-way ANOVA was performed within each bread type. The presence of different letters above bars for saponin type indicates significantly different mean (α<0.05)...... 72

26 Partitioning (%) of saponins in soy-chickpea breads in aqueous fraction during small intestinal phase of digestion in presence of bile extract. Data are mean ± SD, n = 5. One-way ANOVA was performed within each bread type; different letters refer to significant difference (α<0.05)...... 73

27 Uptake (% medium) of saponins from digested soy-chickpea breads by Caco-2 cells. Cultures of Caco-2 cells were exposed to aqueous fraction generated during simulated digestion of soy-chickpea breads. Total saponin content of test media was 19 ± 1, 14 ± 1 and 9.4 ± 0.3 nmol/ mg cell protein for breads containing soy alone or mixed with 1/3 and 2/3 chickpea, respectively. Data are means ± SD for pooled results for the specific saponin in diluted aqueous fraction after digestion of the three breads (n = 5 independent cultures for each digested bread). One-way ANOVA was performed to compare relative extent of uptake of saponins; different letters above bars indicate significant differences in the extent of uptake (α<0.05)...... 77

28 TGA thermograms of raw doughs with different % of soy blend added ...... 79

29 TPA of pocket-type baked formulations (Table 1, section 4.3.1) ...... 80

30 DSC thermograms of pocket-type product formulations (Table 1; section 4.3.1) ...... 81

31 1H spectra (obtained using CPMG) used in the calculation of the spin-spin relaxation time of fresh doughs. Peaks 1, 2 and 3 were attributed to water, carbohydrate and lipid protons, respectively (Lodi et al. 2007) ...... 82

xiv

32 Texture Profile Analysis of fresh and stale soy bread (STD) and soy saponins containing soy bread (SAP), α 0.05 ...... 86

33 Thermogram of fresh soy bread (STD) and soy saponins containing soy bread (SAP) and their state of water: moisture content (MC) and "freezable" water ...... 86

34 Hardness (blue dots) and specific volume (red dots) of the chickpea saponin fortified wheat breads (Table 4; section 4.3.2) ...... 92

35 Chewiness (blue dots) and specific volume (red dots) of the breads formulated with different amounts of soy blend and chickpea protein isolate (Table 5; section 4.3.3) ...... 96

Chapter 1

Introduction

Cardiovascular disease is a major cause of death in the western nations affecting as many as 81.1 million citizens in the US each year and is strongly associated to excessive lipid intake (American Heart Association 2010). Treatment of lipid induced disease is commonly based on drug therapy. However, safety concerns regarding the effect of pharmaceutical agents (i.e. liver dysfunction) and higher health care costs

(Milner 2000) require alternatives to pharmacological therapies. Considerable attention has been given to the potential of dietary intervention as a tool for prevention and treatment of hypercholesterolemia (Milner 2000; Van Kleef et al. 2005). The demand for a food based prevention of cardiovascular disease is continuously increasing, thus leading the development of “functional foods”: foods that that exert a beneficial effect on host health and/or reduce the risk of chronic disease beyond basic nutritional functions

(Charalampopoulos et al. 2002).

Research on “functional foods” as a means to prevent hypercholesterolemia is based on the significant effect of legumes on cholesterol metabolism. Legumes exhibit hypocholesterolemic effects in human and animal models (Setchell and Cassidy 1999;

Erdman 2000; Merz-Demlow et al. 2000; Lucas et al. 2001; Lee et al. 2005). Legumes are rich sources of bioactive phytochemicals such as isoflavones, saponins, lignans, and

1 phytates in addition to proteins of high biological quality. Saponins ability to bind bile acids, the structural similarity to cholesterol and the surfactant behavior suggest that saponin-rich foods may prevent cardiovascular diseases by lowering cholesterol bioaccessibility (Kerwin 2004; Kim et al. 2004).

Chickpea and soy contain high levels of saponins: 1-3.5% of the dry weight

(Kerem et al. 2005). Saponins are a structurally and biologically diverse class of glycosides of steroids and triterpenoids that are widely distributed in many plant food products. The structural diversity of saponins is mainly due to their conjugated moieties

(Francis et al. 2002). Both soy and chickpea saponins contain primarily triterpenoidal saponins conjugated to a maltol residue (2,3 dihydro-2,5 dihydroxy-6 methyl-4 H pyranone, also known as DDMP) that absorbs light at 292 nm and is easily hydrolyzed, thus converting to type B saponins, by increasing temperature and/or non-neutral pH (Hu et al. 2002; Hu et al. 2004a; Kerem et al. 2005; Lee et al. 2005; Heng et al. 2006).

Saponins are tolerated as was demonstrated in rodent fed diets containing up to 3% by weight soy saponins with no adverse effect on growth, organ weights or intestinal morphology (Hu et al. 2004a; Lee et al. 2005). These studies have been limited to soybean and soy derived products, while other commonly consumed legumes such as chickpeas (Cicer arietinum L.) are yet to be investigated. Such discrepancy was probably due by the increasing demand of soy based foods, in contrast to the lower popularity of chickpea products, mainly limited to Middle Eastern countries. Chickpea is a palatable bean and its mild flavor is highly desirable (Williams and Singh 1987). The prevalent cultivar in Israel is cv. Yarden (kabuli type) and ICC 4958 (a wilt-resistant desi type cultivar) is globally cultivated.

Appropriate methods for extraction, separation and quantification of saponins are required to study their role in cholesterol metabolism. Sensitivity and time efficiency are necessary to obtain reliable information, especially when working with low saponin systems such as foods and biological samples.

Extraction of saponins from soy and soy derived products is based on the solvent extraction (hydro alcoholic solution) of the raw material. Berhow et al. (2002) added a further step to reduce fat content of soybeans (about 20% w/w, Berhow et al. 2006) consisting in a Soxhlet extraction with hexane. The extract so obtained is then re dissolved in 80% aqueous methanol and injected into preparative HPLC equipped with a reversed phase C18 column. The mobile phase is typically a solution of acidified water

(solvent A) and acidified acetonitrile (solvent B). Gradient elution is used increasing solvent B percentage from 40 to 100% (Murphy et al. 2008) or from 30 to 80% (Berhow et al. 2006) to separate the different types of saponins (A, B and DDMP types). PDA detector is used to detect saponins absorbance at 205 and 292 nm. Less information is available on the extraction of saponins from chickpea beans. Kerem et al. (2005) proposed a microwave-assisted extraction. Extraction conditions (temperature, pH, dielectric properties of the solvent) play a fundamental role in DDMP stability. Complete quantification of soy saponins types A and B has been accomplished in several papers by

PDA (Berhow et al. 2006; Murphy et al. 2008); both methods are long (about 50 minutes) therefore DDMP might degrade during the experiment. PDA has low sensitivity so it is of difficult application to biological samples and food that typically have low saponins concentration. Mass spectroscopy has been applied to improve the study of these compounds but the only method developed to quantify all types of soy and chickpea

3 saponins was extremely long (80 minutes) thus compromising the DDMP stability

(Decroos et al. 2005).

Saponin-rich ingredients such as soy flour and soy milk powder were incorporated into baked products such as bread resulting in textural changes. Loaf volume decreased with soy addition due to higher density of the product (Vittadini and

Vodovotz 2003; Lodi et al. 2007; Lodi and Vodovotz 2008; Nilufer and Vodovotz 2008).

High water holding capacity of soy proteins has been considered responsible of increased dough hydration, thus higher density of the bread (Kinsella 1979; Doxastakis and others

2002). Another legume, chickpea, has been studied as a partial substitute of wheat flour in bread making (Figuerola et al. 1987; Fernandez and Berry 1989). Substitution at as high as 15% did not affect the textural properties and, in fact, resulted in high acceptability of the bread (Figuerola et al. 1987). Thermal analyses and nuclear magnetic resonance confirmed these findings by showing higher dough hydration in soy products

(Lodi and Vodovotz 2008; Nilufer and Vodovotz 2008). Such findings suggested a threshold for wheat substitution with legume-based flour of about 10-15%, below which texture can benefit of enhanced hydration (Kinsella 1979) and stabilizing effect of soy polar lipids on the gluten-amylose network (Gan et al. 1995) resulting in lower staling rate (Vittadini and Vodovotz 2003). Substitution rate higher than 15% resulted in unacceptable loaf volume and chewy texture (Vittadini and Vodovotz 2003; Nilufer and

Vodovotz 2008).

Such properties suggest that soy and other legumes, such as chickpea, could be incorporated into baked goods formula to deliver hypocholesterolemic activity without deleterious textural changes.

Chapter 2

Statement of the Problem

Cardiovascular disease is a major cause of death in the western nations that affects as many as 81.1 million citizens in the US each year (American Heart Association

2010). The raise of safety concerns regarding the use of pharmaceutical agents brought considerable attention to the potential of dietary intervention as a tool for prevention and treatment of hypercholesterolemia (Milner 2000; Van Kleef et al. 2005). Recent studies showed that soy and chickpea saponins might exert a main role in preventing cardiovascular disease (Southon et al. 1988; Harwood et al. 1993; Matsuura 2001;

Kerwin 2004; Kim et al. 2004; Yamsaengsung et al. 2010) but an in depth understanding of how saponins affect cholesterol metabolism and of saponins bioaccessibility is yet to be achieved. Furthermore, accurate analytical techniques need to be optimized in order to improve the identification and quantification of saponins in biological and food samples.

Legume based ingredients have been recently used to develop functional breads meant to prevent cardiovascular disease (Vittadini and Vodovotz 2003; Nilufer et al.

2008) but their saponin content was estimated to be very low, in the order of 0.1-0.3 %

(Kerem et al. 2006). Development of saponin-rich baked goods is a critical step in the investigation of saponins effect on cholesterol metabolism and on physicochemical properties of baked products.

Hypothesis: the use of saponin-rich ingredients in the development of baked goods will deliver nutritional benefits and acceptable textural qualities. Saponins will exert hypocholesterolemic activity on in vitro model that will be efficiently identified and quantified with accurate analytical techniques.

The specific objectives of this work were:

- 2.1 Aim 1: To develop rapid and sensitive methods for separation, identification

and quantification of soy and chickpea saponins in plant materials and bread

In order to study the stability and bioavailability of saponins in the food matrix we

developed an efficient method for the isolation of saponins from food ingredients

by column chromatography. Furthermore, we developed an accurate method for

identification and quantification of soy and chickpea saponins utilizing LC-MS.

- 2.2 Aim 2: To determine the influence of saponins on micellarization of 14C-

cholesterol using in vitro digestion and to determine saponins stability,

bioaccessibility and intestinal cell uptake using the coupled model in vitro

digestion/Caco-2 human intestinal cells

We hypothesize that saponin isolates from soy and chickpea will reduce

cholesterol absorption in vitro. Saponin isolates, as well as saponin-rich baked

goods, were analyzed. In vitro digestion was performed to investigate the

potential for saponins to reduce cholesterol incorporation into micelles during

digestion. The coupled model in vitro digestion/Caco-2 human intestinal cells was

used to investigate saponin bioaccessibility in terms of micellarization, uptake and

stability.

- 2.3 Aim 3: To develop functional baked goods with elevated saponin content

We hypothesize that saponin isolates and saponin-rich ingredients can be

incorporated into several baked goods formulations without altering the quality of

the products. Various saponin-rich products were investigated including pocket-

type doughs using a soy blend, wheat and soy breads added of saponin extracts,

and breads containing a blend of chickpea and soy. All new products were

assessed for texture, water dynamics and saponin content/distribution.

Chapter 3

Literature Review

3.1 Saponins

3.1.1 Structure

Saponins are a diverse class of glycosides commonly found in several plants, mostly legumes such as soy and chickpea (Francis et al. 2002; Kerem et al. 2005). They occur in a large number and a wide variety of plants but only about 28 of these are regularly used as food by man. The more commonly eaten of these are soybeans, chickpeas, peanuts and spinach (Oakenfull 1981). Saponins are naturally occurring surface-active glycosides. They are mainly produced by plants, but also by lower marine animals and some bacteria (Riguera 1997; Yoshiki et al. 1998). They derive their name from their ability to form stable, soap-like foams in aqueous solutions.

Saponins are steroid (neutral sapogenin) or triterpenoid (acid sapogenin) compounds glycosidically linked to a sugar moiety (Price et al. 1987; Yoshiki et al. 1998,

Figure 1). The sugar moiety can be attached at the C3 position of the A ring

(monodesmosidic saponins); another sugar can be attached at the C26 position

(bidesmosidic saponins). The sugar moiety usually exists as galactose, glucose, glucuronic acid, methylpentose, xylose, and/or rhamnose (Kensil 1996, Table 1). Based

8 on the aglycone and on the sugar moiety, saponins are classified into different groups.

For example, soy contains 4 different groups: A (aglycone A and two sugar moieties), B

(aglycone B and one sugar moiety), E (aglycone B oxidized and one sugar moiety), and

DDMP (aglycone B with one sugar mojety and one 2,3 dihydro-2,5 dihydroxy-6 methyl-

4 H pyranone, unit (DDMP; Figure 2, Berhow et al. 2002). Chickpea contains mainly βg saponin (a DDMP type) and lower amounts of Bb and Be (Kerem et al. 2005).

Figure 1 - Structures of sapogenins: triterpenoids (a) and steroid (b) (Francis et al. 2002)

Table 1 - Structure of soy saponins (Güçlü-Üstündağ and Mazza 2007)

3.1.2 Bioactivity of saponins

Saponins role in plants is mainly protection against insects, microorganisms, and fungi and are classified as phytoanticipins (saponins released by plant‟s enzymes in response to tissue damage) and phytoprotectants (general protection). They are present more abundantly in immature plants. Some scientist believes that saponins can be a source of monosaccharide (Morrissey and Osbourn 1999).

Bioactivities in herbivorous/omnivorous animals are several and affect organoleptical quality of saponins containing food, as well as membrane stability, nutrient uptake, glycemia, nervous system, tumors growth, proteins and lipids digestion.

Saponins are responsible for the bitter taste of foods such as tea, legumes and chocolate, along with flavones, isoflavones, catechins and caffeine (Drewnowski and

Gomez-Carneros 2000; Aldin et al. 2006).

Membrane activity is due to the high affinity of the saponins aglycone

(sapogenin) to the membrane sterols, particularly cholesterol with which they form insoluble complexes in the membrane plane (Bangham and Horne 1962). This may explain the hemolytic activity of these compounds, especially monodesmosidic saponins, due to the higher exposure of the aglycone to the membrane surface (Glauert et al. 1962).

Saponins also exert a hypotensive effect caused by the inhibition of the calcium dependent potassium channels with consequent membrane depolarization, altered calcium metabolism (PKC inhibited) and channels deionization (Oh et al. 2001).

Saponins exerted toxic activity in insects and fish (hemolytic activity) but there was no evidence of their activity on humans when ingested orally (Francis et al. 2002).

Nutrient uptake through the gastrointestinal tract is influenced by the modified membrane equilibrium described above with consequent reduction of TPD (Transmural

Potential Difference) which inhibits the active transport and increases the passive diffusion of nutrients not usually absorbed (for example, less SGLT1-mediated glucose uptake and increased L-glucose uptake) (Gee et al. 1989; Onning et al. 1996). Saponaria and tomatine showed this effect (Carlson 2009).

Glycemia has been shown to decrease with addition of fenugreek saponins (Petit et al. 1993). Possible mechanisms involve higher insulin secretion by higher β–cells stimulation or most likely a suppression of glucose transport from stomach to intestine and an inhibition of glucose uptake through intestinal membranes of enterocytes

(Matsuda et al. 1999). Fiber-like effect could be another explanation since fenugreek has high viscosity and high fiber content (similar to guar gum, Yoshikawa 2001).

Ginseng saponins showed neurotrophic and neuroprotective effects most likely due to the membrane stabilization (altered calcium and potassium channels) in brain damaged rats (Rudakewich 2001).

The α-tomatine saponin found in tomatoes showed an inhibition of tumor cell line growth and apoptosis. The mechanism apparently involves 2 steps: 1st) non specific toxicity due to a detergent action with consequent cells aggregation (Mimaki et al. 2001);

2nd) specific cytotoxicity due to the sugar moiety (after the aglycone allowed saponins to traverse the cell membrane) which involves an apoptotic process with the activation of caspases, apoptotic enzyme (Kuroda et al. 2001).

Another possible bioactivity is the antioxidant activity of group B soy saponins due to the presence of DDMP (2,3 dihydro-2,5 dihydroxy-6 methyl-4 H pyranone, Figure

2) which scavenges superoxides thus inhibiting biomolecular damage (Yoshiki and

Okubo 1995; Yoshiki et al. 1998; Hu et al. 2002).

Figure 2 - Structure of group B saponins I (Bb), II (Bc), III (Bb‟), IV, V (Ba) and group DDMP (2,3 dihydro-2,5 dihydroxy-6 methyl-4 H pyranone) saponins βg, βa, γg, γa, αg (Hu et al. 2002)

Protein uptake can be inhibited (so far it was shown only for soy proteins) by the formation of sparingly absorbable protein-saponins complexes which prevent chymotrypsin attack on the proteins (Potter et al. 1993). Soyasaponin seemed to slightly activate α-chymotrypsin, but the effect was not great (Ikedo et al. 1996).

Cholesterol functions in mammals are several: membrane component (fluidity), bile acids synthesis (precursor, it‟s oxidized in liver and eventually conjugated to aminoacids such as taurine and glycine), lipid soluble vitamins metabolism

(comicellarization), and vitamin D synthesis (precursor). Excessive levels of cholesterol

13 may lead to coronary heart disease, which is the major cause of death in the western nations that afflicts as many as 16.8 and 0.5 million citizens in the US and Israel, respectively, each year (American Heart Association 2010). Lipid metabolism is influenced by saponins of different sources (soy, garlic, and fenugreek) and at least 3 possible mechanisms are described in the literature (Southon et al. 1988; Harwood et al.

1993; Matsuura 2001). It has been hypothesized that saponins compete with cholesterol for micellarization and receptors by forming large mixed micelles. The high hydrophobicity of saponins enlarges micelles reducing their solubility and thus cholesterol absorption and resulting in an extended residence time of cholesterol in the intestinal lumen. A second possible hypocholesterolemic mechanism is the increased cholesterol conversion into bile acids by oxidation in the liver and subsequent elimination via fecal of about 5% bile while the rest of the bile is re absorbed (Oakenfull and Sidhu

1990). The third mechanism is less specific as it involves a higher gastric emptying due to a fiber-like effect of saponins (some saponins are very viscous, Oakenfull and Sidhu

1990). To be noted that the hypocholesterolemic effect is exerted by saponins and fiber, while the hypolipidemic effect by fiber alone.

Further research is needed to elucidate the extent saponins reduce cholesterol absorption and, if so, the mechanisms. Furthermore, limited information is available on saponins bioaccessibility and bioavailability (Hu et al. 2004a). Knowledge obtained from such researches will lead to more specific dietary intervention as a tool for prevention and treatment of hypercholesterolemia (van Kleef et al. 2005).

3.2 Analysis

Proper analytical methods are fundamental for the study on saponin effect on cholesterol metabolism. Isolation, identification and quantification of saponins in food and biological samples must be optimized in order to achieve accurate measurements.

Although methods have been published to describe extraction, separation and quantification of saponins from soybeans and soy products, a lack of appropriate methods to analyze and quantify DDMP type saponins has contributed to the scarcity of data on these compounds typically found in soy and chickpea.

3.2.1 Extraction and separation

Extraction of saponins from soy and soy derived products was described in details in several studies (Berhow at al. 2002, 2006; Hu et al. 2002; Murphy et al. 2008). The extraction begins by stirring the raw material at room temperature in a hydro alcoholic solution (70% ethanol) in order to solubilize the saponins. The solution obtained is then filtered and centrifuged. The supernatant is condensed in a rotoevaporator at <30°C under reduced pressure to evaporate the organic solvent and then freeze-dried to remove the water. Berhow et al. (2002) added a further step to remove fat from the soybean preparations (about 20% w/w) consisting of a Soxhlet extraction with hexane.

The extract is then re dissolved in 80% aqueous methanol and injected into a preparative HPLC equipped with a reversed phase C18 column. The mobile phase is typically a solution of 0.05% trifluoroacetic acid (TFA) in water (solvent A) and 0.05%

TFA in acetonitrile (solvent B). Gradient elution is used, increasing solvent B percentage from 40 to 100% (Murphy et al. 2008) or from 30 to 80% (Berhow et al. 2006) to

15 separate the different types of saponins (A, B and DDMP types). PDA detector is used to detect saponins absorbance at 205 and 292 nm (Figure 3). The fractions obtained are then condensed by rotoevaporator and freeze-dried.

Figure 3 - HPLC chromatogram of soy saponins in soy protein isolates at 205 nm indicating formononetin (isoflavone) and saponins V (Ba), I (Bb), II (Bc), αg, βg and βa (Hu et al. 2002)

Less information is available on the extraction of saponins from chickpea beans.

Kerem et al. (2005) proposed a microwave-assisted extraction. Grounded chickpea seeds were defatted with hexane and further dissolved in 70% aqueous ethanol and processed in a microwave oven at 60°C for 20 minutes. Preparative HPLC was used similarly to the methods described by Berhow, Hu and Murphy using acidified water with formic acid

16 and methanol as solvents. Samples were kept at -20°C until analysis. Saponin βg was the main type found in chickpea.

It is important to maintain a temperature as low as possible to preserve the DDMP type saponins which are the most abundant in both soy and chickpea and are heat labile.

Heng et al. (2006) investigated in depth the stability of DDMP saponins as a function of temperature, time and chemical environment (ethanol, pH). They found that DDMP is unstable at temperature above 30°C for more than one hour during the extraction procedure. Additionally, pH plays an important role in DDMP stability since either acidic or basic condition of the matrix catalyzes the hydrolysis of the DDMP type with the release of maltol and the conversion of the saponin to type B (proposed mechanism in

Figure 4).

Figure 4 - Proposed mechanism for DDMP saponin decomposition at acidic and alkalin pH. R is the saponin B aglycone. IA and IB are the intermediate products catalyzed by acid and base respectively (modified version of figures from Heng et al. 2006)

Ethanol preserved the DDMP from this conversion possibly because of lower dielectric constant of ethanol compared to water (24 vs. 82 at 20°C; Heng et al. 2006) thus ethanol is a less favorable environment for the formation of charged intermediates such as IA, IB and III (Figure 4). Extraction should be carried at temperature <30°C and the acidic solvent should be evaporated as rapidly as possible to preserve the DDMP saponins.

3.2.2 Quantification

Methods for quantification of soy saponins types A and B have been described previously (Berhow et al. 2006; Murphy et al. 2008). HPLC-PDA has been used to quantify saponins by using group A soyasaponin fraction as standards to quantify the group A saponins and Bb soyasaponin standard to quantify the type B saponins (Berhow et al. 2006). HPLC-PDA has been used to quantify some DDMP type saponins using standards of saponin αg, βg and βa (Murphy et al. 2008). Both methods are long (about

50 minutes) therefore DDMP may degrade during the analysis.

PDA has low relative sensitivity so it is not appropriate for cell studies such as

Caco-2 cells and for food samples which typically have low saponins concentration (i.e. legume based breads and chows). Mass spectroscopy has been used to improve the study of these compounds. The majority of LC-MS methods were developed for the analysis of type A saponins (Shiraiwa et al. 1991; Gu et al. 2002; Dalluge et al. 2003) and type B saponins (Berhow et al. 2002; Gu et al. 2002; Dalluge et al. 2003). Few methods were optimized for the analysis of DDMP type from soybeans (Kudou et al. 1993; Decroos et al. 2005; Sagratini et al. 2009).

Decroos et al. (2005) developed a LC-MS method to identify and quantify soy saponins profile, DDMP types included, based on their molecular weight and mass/charge ratio (m/z). Standards of saponin Ab, Bb, αg and βg were used for quantification. The method was efficient but extremely long (80 minutes, Figure 5) therefore not suitable for the study of DDMP type saponins. Full scan mass spectrum was obtained with an ion-trap mass spectrometer. Such method was validated for food samples that are rich in saponins (soybeans) but not for biological samples which require higher sensitivity and short time. Accurate identification of compounds was obtained with a mass spectrometer equipped with a time of flight analyzer, performing full scans over a broad mass/charge range (200-1500 m/z, Berhow et al. 2002; Gu et al. 2002;

Decroos et al. 2005; Sagratini et al. 2009). This instrumental setting allows accurate identification of compounds, necessary when co elution occurs, but lacks of sensitivity, required for low concentration samples. Triple quadrupole analyzer was used for quantification of soy saponins in soy nutraceutical tablets (Dalluge et al. 2003). Such setting resulted in high sensitivity but long run time (65 minutes), thus raising concerns on DDMP saponins stability in the acidified (with TFA) solvents.

Figure 5 - RP-HPLC elution profiles of extracts from two cultivars of soybeans (A and B) recorded by the Evaporative Light Scattering Detection (ELSD; Decroos et al. 2005)

3.3 Saponin-rich baked products

3.3.1 Formulation

Bread is a staple food around the world (Mondal and Datta 2008). Typically dough consists of 62% flour, 31% water, 2.5% sugar, 2.0% shortening, 1.25% leavening agent yeast, and 1.25% salt (Mondal and Datta 2008) and each ingredient has specific functionality in the dough (Mondal and Datta 2008). Additives are commonly included into the basic formulation of baked goods due to today‟s mechanization, large scale production and increased consumer demand for high quality, convenience and longer shelf life (Stampfli and Nersten 1995).

Emulsifiers are the most common additives used to improve bread texture and shelf life. Emulsifiers belong to the class of compounds called surface-active agents or surfactants and their structure consists of a hydrophobic tail and a hydrophilic head

(Stampfli and Nersten 1995). Surfactants are classified, based on the composition of their head, in the following categories: anionic (based on permanent anions, for example diacetyl tartaric acid esters of mono- and di-glycerides); cationic (based on permanent cations, not used in foods); Zwitterionic (positive and negative charge coexisting on the same molecule, for example lecithin); non-ionic (no permanent charges, for example mono- and di-glycerides, Rosen 2004). The great interest of the bakery industry for surfactants arises from their potential use in a myriad of applications: dough strengthener, improvement of dough hydration and water retention, improvement of gas retention and crumb structure, improvement of slicing characteristics of the bread, and longer shelf life

(Stampfli and Nersten 1995). Nowadays, most of the commonly used surfactants are of synthetic origin; but an increasing interest towards natural substitutes has brought the attention to the development of plant based products (Kjellin and Johansson 2010).

Lecithin and saponins are the two major natural surfactants that have been studied.

Saponins are non-ionic surfactants that showed interesting activity in food systems. Two types of saponins, Yucca schidigera and Quillaya saponaria, have been used for their foaming ability in food systems. Their applications included: emulsifier in oil-based flavors for candies; emulsifier in mayonnaise to prevent oil separation; leavening agent in the bakery industry; emulsifier in oil in water emulsions; stabilizer of coffee creamer; dispersing agent for coating waxes (Kjellin and Johansson 2010).

Nonetheless, application of these saponins as surfactant is limited by their weak action and their low abundance in plants (Kjellin and Johansson 2010) as well as by their bitter flavor (Price et al. 1985; Ridout et al. 1991; Aldin et al. 2006).

3.3.2 Textural properties and state of water

Consumer acceptability of fresh bread is mostly determined by its texture (Brady and Mayer 1985). Texture Profile Analysis (TPA) utilizing an Instron is performed to mimic the mechanical changes occurring to the food during mastication. Hardness, chewiness, springiness and cohesiveness are instrumental parameters used to describe the sensorial experience that humans perceive during eating.

Formulation of bakery products significantly contributes to textural properties.

The structure of baked products is based on the gluten-amylose network developed during mixing, proofing and baking. Gluten exhibits cohesive, elastic and viscous properties that combine the extremes of its two components, glutenin and gliadin

(Gallagher et al. 2004). The gluten-amylose system is a major determinant of the important properties of dough (extensibility, resistance to stretch, mixing tolerance, gas holding ability), which encloses the starch granules and fiber fragments (Gallagher et al.

2004). Complete substitution of wheat flour with gluten-free flours (soy, barley, chickpea, buckwheat and so on) resulted in lower loaf volume and decreased elasticity

(Fernandez and Berry 1989; Gallagher et al. 2004; Nilufer et al. 2008) due to an insufficient interaction of proteins and starch with water and increased fiber content

(Gallagher et al. 2004; Fessas et al. 2008; Nilufer et al. 2008). Gluten substitution can result in acceptable texture when various processing factors such as order of addition of

23 ingredients and interactions between components are considered (Friend et al. 1982;

Vittadini and Vodovotz 2003; Lodi et al. 2007; Lodi and Vodovotz 2008; Nilufer et al.

2008; Serventi et al. 2009). For example, the water binding capacity of soy proteins enhanced starch gelatinization thus resulting in acceptable texture especially for thin baked products.

Water plays a critical role in the textural qualities of baked goods, either by enhancing the molecular mobility of polymer chains or by acting as a coordination agent between them (Chinachoti and Vodovotz 2000). Water dynamics can be elucidated with a calorimetric approach (Schiraldi et al. 1996; Fessas and Schiraldi 2001) such as

Thermogravimetric Analysis (TGA). In dough systems water was observed to partition between coexisting phases. The vaporization rate revealed that water is released in two main steps, the first corresponding to a mere diffusion process through starch, the second being related to the desorption of water more tightly bound to the gluten network (Fessas and Schiraldi 2001). Overall dough moisture, extent of mixing and dough resting time after mixing all can modify the water partitioning between phases and the way water is released during the analysis scan (Fessas and Schiraldi 2001).

Other methods such as Differential Scanning Calorimetry (DSC) can be used to investigate thermal properties such as first order transitions (involving changes in heat capacity and latent heat flow) and second order transitions (involving only changes in heat capacity, Schiraldi 1996). First order transitions (melting or recrystallization) include amylopectin melting (studied to investigate starch gelatinization) and ice melting (studied to measure the amount of “freezable” and “unfreezable” water). The amount of water that is not mechanically entrapped or bound is called “freezable” water (%FW). High

24 amounts of %FW depict poor hydration of the system which may lead to poor textural qualities of the baked products (Fessas and Schiraldi 1998; Baik and Chinachoti 2000).

The amount of “freezable” water is also influenced by the moisture content (a minimum of 20% moisture content is needed in order to observe ice melting in the DSC for most high-starch systems) thus caution must be used when interpreting %FW data and its significance. Second order transition of interest for food technologists is glass transition, describing the transition occurring in amorphous solids, from glassy (rigid structure) to rubbery (more mobile, Fessas and Schiraldi 1998) and provides useful information on state of the polymers during storage or change in formulation (Matuda et al. 2005).

Proton Nuclear Magnetic Resonance (1H NMR) has been commonly used to study the impact of water mobility and polymer mobility on macroscopic parameters such as

1 texture (Engelsen et al. 2001; Seow and Teo 2006). H NMR relaxation tests T1 and T2 are used to measure molecular mobility of protons which is associated to water mobility thus providing information on structural motions at molecular level in glassy, rubbery and crystalline solids.

The T1 relaxation time provides information about the spin-lattice and the interaction between water protons and the lattice. T2 measures the spin-spin relaxation time which represents the interaction of the protons with the surrounding protons.

Mechanisms contributing to spin–lattice relaxation processes also affect spin–spin contributions. T1 and T2 values vary depending on the correlation time, τC, which is the time spent by molecules in a given orientation and enables the determination of the state of the compound analyzed: non-viscous liquid, viscous liquid, non-rigid solid, rigid lattice (Figure 6, Cornejo and Chinachoti 2003). Short correlation time is associated to

25 high moisture systems as increasing water in the formulation induces longer relaxation times. Changes in the formulation, such as addition of non-native proteins and fiber, have different effects based on the characteristics of the baked product. In the case of soy tortillas and soy-almond bread, the increased amount of fiber and non-native proteins resulted in significantly shorter T1 and slightly longer T2 (Lodi et al. 2007; Serventi et al.

2009). Such molecular changes suggested that these systems were comparable to viscous liquids (Figure 6) and shifted toward a more solid-like system due to the addition of fiber and soy proteins, probably due to higher water binding.

Figure 6 - Trends of T1 and T2 as a function of the correlation time (Cornejo and Chinachoti 2003)

3.3.3 Functional ingredients

Saponins occur in a large number and a wide variety of plants but only about 28 of these are regularly used as food by man. The more commonly eaten of these are soybeans, chickpeas, peanuts and spinach (Oakenfull 1981). Saponin content of these plant materials is low: soybeans contain 0.22-0.47% saponins; chickpeas contain 0.23%

(Güçlü-Üstündağ and Mazza 2007). Therefore, saponin content in the diet is very low. A soy bread developed in our laboratory, however, contains 0.1-0.3% saponins (Kerem et al. 2006).

Legumes are a natural source of high quality proteins (Jones and Divine 1944) and saponins (Potter 1998; Zhang et al. 2003) that have been extensively used to increase the biological value of baked products such as bread (Volz et al. 1945; Yáñez et al.

1982). More recently, legume proteins have been incorporated into bread formulation in the form of soy flour and soy milk powder (Nilufer et al. 2008) with no significant adverse effect on textural quality as shown by thermal analyses (Vittadini and Vodovotz

2003) and nuclear magnetic resonance (Lodi et al. 2007). Higher water holding capacity of soy proteins has been shown to enhance dough hydration (Kinsella 1979; Doxastakis and others 2002) resulting in soft and elastic texture as depicted by thermal analyses of soy bread (Vittadini and Vodovotz 2003; Lodi and Vodovotz 2008).

Chickpea flour addition increased crumb hardness and resulted in darker, yellower crumb (Yamsaengsung et al. 2010). Wheat flour substitution of more than 10% chickpea flour significantly decreased organoleptical quality of bread resulting in harder, less elastic and less acceptable product (Fernandez and Berry 1989). Similar findings were reported for 15% substitution (Figuerola et al. 1987). These authors did not report

27 specific explanations to these changes. Previous studies on soy bread can serve as model system for chickpea addition due to the similarities between soy and chickpeas: legumes containing no gluten but high water binding capacity protein; no starch but oligosaccharides and higher fiber (Kinsella 1979). The reduction of gluten content and the high water binding capacity of legume proteins developed a dense matrix thus leading to a firm and less elastic texture; this effect was particularly relevant for high volume matrices such as bread.

Previous studies on the enrichment of foods with saponin isolates reported emulsifying activity in baked products (Martin and Briones 1999; Kjellin and Johansson

2010) along with intense bitterness (Price et al. 1985; Ridout et al. 1991; Aldin et al.

2006). Price et al. (1985) attributed bitterness of peas to soy saponin Bb and determined a linear relationship between the concentration of soy saponin Bb over the range of 0.03% to 1% and the bitterness sensory response. Drewnowski and Gomez-Carneros (2006) measured the recognition threshold for soy germ, flakes and isolate extracts diluted with either water or milk. Panelists recognized bitterness at lower concentration in the soy flakes extracts, which contained the most DDMP saponin among all extracts (62 vs. 10

µmol DDMP saponin/l 70% ethanol). The authors suggested that the malonyl-β-glucoside isoflavone and the DDMP saponins may be the source of bitterness of these extracts.

Interestingly, recognition threshold was higher in milk than in water (338 vs. 8 µM saponins from flakes extract; Drewnowski and Gomez-Carneros 2006) indicating that the bitterness of the extracts was less detectable in milk. Therefore, saponins confer bitterness at concentration typically found in legume based bread (0.1-0.3 % w/w; Kerem

28 et al. 2006) which could be attenuated by the presence of a viscous matrix (as is milk compared to water).

3.3.4 Development of functional foods

Appropriate delivery system must be chosen in order to assure saponins stability throughout processing. In the development of a functional food two main steps must be taken into account:

- identify the relationship between food component and health benefit;

- develop suitable food vehicle for bioactive component (Clydesdale 2004).

Relationship between legume proteins and lowered cholesterolemia has been demonstrated by several studies on animal and human subjects (Matsuura 2001; Setchell

2001). Hypocholesterolemic activity of saponins has been previously investigated by

Caco-2 cells (Carlson 2009) but further studies are needed for a complete understanding of the effect of different saponin types on the human metabolism. Stability, bioaccessibility and intake of these compounds need to be examined in the gastrointestinal tract (chyme, micelles, uptake across intestinal epithelial cells) by Caco-2 cells (Hu et al. 2004a; Carlson 2009). Saponins are well tolerated as was demonstrated in rodents fed diets containing up to 3% by weight soy saponins with no adverse effects on growth, organ weights, or intestinal morphology (Lee et al. 2005; MacDonald et al.

2005). Once bioaccessibility and activity of saponins is established a food vehicle can be developed.

A previous study (Otsuki 2000) suggested that ingestion of foods containing protein and lipids induces release of bile into the duodenum thus legume-based bread could be a suitable food vehicle for saponins.

The mentioned bread must be effective. Effectiveness of a functional food depends upon its efficacy and compliance. Efficacy is the extent to which a bioactive component exerts its intended activity. Compliance is the degree to which a consumer adheres to the recommended usage of the functional food. Efficacy is influenced by stability and bioavailability of the bioactive components. Stability is important since a compound has to be stable in the food matrix if it is to be functional at the time of consumption. Bioavailability is fundamental to have the component exerting its activity at the intended moment and at the desired level.

Stability and bioavailability are influenced by several parameters:

- physical form (solid, liquid, crystalline, amorphous, microencapsulated, coated)

(Clydesdale 2004);

- chemical form (in example, ferrous iron is more bioavailable than ferric iron but it

causes more oxidative reactions (Fairbanks 1994);

- diet effect (nutrients ingested with functional food can modify its activity (in

example, vitamin C increases iron absorption, radishes decrease carotene

bioavailability) (Olivares 1997);

- processing effect (common treatments such as fermentation, concentration,

pasteurization, canning, bottling, and others can influence nutrients concentration

and bioavailability) (Clydesdale 2004);

- environment (soil fertilization, temperature, rain, pesticides use; Clydesdale 2004)

(Clydesdale 2004).

Compliance is achieved when the intended amount of food is consumed.

Chapter 4

Research Methods

4.1 Specific Aim 1 Objective

To develop a rapid and sensitive LC-MS method for separation, identification and quantification of soy and chickpea saponins in plant materials, bread and excretions.

Figure 7 – Experimental design for aim 1

4.1.1 Saponins Extraction

Saponin containing ingredients were screened with the following HPLC-PDA method. Samples were dissolved in 80% aqueous methanol and injected onto a Symmetry

C18 column, 4.6mm x 75mm ID, 3.5µm particle size (Waters Corp., Milford, MA).

Composition of mobile phase solvents was as follows: A= 0.05% Trifluoroaetic acid

(TFA) in water; B= 0.05% TFA in acetonitrile (ACN). A gradient of A and B was used to separate the synthesized compounds at a flow rate of 1.5 ml/min and temperature of

30°C. An in-line photodiode array detector was used to identify and quantify saponins absorbing at the following wavelengths: 205 and 292 nm.

Saponins standards were isolated from soy and chickpea ingredients (Figure 7) in

Dr. Schwartz‟s lab using a modified version of Dr. Hu‟s method (2002). Ingredients used for this study were the following: soy flour (Baker‟s Soy Flour, ADM Protein Specialties

Division, Decatur, IL USA); soy milk powder (Devansoy Farms, Carrol, IA, USA); soy proteins (931, ADM Protein Specialties Division, Decatur, IL USA); chickpea flour

(Bob‟s Red Mill, Milwaukie, Oregon, USA); chickpea proteins isolate (kindly donated by

Dr. Kerem, The Hebrew University of Jerusalem, Israel). Caution was used to assure preservation of the DDMP-type saponins from converting to type B as occurs in common extraction conditions (Heng et al. 2006) by working in a dark environment and keeping temperature <30°C. Raw material in the amount of 15 g (Hu used 1 g) was stirred in 100 ml of 70% aqueous ethanol for 2.5 hours at room temperature. The solution was filtered and centrifuged; the obtained supernatant was allowed to dry by rotoevaporator (to remove ethanol) and freeze-dryer (to remove water). The dry powder was re-dissolved in

80% aqueous methanol as described previously (Berhow et al. 2002; Hu et al. 2002) but

33 at a higher concentration (100 mg/ml vs. 1 mg/ml). Modifications were applied to literature methods in order to achieve the highest yield of saponins. Dissolved raw extract in the amount of 1 ml were filtered through nylon filters (0.2 µm) and centrifuged for 2 minutes before injection into the semi-preparatory HPLC. Two solvents were used: solvent A (0.05% trifluoroacetic acid in water); solvent B (0.05% trifluoroacetic acid in acetonitrile). Isocratic elution was performed for chickpea products (50% A, 50% B) while a gradient was used for soy products (from 40% B to 60% B).

Each run obtained from the semi-preparative HPLC was immediately allowed to dry by applying vacuum in rotoevaporator with temperature <30°C and subsequently freeze-dried. Once a dry powder was obtained it was collected in 4 ml vials, sealed with parafilm to avoid moisture absorption and stored in a -80°C freezer. All experiments were performed in the dark in order to prevent reaction of the antioxidant compounds (DDMP type saponins). Saponins fractions were run in HPLC-PDA to determine their purity by specific absorbance and elution time. Standards of soyasaponin Ab and Bb, kindly donated by Dr. Berhow, were used for identification of type A and B saponins. Type

DDMP saponins were identified by specific absorbance at both 205 and 292 nm

(absorbance at 292 nm is due to the DDMP moiety) and retention time and their fraction collected. A standard of the DDMP type saponin βg was extracted from chickpea flour and used for quantification of DDMP saponins.

4.1.2 LC-MS Analysis

Saponin profile and content of several ingredients was determined by LC-MS

(Figure 7) and such method was also used for analysis of samples from in vitro digestion

(Aim 2). Chromatographic separation of soy saponins was achieved by reversed phase on a Waters Acquity UPLC, equipped with a binary pump, autosampler, column oven and degasser (Acquity, Waters Corp., MA, USA). An Acquity BEH C18 column

(2.1mm×50mm, 1.7µm, Waters Corp., Milford, MA, USA) was held at 40°C and a solvent system of 0.1% (v/v) formic acid in water (A) and 0.1% (v/v) formic acid in acetonitrile (B) delivered at a flow rate of 0.8 mL/min. The gradient elution was applied as follows: 0–2 min, 30.0–43.8% B; 2–2.8 min, 43.8–60.0% B; 2.8–3.8 min, 60.0-95.0%

B; 3.8–5 min, re-equilibration to initial conditions. The injection volume was 1µL for ingredients and breads, 5µL for digesta and aqueous fractions, 15µL for cell uptake samples and the autosampler was kept at 25°C.

HPLC eluate was split approximately 1:10 and interfaced with a Quattro Ultima triple quadrupole mass spectrometer (Micromass, Manchester, UK) operated in negative ion electrospray mode. Acquisitions were performed by selected ion recording (SIR). 17

Channels were set (one for each soy saponin analyzed) and source conditions were as follows: capillary voltage, 2.8 kV; source temperature, 110°C; desolvation temperature,

480°C; cone gas flow rate, 102 L/h; desolvation gas flow rate 801 L/h and cone voltage,

50 V. High purity nitrogen was used as desolvation and nebulizing gas and high purity argon was used as collision gas (1.87×10−4 mbar). All data were acquired and peak areas integrated using Masslynx 4.1 software (Waters Corp., Beverly, MA).

Standards of saponin Ab and Bb were kindly donated by Dr. Berhow (USDA,

Peoria, IL) while standard of saponin βg was isolated in Dr. Schwartz‟s laboratory

(4.1.1).

4.2 Specific Aim 2 Objective

Influence of saponins on micellarization of 14C-cholesterol upon in vitro digestion and saponins bioaccessibility using the coupled model in vitro digestion/Caco-2 human intestinal cells.

Figure 8 - Experimental design for aim 2

4.2.1 Saponins effect on cholesterol micellarization studied by in vitro digestion

Details of the in vitro digestion proposed are shown in Figure 8. Soy saponins isolated in Dr. Schwartz‟s laboratory, as well as commercial mixture of phytosterols and an extract of tomatine (Chromadex, Irvine, CA) were analyzed. Saponins effect on

14 cholesterol micellarization was determined by others (Carlson 2009). Briefly, 3,4 C2- cholesterol (2 mg) in 0.5 ml corn oil with 5% either none or relevant quantities for soy and chickpea-saponins and soy-chickpea breads (2g) were added to a 50 ml polypropylene tube for in vitro digestion containing yogurt as test meal. Saponin isolates and cholesterol were combined at different molar ratio, from 1:5 to 1:20 cholesterol:saponin isolate. Simulated gastric and small intestinal digestions has been described previously (Garrett et al. 1999). At the completion of digestion, samples were centrifuged (167,000 g) for 30 min to separate residual solids. Aqueous fractions were filtered (0.22 µm) in order to collect the micellar fraction to assess the effect of saponins on cholesterol partitioning into micelles during digestion (Figure 8). Samples with each test concentration of saponins were subjected to eight independent digestions.

Scintillation counting was used to determine cholesterol content in digesta and aqueous fractions. The ratio (%) of cholesterol concentration in the aqueous fraction over the digesta was used as an indicator of cholesterol micellarization.

4.2.2. Saponins bioaccessibility upon in vitro digestion Reagents porcine pepsin, porcine lipase, porcine pancreatin, porcine bile extract were purchased from Sigma Chemical Co. (St. Louis, MO). Bread (1.0 g) containing either soy alone, 1/3 chickpea and 2/3 soy, 2/3 chickpea and 1/3 soy was homogenized

(Ultra Turrax 339619, Tekmar Company) in 7 ml phosphate buffered saline and

37 transferred to 50 ml polypropylene tube. The sample was subjected to simulated oral, gastric and small intestinal digestion using procedures previously described (Garrett et al.

1999; Chitchumroonchokchai et al. 2004). Briefly, simulated saliva (7 ml) containing

3000 unit α-amylase was added to the homogenized bread mixture and incubated in shaking water bath (85 rpm) at 37 C for 10 min. For gastric digestion, the pH of the mixture was decreased to pH 3.0 with 1M HCl, porcine pepsin added, and the volume increased to 40 mL with salt solution (120 mmol/l NaCl, 5 mmol/l KCl, and 6 mmol/l

CaCl2). The bolus was incubated at 37˚C in a shaking water bath. After 60 min, pH of the chyme was increased to pH 6.0 with 1 mol/l NaHCO3 before adding pancreatin, pancreatic lipase and bile extract (final concentrations 0.4, 0.2 and 2.4 mg/ml, respectively). The pH was adjusted to 6.5 with 1 mol/L NaOH, the final volume increased to 50 ml with saline, and the reaction mixture incubated at 37˚C in a shaking water bath for 120 min. An aliquot of chyme was centrifuged at 12,000 x g, 4 C, for 45 min (Avanti J-E, Beckman, Palo Alto, CA) to separate the aqueous fraction from undigested materials. Supernatant was passed through a syringe filter (0.2 µmeter pores) to collect aqueous or bioaccessible fraction. In a separate experiment, bile extract was deleted during small intestinal phase of digestion to assess the role of mixed micelle formation in partitioning of saponins in the aqueous or bioaccessible fraction. Aliquots of chyme and aqueous fractions were stored under nitrogen gas at -80˚C until analysis.

Caco-2 cells (HTB37, ATCC) were maintained in T75cm3 flasks as previously described (Chitchumroonchokchai et al. 2004). Experiments used highly differentiated cultures of Caco-2 cells (11-14 days post-confluency) at passages 26-28. Aliquots of aqueous fractions from simulated digestion were diluted 1:4 with basal Dulbecco„s

38 minimum essential medium (DMEM) and added to washed monolayers of Caco-2 cells.

Cells were incubated in a humidified atmosphere of 95% air, 5% CO2 at 37˚C for 4h to examine cellular uptake of saponins. Media were aspirated and monolayers washed with ice cold phosphate buffer saline (PBS) containing 2 g albumin/l, followed by two washes with albumin-free cold PBS. Monolayers were scraped from the surface of the culture dish, collected in cold PBS, and centrifuged for 5 min at 80 g, 4˚C. The supernatant was discarded, and cell pellets were stored under nitrogen gas at -80˚C until analysis.

Breads were extracted with a modified version of Hu et al. (2002). Breads were ground to a fine paste with a grinder (Oskar Jr. chopper Plus, Sunbeam, Boca Raton, FL) and aliquots of 150 mg were mixed with 3 mL of 70% aqueous ethanol in 4 ml vials. The mixture was sonicated (FS30H Fisher Scientific, Fair Lawn, NY) for 2 min, filtered through nylon filters (0.2 µm, 13 mm, Alltech Associates Inc, Deerfield, IL) and centrifuged (micro centrifuge 5424 Eppendorf, Hauppauge, NY) for 2 min prior to LC-

MS injection.

Digested and aqueous fractions of soy-chickpea breads were thawed at room temperature and vortexed (model no. 231, Fisher Scientific, Fair Lawn, NY) for 1 min.

Aliquotes of 300 ml were mixed with 700 mL of ethanol, sonicated for 2 min, then filtered and centrifuged prior to LC-MS injection.

Test medium (0h incubation) and spent medium (4h incubation) were extracted with the procedure described for digested bread samples. Cell uptake pellets were resolubilized in 500 ml 70% aqueous ethanol, sonicated for 2 min, then filtered and centrifuged prior to LC-MS injection.

4.3 Specific Aim 3 Objective

To develop functional baked goods meant to lower cholesterol absorption by incorporating saponins and saponin-rich ingredients into wheat and soy bread.

Pocket-type doughs + Soy bread + soy Wheat bread + chickpea Soy bread + soy blend saponins saponins chickpea proteins

0% 10% 20% 26% 0% 1% 0% 1% 3% no 1/3 2/3

Macroscopic state of water Textural characterization (TGA, DSC) (TPA)

Molecular state of water 1 ( H NMR)

Figure 9 - Experimental design for aim 3

4.3.1 Development of pocket-type soy doughs

The effect of soy on pocket-type doughs was studied. Four pocket-type dough formulations were developed with 0, 10, 20 and 26% soy blend (HealthyHearth™ Baking

Blend, Columbus, OH, consisting of 2/3 soy flour, 1/3 soy milk powder) as shown in

Table 2. The soy blend contained 50% proteins, 28% carbohydrates, 9% fat, 5.5% moisture, 5% fiber, 2.5% ash. Wheat flour was purchased from General Mills (Grand

Rapids, MI). The following process was used to make the doughs: ingredients were

40 scaled, and then mixed in a high speed mixer until combined for about 1 minute (cuisine mixer, Kitchen aid, K5SS, St. Joseph, MI). Dough was then be allowed to rest for 10 minutes (hydration) and rolled in a sheeter (Atlas model 150mm deluxe, Italy).

All doughs (130x45x3mm, weight ~5 g) were analyzed before and after baking:

120 s at 182°C in a toaster oven (Hamilton Beach Toastation, Washington, NC). Each analysis was performed 3 minutes after baking.

Ingredient (%) Control 10% 20% 26% Wheat Flour 61 50 39 30 HealthyHearth™ Baking Blend 0 10 20 26 Water 36 37 38 41 Yeast 1 1 1 1 Salt 2 2 2 2

Table 2 - Pocket-type dough formulations developed in this study

4.3.2 Development of saponin-rich breads

Soy saponin-soy bread was prepared using a patented process (Vodovotz and

Ballard 2009). An isolate of soy saponins, kindly donated by Dr. Kerem, was added to the formulation of soy bread in the amount of 1% w/w (Table 3; Figure 9). Ingredients were combined and mixed in a 5-quart Kitchen Aid Mixer until uniform dough is obtained. The dough was then hand-kneaded into 100 g pup loaves, proofed at 48°C and humidity >90% (CM2000 combination module, InterMetro Industries Corp, Wilkes-

Barre, PA) for 1 hour and baked at 163°C inject air oven (Model: JA14, Doyon, Liniere,

Quebec, Canada) for 20 minutes. Bread was allowed to cool at room temperature on a cooking rank for 3 hours prior to the analysis.

Ingredient (%) STD SOY Water 45.3 44.8 Soy Flour 19.9 19.7 Wheat Flour 17.5 17.3 Soy Milk Powder 6.6 6.5 Sugar 4.5 4.4 Gluten 2.2 2.2 Shortening 1.7 1.7 Yeast 0.9 0.9 Salt 0.9 0.9 Dough Conditioner 0.2 0.2 Soy Saponin Isolate - 1

Table 3 - Control bread (STD) and soy saponin-soy bread (SAP) formulations

Two chickpea saponin-wheat breads were prepared by adding a food grade isolate of chickpea saponins (kindly donated by Dr. Kerem) to the formulation of wheat bread (AACC method 10-10B modified in order to reduce yeast and sugar; Table 4) in the amount of 1 and 3% w/w (Figure 9). Bread preparation was similar to the one used for the soy saponin-soy bread with the only exception that the loaf size was 300 g instead of 100 g in order to study saponin effect on loaf volume by rapeseed displacement.

Breads were analyzed only after cooling at room temperature as noted previously.

Ingredient (%) Control 1% CP 3% CP Wheat Flour 56.6 55.8 54.2 Water 36.0 36.1 36.0 Sugar 2.3 2.1 1.6 Shortening 1.7 1.7 1.7 Salt 1.7 1.7 1.7 Yeast 1.2 1.2 1.2 Dough Conditioner 0.5 0.5 0.5 Chickpea Saponin Isolate - 1 3

Table 4 - Control bread (0%) and chickpea saponin wheat breads (1 and 3%) formulations

4.3.3 Development of breads containing saponin-rich ingredients

Chickpea-soy breads were prepared using a patented process (Vodovotz and

Ballard 2009) as described in point 4.3.2. Saponin-rich ingredients such as chickpea proteins isolate (kindly donated by Dr. Kerem) were incorporated into the soy bread formulation at different proportions for the development of texturally acceptable bread

(Table 5; Figure 9). The chickpea proteins isolate substituted part of the soy blend in the attempt to increase the amount of saponins, DDMP type especially (βg). Chickpea proteins isolate (~50% proteins) substituted for soy blend at different proportions (1/3 and 2/3) in order to determine what rate of substitution will have a better impact on bread texture.

Ingredient (%) SOY 1/3 CPP 2/3 CPP Water 45.3 45.3 45.3 Soy Flour 19.9 13.3 6.6 Wheat Flour 17.5 17.5 17.5 Soy Milk Powder 6.6 4.4 2.2 Chickpea Proteins - 8.8 17.7 Sugar 4.5 4.5 4.5 Gluten 2.2 2.2 2.2 Shortening 1.7 1.7 1.7 Yeast 0.9 0.9 0.9 Salt 0.9 0.9 0.9 Dough Conditioner 0.2 0.2 0.2

Table 5 - Bread formulations containing no (SOY), 1/3 (1/3) and 2/3 (2/3) chickpea proteins

4.3.4 Texture Profile Analysis (TPA)

Texture Profile Analysis test (74-09 AACC, 2008) was performed using an

Instron Micro System 5542 (Instron, Canton, MA) and data was analyzed using the software Bluehill 2 (Instron, Norwood, MA). TPA was used to simulate mastication by a double compression (40%) at a speed of 100 mm/min to 25x25x15mm dough samples.

Doughs (130x45x3 mm each) were shaped with an electric carving knife (Toastmaster,

Boonville, MO) in order to obtain the required length and width. Then, 5 doughs were stacked in order to obtain the required thickness. Parameters determined included: hardness, springiness, cohesiveness, gumminess and chewiness. Definitions of these parameters can be found elsewhere (Bourne 2002).

4.3.5 Thermal Gravimetric Analysis (TGA)

Moisture content and water distribution were determined by a thermogravimetric analyzer (Q 5000 TA, New Castle, DE). About 20 mg of each sample were placed in a previously tared stainless-steel pan (PerkinElmer, Boston, MA) and heated from 25°C to

150°C at the rate of 5°C/min. Total moisture loss was obtained by subtracting the final weight from the initial weight (Universal Analysis Software, Version 4.2, TA

Instruments, New Castle, DE). The first derivative curve of weight loss was studied in order to determine water distribution among the products.

4.3.6 Differential Scanning Calorimetry (DSC)

Thermal transitions indicative of the state of components in a system were determined by Differential Scanning Calorimeter (DSC). About 10 mg of dough were

45 placed in large volume stainless steel sample pans and lids fitted with o-rings

(PerkinElmer, Boston, MA). Sample and reference pans were placed inside the DSC equipped with a refrigerated cooling system (Q 100, TA Instruments, New Castle, DE).

Samples were heated from -50°C to 100°C at the rate of 5°C/min. Transitions observed in the thermograms were analyzed using the Universal AnalysisTM software (TA

Instruments, New Castle, DE). Enthalpies of transitions were estimated by integration of peaks and were expressed in J/g. Percent “Freezable” Water (%FW) were calculated by using equation 1 where λ is the specific latent heat of fusion of water (334 J/g).

(1)

4.3.7 Nuclear Magnetic Resonance (NMR)

1 Proton Nuclear Magnetic Resonance ( H NMR) relaxation experiments T1 and T2 were performed to study proton mobility, which is correlated to water mobility, of the pocket-type doughs using a Bruker NMR DMX 300 MHz: Saturation Recovery (Derome and others 1987) and the CPMG sequences (Carr and Purcell 1954), respectively. Both experiments were characterized by a spectral width of 18 MHz, acquisition time of 456 ms and pulse length of 12.50 µs. 4 scans were performed during each T1 experiment while 8 scans were performed during T2.

In the T1 experiments, the intensity of the relaxation peak as a function of recovery time were fit to equation 2 to determine the value of T1 of water. Relaxation time T2 was measured using a Car-Purcell pulse train with a spacing of 1 ms; the intensity

46 of the relaxation peak with pulse train length was measured to determine the T2 of water by fitting to equation 3.

t (2) T1 INTENSITY fit: I t I 0 P exp T1

t (3) T2 INTENSITY fit: I t P exp T2

1 1 T measured T dd R (4) 2 2 ex

The correlation time (τc) and water exchange rate were estimated from the measured relaxation rates using the standard equations for T1 and T2 and assuming dipolar relaxation and isotropic tumbling (Gordon et al. 2006). In addition, an exchange

dd rate, Rex, was added to the dipolar transverse relaxation rate, T2 , by equation 4 in order to fit the measured T1 and T2 data. All calculations were performed with Mathematica7

(Wolfram Research, Inc., Champaign, IL).

4.3.8 Color Analysis

Changes in color were monitored with a colorimeter (CR-300, Minolta, Osaka,

Japan). The CIE parameters L*, a* and b* were measured with an illuminant of D65 and a standard observer of 2°. L* represents the lightness with values from 0 (black) to 100

(white), which indicates a perfect reflecting diffuser. Chromatic components are represented by a* and b* axes. Positive values of a* are red and negative values are

47 green, whereas positive values of b* are yellow and negative values are blue. ΔE* (total color difference) was calculated using equation 5 and using formulation called control as the reference point. Crust color was measured on four different zones of the top of the whole bread.

ΔE = √(ΔL*)2 + (Δa*)2 + (Δb*)2 (5)

4.3.9 Loaf Volume Measurement

Specific loaf volume was measured using the standard rapeseed displacement method (10-05 AACC, 2000) on intact loaves. Each loaf was weighed and the specific loaf volume was obtained from the ratio of volume and weight. The reported data are the result of at least three loaves per type of bread.

4.3.10 LC-MS

Saponin content and composition of the chickpea saponin-rich wheat breads were determined by LC-MS as described in section 4.1.2.

Chapter 5

Results and Discussion

5.1 Specific Aim 1

5.1.1 Saponins extraction

Isolation of saponins from food sources was achieved by optimizing a method described previously (Hu et al. 2002). Hu‟s method was developed to allow isolation on small quantities of saponins from food matrices in order to obtain a profile of their saponin content. The current modifications allowed a higher yield (15 times higher) due to a higher concentration of the raw material in the extraction solvent. Furthermore, based on Heng‟s findings (2006) temperature, pH and extraction time were controlled in order to minimize the conversion of DDMP saponins to type B saponins resulting in a significant quantity of the DDMP type. Several ingredients were analyzed by HPLC-PDA in order to characterize their saponins profile and to choose the raw material for the isolation of these phytochemicals. Chromatograms of soy flour and chickpea flour are shown in Figure 10 as compared to standards of saponins Ab and Bb. Protein isolates from soy and chickpea (Figure 11 A, B) and soy milk powder (Figure 12) were screened as well. Flours (Figure 10) were characterized by a different saponin profile compared to the proteins (Figure 11) consisting in predominantly DDMP saponins compared to type B

49 saponins. The processing conditions for the isolation of proteins may have induced the conversion of DDMP saponins to type B (Heng et al. 2006). Chromatogram of soy proteins did not show peaks attributable to type A saponins (Figure 11). Similarly, chromatogram of soy milk powder (Figure 12) did not show the presence of Type A saponins; soy milk powder saponin profile was characterized by higher presence of type

B suggesting a conversion from DDMP type to type B (Figure 12).

2 296.5 202.1 A

Ab βg Bb 0

2 295.3 202.1 B βg

Absorbance(AU) 0 Bb C 0.2 Ab

0.0 5.00 10.00 15.00 20.00 25.00 30.00 Time (min)

Figure 10 - HPLC-PDA chromatogram at 205 nm of soy flour (A); chickpea flour (B) at 50 mg/ml, 50 µl injections; Ab and Bb standards (C), 1 mg/ml, 50 µl injections

2.00 βg 1.50 Bb A

1.00 AU

0.50

0.00 5.00 10.00 15.00 20.00 25.00 30.00 Minutes Bb

2.00 βg B Absorbance (AU) Absorbance

AU 1.00

0.00 5.00 10.00 15.00 20.00 25.00 30.00 TimeMinutes (min)

Figure 11 - HPLC-PDA chromatograms at 205 nm of protein extracts from soy (A) and chickpea (B) at 50 mg/ml. 50 µl injections

292 nm 2.00 βg Ab Bb βg

0.00 Absorbance (AU) Absorbance 0 5.00 10.00 15.00 20.00 25.00 30.00 Time (min)

Figure 12 - HPLC-PDA chromatograms at 205 nm of soy milk powder at 50 mg/ml, 50 µl injection

Based on these results, soy and chickpea flour were selected to isolate a mixture of saponins representative of the profile in that commodity. Isolates of saponin types A, B and DDMP were obtained from soy and chickpea flours (Figure 15 A, B, C), while a standard of the major soy and chickpea saponin, βg, was isolated from chickpea flour

(Figure 15 D).

Preparative HPLC yielded chromatograms characterized by broader peaks compared to the analytical HPLC (Figures 13 and 14) due to the high concentration (100 mg/ml) and volume (1000 µl) of each injection and flow rate (10 ml/min; Figures 13 and

14). Such conditions reduced peak resolution and thus the ability to identify saponins by chromatography. Since chromatography was not a reliable tool for saponin identification, the separation method was further optimized. Fractions of eluates from the semi- preparatory HPLC were collected at specific time ranges and analyzed by LC-MS for saponins and potential contaminant (i.e. isoflavones). This procedure allowed us to determine the precise time ranges at which to collect the desired saponins and thus obtain high purity isolates.

3.50 205 nm 260 nm 292 nm

βg

AU Absorbance (AU)Absorbance

0.00

0 7 TimeTime (min) (min)

Figure 13 - PDA chromatogram of chickpea flour in semi-preparatory HPLC at 205, 260, 292 nm at 100 mg/ml, 1 ml injection

2.40 205 nm DDMPs 260 nm 292 nm

AU Bs

Absorbance (AU) Absorbance As

0.00 5 16 TimeTime (min) (min)

Figure 14 - PDA chromatogram of soy flour in semi-preparatory HPLC at 205, 260, and 292 nm at 100 mg/ml, 1 ml injection

Bb Ab Ab

0.20 A B

-0.25 AU

0.10 AU

AU AU

-0.30 0.00 5.00 10.00 15.00 20.00 25.00 30.00 5.00 10.00 15.00 20.00 25.00 Minutes Time (min) TimeMinutes (min)

βg βg Bb Bb 0.12 Ab C 0.30 0.10 292 nm D

0.20

AU AU AU 0.08 0.10 0.06

5.00 10.00 15.00 20.00 25.00 5.00 10.00 15.00 20.00 TimeMinutes (min) TimeMinutes (min)

Figure 15 - HPLC-PDA chromatograms at 205 nm of saponin extracts: soy type A (A), soy type B (B), soy DDMP (C) and chickpea βg (D) at 1 mg/ml, 10 µl injections

5.1.2 LC-MS Analysis

Liquid chromatography was an excellent tool for identification and quantification of major soy saponins, but it lacked sensitivity for the analysis of minor soy saponins.

Furthermore, run time was long (30 minutes; Figures 10, 11, 12). In order to obtain a full profile of soy saponins in the ingredients analyzed, HPLC was coupled to a mass spectrometer as described in section 4.1.2. The superior sensitivity of this analytical technique allowed complete saponins profiling and reduced exposure of the analytes to solvents (run time 7 minutes; Figures 16, 17, 18, 19, 20). Peaks were identified by molecular weight of saponins under the specified analytical conditions (negative mode).

Soy flour contained all soy saponins known (17 compounds) with a predominance of the DDMP type (Figure 16). The Type B saponins were ~50% of the DDMP while

Type A saponins were ~40% of the DDMP type (Figure 16). Type A saponins in soy flour were represented mainly by Ab and Af (Figure 13; Table 6). Saponin Ab (molecular weight 1437 g/mol) is a bidesmosidic glycoside whose sapogenol is attached to two sugar chains (one consisting of arabinose and triacetylated xylose, and the other consisting of glucuronic acid, galactose and glucose; Güçlü-Üstündağ and Mazza 2007). Retention time for Ab using this analytical method was 1.10 minutes (Figure 16). Saponin Af is the product of glucose hydrolysis from Ab and its retention time was 1.33 minutes. Minor amounts of Ac, Ad and Ah were also detected (Figure 16; Table 6). The Type B in soy flour mainly consisted of Bb and Bc (Figure 16; Table 6). Saponin Bb (molecular weight

943 g/mol) is a monodesmosidic glycoside whose sapogenol is attached to one sugar chain, consisting of glucuronic acid, galactose and rhamnose (Güçlü-Üstündağ and

Mazza 2007). Retention time of Bb using this analytical method was 1.80 minutes

(Figure 16). Saponin Bb‟ is the product of rhamnose hydrolysis from Bb and its retention time was 1.92 minutes (Figure 16). Minor amounts of Ba, Bc‟, Bd E, and Be E were detected (Figure 16; Table 6). The DDMP type saponins in soy flour mainly consisted of

βg and βa (Figure 16; Table 6). Saponin βg (molecular weight 1069 g/mol) is the product of the bond of Bb saponin to a DDMP moiety (Güçlü-Üstündağ and Mazza 2007).

Retention time of βg in this analytical method was 2.37 minutes (Figure 16). Saponin βa is the product of the bond of Bc saponin to a DDMP moiety and its retention time was

2.47 minutes (Figure 16) (Güçlü-Üstündağ and Mazza 2007). Minor amounts of αg, γa, and γg were detected (Figure 16; Table 6).

Unlike soy, chickpea flour contained only three saponins: Bb, Be E and βg

(Figure 17; Table 6) as found previously (Kerem et al. 2005). Traces of Bb‟, Bc, βa, and

γa were detected (Table 6). The predominant saponin in chickpea flour was βg, followed by Bb (βg minus DDMP moiety) and Be E (oxidation product of Bb) (Figure 17; Table

6). These findings suggested that chickpea flour contained only βg, which partially converted into other forms upon hydrolysis and oxidation caused by either processing or storage of the ingredient.

Previous studies have shown that saponins are mostly bound to proteins (Francis et al. 2002; Güçlü-Üstündağ and Mazza 2007). Therefore, we analyzed soy and chickpea protein isolates and characterized their saponin profile. Both soy and chickpea protein isolates contained three times more saponins than the flours: 1440 vs. 464 μg/g (soy) and

1058 vs. 299 μg/g (chickpea, Table 6). Similarly, soy milk powder contained almost three times more saponins than soy flour (1025 vs. 464 μg/g; Table 6). Mass spectra of these ingredients revealed a higher amount of Type B saponins and a lower amount of DDMP

Type saponins compared to the flours (Figures 18, 19, 20). Moreover, the sensitivity of the LC-MS method developed enabled the complete characterization of the saponin profile of the different ingredients. The LC-MS method previously described was also used to analyze the standards that we isolated for saponins quantification of food and biological samples. The chromatogram at 205 nm (Figure 21 A) and the spectrum (Figure

21 B) show the high purity of the equimolar solution of Ab, Bb and βg saponins.

Standards of Ab and Bb were kindly donated by Dr. Berhow, while βg was isolated in Dr.

Schwartz‟s lab.

Figure 16 – LC-MS spectrum of saponins in soy flour (50 mg/ml, 1 µl injection). Peak labels represent specific saponins

Figure 17 - LC-MS spectrum of saponins in chickpea flour (50 mg/ml, 1 µl injection). Peak labels represent specific saponins

Figure 18 - LC-MS spectrum of saponins in soy protein isolate (50 mg/ml, 1 µl injection). Peak labels represent specific saponins

Figure 19 - LC-MS spectrum of saponins in chickpea protein isolate (50 mg/ml, 1 µl injection). Peak labels represent specific saponins

Figure 20 - LC-MS spectrum of saponins in soy milk powder (50 mg/ml, 1 µl injection). Peak labels represent specific saponins

Figure 21 – Chromatogram at 205 nm (A) and mass spectrum (B) of equimolar solution of saponins Ab, Bb and βg, injection volume 5 μl. Peak labels represent specific saponins

saponin MW soy chickpea soy chickpea soy milk (g/mol) flour flour proteins proteins powder (µg/g) (µg/g) (µg/g) (µg/g) (µg/g) A type Ab 1437 61 ± 9 nd 147 ± 3 nd 136 ± 4 Ac 1421 14 ± 2 nd 39 ± 1 nd 26 ± 1 Ad 1407 3.3 ± 0.5 nd 9.9 ± 0.2 nd 6.7 ± 0.3 Af 1275 33 ± 5 nd 91 ± 2 nd 61 ± 2 Ah 1245 9.8 ± 1.5 nd 29 ± 1 nd 18 ± 0 B type Ba 959 1.2 ± 0.2 nd 20 ± 1 nd 17 ± 0 Bb 943 21 ± 3 12 ± 5 342 ± 14 310 ± 1 215 ± 4 Bb‟ 797 1.3 ± 0.2 0.12 ± 0.07 32 ± 4 3.6 ± 0.0 15 ± 0 Bc 913 7.2 ± 1.2 0.11 ± 0.03 153 ± 6 3.2 ± 0.2 85 ± 2 Bc‟ 767 0.62 ± 0.09 nd 15 ± 1 nd 6.2 ± 0.1 E type Bd E 957 0.04 ± 0.03 nd 1.0 ± 0.7 nd 1.8 ± 0.0 Be E 941 1.4 ± 0.2 1.7 ± 0.7 31 ± 1 38 ± 0 23 ± 0 DDMP type αg 1085 9.4 ± 1.9 nd 0.23 ± 0.16 nd 0.22 ± 0.03 βg 1069 193 ± 31 278 ± 113 320 ± 15 686 ± 13 263 ± 5 βa 1039 79 ± 14 2.7 ± 1.1 156 ± 7 9.9 ± 0.7 116 ± 3 γa 893 16 ± 7 4.4 ± 1.9 41 ± 1 10 ± 0 26 ± 1 γg 923 13 ± 9 nd 15 ± 1 nd 8.9 ± 0.2

Total 464 ± 77 299 ± 121 1440 ± 55 1058 ± 16 1025 ± 24

Table 6 - Saponin content of soy and chickpea ingredients indicated as μg saponins/g ingredient

5.2 Specific Aim 2

5.2.1 Saponins effect on cholesterol micellarization studied by in vitro digestion

Soy and chickpea saponin effect on cholesterol micellarization was studied by others (Carlson 2009) in order to determine whether the hypocholesterolemic activity of saponins is related to the inhibition of cholesterol micellarization. Soy saponins, as well as soy saponins type A and DDMP were isolated from soy flour. Chickpea saponins

(mostly βg) were isolated from chickpea flour.

Soy and chickpea saponins, as a mixture and as separate types, were compared to a control (cholesterol alone) and to other compounds (phytosterols, tomatine and tomatidine) of known hypocholesterolemic activity (Carlson 2009). A solution containing

14 3,4 C2-cholesterol, 0.5 ml corn oil and 5% samples was added to the test meal (yogurt) and subjected to simulated digestion, centrifugation and filtration. Saponins were examined at different concentrations, measured as cholesterol:saponin molar ratio (1:5,

1:10, 1:12.5, 1:20). Micellarization of cholesterol was measured as the percentage of cholesterol in the starting material that partitioned in the filtered aqueous fraction after simulated digestion.

Results indicated a micellarization of ~90% cholesterol in soy and chickpea saponins, at all concentrations tested (cholesterol:saponin 1:10 and 1:20), thus not significantly different from the control (~90%; Figure 22). Phytosterols, tomatine and tomatidine significantly reduced cholesterol micellarization to as low as 30%

(cholesterol:tomatidine 1:10; Figure 22). These findings suggested that the hypocholesterolemic activity of soy and chickpea saponins is not related to inhibition of cholesterol micellarization, in contrast with previous preliminary findings (Kerem et al.

2006) that reported a micellarization of 80% cholesterol in the presence of soy saponins.

Absence of significant effect on cholesterol micellarization could be attributed to the low power of inhibition of soy saponins or to a different preparation of the extracts.

Figure 22 - Micellarization of cholesterol during simulated digestion of test meal (yogurt) with and without various saponins (Carlson 2009)

5.2.2 Saponins bioaccessibility upon in vitro digestion

Stability of Saponins from Soy and Chickpea Ingredients during Bread Making

Saponin sources in the soy-chickpea breads were soy flour, soy milk powder and chickpea protein isolate. Seventeen saponins (types A, B, E, DDMP) were identified and quantified in soy flour and soy milk powder by LC-MS analysis (Table 7). Total saponin content of soy flour was 464 ± 77 µg/g with DDMP type most abundant, followed by saponin types A, B and a trace of E. βg was the predominant saponin in the soy flour

(193 ± 31 µg/g). Total saponin content of soy milk powder was twice that of soy flour

(1025 ± 24 µg/g, Table 7). βg saponin was the most abundant saponin in soy milk powder

(25% of total) and the amounts of types B and DDMP saponins were similar to that in soy flour. Only seven saponins were identified and quantified in the chickpea protein isolate. Total saponin content was comparable to that of soy flour. βg and Bb saponins were most abundant in the chickpea protein isolate and A type saponins were not detected.

Saponin content for the three breads was calculated by considering the percentage of the specific ingredients in the final formulation. The three breads had similar total saponin content but different saponin profiles (Table 7). Partial substitution of the soy ingredients with chickpea protein decreased amount of type A saponins and increased the content of type B, E and DDMP saponins in the bread (Table 7).

ingredients calculated saponin profile in breads soy soy milk chickpea SOY 1/3 CPP & 2/3 CPP & saponin flour powder protein 2/3 SOY 1/3 SOY type (µg/g) (µg/g) (µg/g) (µg/g) (µg/g) (µg/g) A type Ab 61 ± 9 136 ± 4 nd 80 54 26 Ac 14 ± 2 26 ± 1 nd 17 11 5.6 Ad 3.3 ± 0.5 6.7 ± 0.3 nd 4.1 2.8 1.4 Af 33 ± 5 61 ± 2 nd 40 27 13 Ah 9.8 ± 1.5 18 ± 0 nd 12 8.0 3.9 B type Ba 1.2 ± 0.2 17 ± 0 nd 5.3 3.6 1.7 Bb 21 ± 3 215 ± 4 155 ± 1 70 98 126 Bb‟ 1.3 ± 0.2 15 ± 0 1.8 ± 0.0 4.7 3.8 2.7 Bc 7.2 ± 1.2 85 ± 2 1.6 ± 0.1 27 19 9.7 Bc‟ 0.62 6.2 ± 0.1 nd 2.0 1.4 0.65 ±0.09 E type Bd E 0.04 1.8 ± 0.0 nd 0.50 0.34 0.16 ±0.03 Be E 1.4 ± 0.2 23 ± 0 19 ± 0 6.9 11 15 DDMP type αg 9.4 ± 1.9 0.22 ± nd 7.1 4.8 2.4 0.03 βg 193 ± 31 263 ± 5 343 ± 8 210 254 299 βa 79 ± 14 116 ± 3 5.0 ± 0.4 88 61 32 γa 16 ± 7 26 ± 1 5.0 ± 0.2 19 14 9.5 γg 13 ± 9 8.9 ± 0.2 nd 12 8.0 3.9

Total 464 ± 77 1025 ± 24 529 ± 8 605 581 552

Table 7 - Saponin content of soy-chickpea ingredients and estimated saponin content of soy-chickpea breads.

Saponin stability during the preparation of bread was determined by analyzing the soy-chickpea breads. Recovery of total saponins in soy bread was approximately 75%. In general, types A, B and DDMP saponins were relatively stable under bread making conditions (recovery ~70-80%) with lower recovery of type E saponins (Figure 23).

Recovery of type A saponins was not markedly affected by substitution of a portion of the soy flour with chickpea protein isolate. Interestingly, recoveries of type E and DDMP saponins in soy-chickpea breads were significantly lower (p = 0.006 and 0.018, respectively) than their recoveries from soy bread (Figure 23).

The low recovery of type E saponins in all three breads was likely due to the reactive ketone group at position 22 of the triterpenoid aglycone (Güçlü-Üstündağ and

Mazza 2007). The loss of DDMP saponins reflects hydrolysis of the maltol moiety. The acetal linkage is susceptible to hydrolysis during typical bread making conditions that include high temperature (>30°C), slight acidic pH, and presence of water (Heng et al.

2006). The extent of the hydrolysis was greater in the breads with soy and chickpea than for the bread with soy alone. This suggests that the soy matrix stabilizes the acetal linkage during preparation of the bread. Partial substitution of the soy blend with chickpea protein also decreased the fat and fiber contents of the bread. Chickpea protein isolate contained 3% fat and 3% fiber, whereas soy blend contained 9% fat and 5% fiber.

Soy fat may inhibit degradation of E saponins and hydrolysis of DDMP saponins due to its lower dielectric constant (6.3 kV for soybean oil (Cannon et al. 1999) than that of other common food starches (e.g. ~50 kV for corn starch (Cannon et al. 1999). As there was no increase in type B saponins associated with loss of DDMP, degradation appeared to differ from that described by Heng et al. (2006). Degradation of DDMP saponins was

68 proposed to follow two possible paths based on the conditions of the matrix (acidic or alkaline pH). Hydrolysis in media with high dielectric constant such as water containing starch favored generation of an ionized intermediate. This intermediate was proposed to undergo molecular rearrangement with subsequent release of maltol and B saponin (Heng et al. 2006). However, we did not observe increased recovery of type B saponins in breads containing chickpea compared to bread with soy alone (Figure 23). Therefore, an alternate degradation process may have occurred, possibly by reduction of the ketone group characteristic of saponins E and DDMP.

100 a a SOY 1/3 CPP & 2/3 SOY a 2/3 CPP & 1/3 SOY 80 a a

a 60 a

a b 40 b b 20 b

Recovery after bread after making (%) Recovery

0 A B E DDMP Saponin Type

Figure 23 - Recovery (%) of saponin groups in breads from ingredients. Data are mean ± SD, n = 3. One-way ANOVA was performed to compare recovery of each saponin type across the three breads. Presence of a different letter above the error bars within the saponin type indicates that recovery differed significantly (α<0.05) in the different matrices.

Recovery and Digestive Stability of Saponins from Digesta

Type B saponins were highly stable during digestion of soy bread (~100% recovery) (Figure 24). In contrast, recovery of types A and DDMP saponins after simulated digestion was ~60%. Hydrolysis of glycosidic linkages and of the maltol moiety during in vitro digestion may explain the partial loss during digestion (Figure 24).

The high recovery of type B saponins may reflect hydrolysis of DDMP to generate type

B saponins (Figures 24 and 25). Matrix appears to affect apparent stability of type B saponins as their recovery from digested soy and 1/3 chickpea protein bread was similar to that from digested soy and significantly (p 0.001) greater than that from digested bread containing 2/3 chickpea protein (Figure 25). These results suggest that degradation of

DDMP saponins is increased in low fat systems. Within the type A saponin group, recoveries of saponin Ac and Ad were lower than that of other type A saponins (~30% for Ac from all breads, ~40% for Ad from soy bread while below quantification limit in chickpea-containing breads). It is also interesting that recovery of saponins Bd E and Be

E was greater in bread containing one third chickpea protein (187% and 137%, respectively). As saponin Bd E is the oxidized derivative of saponin Ba (Price et al.

1986), high recovery of saponin Bd E suggests that saponin Ba was oxidized during simulated digestion. Also, less than 20% saponin γa was recovered in digested breads, whereas recovery of saponin γg exceeded 120%. Hydrolysis of the glycosidic linkage of rhamnose to the βg nucleus, the most abundant soy saponin, generates γg (Gestetner et al.

1986).

Bb 100 Soy bread

Bc βg βa Ab Af Ac Ba Bc’ Be E γg 0 Ad Ah Bb’ Bd E αg γa

100 Chyme

100 Filtered

Aqueous Relative Relative Intensity (%)

100 Caco-2 cells

0 0 0.50 1.00 1.50 2.00 2.50 3.00 3.50 Time (min)

Figure 24. Representative LC-MS spectra in soy bread, chyme and filtered aqueous fraction generated during simulted digestion and Caco-2 cells incubated with diluted aqueous fraction for 4 h.

160 SOY a 1/3 CPP & 2/3 SOY 140 ab 2/3 CPP & 1/3 SOY a a 120 b a 100 c ab b 80 a a a 60

Post-digestion recovery of saponins (% bread) saponinsof (% Post-digestion recovery A B E DDMP Saponin Type

Figure 25. Recovery (%) of saponin groups in chyme after in vitro digestion of breads. Data are mean ± SD, n = 5. One-way ANOVA was performed within each saponin type. The presence of different letters above bars for saponin type indicates significantly different mean (α<0.05).

Bioaccessibility of Saponins

The efficiency of transfer of type DDMP saponins to the aqueous or bioaccessible fraction during the small intestinal phase of digestion was 50-60% and, similar to that we previously reported for cholesterol in orange juice (Bohn et al. 2007), with no effect of the bread matrix (Figure 26). Partitioning of type A and E saponins in the aqueous fraction of chyme was markedly lower (approximately 30%) and higher (>85%) than that of type DDMP saponins, and also independent of composition of the bread. In contrast, the efficiency of transfer of type B saponins to the aqueous fraction during digestion of

72 bread with 2/3 chickpea protein was significantly less than that of breads with 1/3 chickpea protein and soy alone (~45%, ~65% and 58% respectively). Among the numerous saponins measured, lower extent of partitioning of saponins Ac and Ad was observed for all breads (~20%) as compared to the other type A saponins (~30%).

Furthermore, Bd E partitioned in the aqueous fraction of chyme more efficiently than E saponins for all breads (~100% partitioning vs. ~140% post-digestion recovery, therefore

71% of the amount in chyme).

120 SOY 1/3 CPP & 2/3 SOY 2/3 CPP & 1/3 SOY a 100 a a

80 a ab a a 60 b a

40 a a a

Partitioning inbread) aqueous fraction (% 0 A B E DDMP Saponin Type

Figure 26. Partitioning (%) of saponins in soy-chickpea breads in aqueous fraction during small intestinal phase of digestion in presence of bile extract. Data are mean ± SD, n = 5. One-way ANOVA was performed within each saponin type; different letters refer to significant difference (α<0.05).

Bile extract was deleted during the small intestinal phase of digestion in a parallel experiment to estimate the extent to which saponins are incorporated into micelles.

Partitioning of type B (~25%) and E (~25%) saponins in the absence of bile during small intestinal phase of digestion was markedly decreased from that in chyme lacking bile extract (50-60% and 85% for types B and E, respectively). Furthermore, partitioning of type DDMP saponins in the absence of bile was almost negligible (~2%). Surprisingly, partitioning of type A saponins in chyme lacking bile extract was more efficient than chyme containing bile extract (50-60% vs. ~30%, respectively) for all breads.

Distribution of the various types of saponins within chyme is likely associated with their relative solubilities in water. Type A saponins contain two hydrophilic tails (sugar chains) and an additional hydroxyl group resulting in greater hydrophilicity than other saponin types (Berhow et al. 2006, Güçlü-Üstündağ and Mazza 2007). The bi-desmosidic structure of type A saponins (two sugar chains attached to the aglycone) may have caused their aggregation with bile acids followed by salt formation and precipitation (Hwang and

Damodaran 1994). DDMP saponins are more hydrophobic, necessitating incorporation into micelles during the small intestinal phase of digestion. Types B and E saponins are intermediate and appear to be partitioned into both the aqueous and micelle compartments.

Caco-2 cell uptake of saponins from digested soy-chickpea breads

Accumulated of saponins in the aqueous fraction of chyme by Caco-2 cells after 4 h incubation is presented in Figure 27. Relative uptake (%) of a specific saponin from media containing aqueous fraction generated during digestion of the three breads

74 was not significantly different (p>0.05). Therefore, these data were pooled to simplify presentation. Type B, E and DDMP saponins were detected in cells after 4h incubation with mean uptake of ~1% from medium. DDMP saponins were the most abundant type in

Caco-2 cells, representing ~60% of total content (11, 6 and 3 nmol/mg protein for cultures exposed to aqueous fraction generated during digestion of soy bread and breads with 1/3 and 2/3 chickpea protein isolate). B type saponins accounted for ~35% total cell content (~7 nmol/g protein for all breads), whereas E type saponins accounted for ~5% total cell content (<1.0 nmol/mg protein for all breads). Relative uptake of Bb‟, Bc‟ and

Be E saponins (2.8, 2.6 and 1.8%, respectively) was significantly greater than that of the other saponins detected in cells. Hu et al. (2004a) reported that Caco-2 cells accumulated

~1% soy saponin Bb from apical medium (Hu et al. 2004a). Bb‟ and Bc‟ saponins are the products of hydrolysis of rhamnose from Bb and Bc, respectively, whereas Be E saponin is the oxidized product of Bb saponin (Price et al. 1986, Güçlü-Üstündağ and Mazza

2007). Analysis of culture media at 0h and after 4h incubation provided information on saponins stability in culture environment. Recovery of type B and E saponins in media after incubation was ~100%. In contrast, ~40% of Ab and Af saponins were recovered, while other type A and all type DDMP saponins were below detection limit. Reactivity of acetyl groups and ketone group with compounds present in medium may had caused degradation of type A and DDMP saponins in the culture environment. With only exception for saponin Ab from soy bread (0.90 ± 0.70 nmol/mg protein), type A saponins were not detected in cells. The low concentration of type A saponins in the aqueous fraction of digesta probably resulted in a concentration below detection limit of the MS in

75 the cells. Further investigation is needed to assess spontaneous cellular metabolism and spontaneous degradation of saponins in the cell culture environment.

The above results expand the previous report of Hu et al. (2004) that indicated very limited uptake of soy (and chickpea) saponins by Caco-2 cells. These data align well with in vivo studies showing very limited uptake of saponins (Hu et al. 2004a, Hu et al.

2004b). Nonetheless, numerous investigators have reported health-promoting activities of saponins. Soy saponins significantly affected nutrient uptake by inhibiting their active transport and increasing their passive diffusion (Onning et al. 1996). Inhibition of the calcium dependent potassium channels with consequent membrane depolarization, altered calcium metabolism (PKC inhibited) and channels deionization were responsible of the altered nutrient uptake (Oh et al. 2001). Hypocholesterolemic activity was demonstrated for soy saponins in animals and humans (Francis et al. 2002) although the mechanism is still not well understood. Proposed mechanisms involved soy saponins complexation with cholesterol in the intestine and cholesterol conversion into bile acids by oxidation in the liver (Francis et al. 2002). Furthermore, soy saponins were shown to suppress the release of proinflammatory mediators by lipopolisaccharide-stimulated peritoneal macrophages (Kang et al. 2005). This suggests that health-mediated activities associated with saponins are initiated by interactions of ingested saponins with the gut epithelium and/or metabolites of saponins produced by the gut microbiota induce such responses. Further inquiry into these possibilities is merited.

3.5

3.0 b b

2.5 b

2.0

1.5 a a a

Uptake (% medium) (% Uptake 1.0 a a a a

0.5

0.0 BaBa Bb Bb’Bb' BcBc BcBc'’ BdBdE E BeEBe E alphagαg betagβg betaaβa Saponin

Figure 27. Uptake (% medium) of saponins from digested soy-chickpea breads by Caco- 2 cells. Cultures of Caco-2 cells were exposed to aqueous fraction generated during simulated digestion of soy-chickpea breads. Total saponin content of test media was 19 ± 1, 14 ± 1 and 9.4 ± 0.3 nmol/ mg cell protein for breads containing soy alone or mixed with 1/3 and 2/3 chickpea, respectively. Data are means ± SD for pooled results for the specific saponin in diluted aqueous fraction after digestion of the three breads (n = 5 independent cultures for each digested bread). One-way ANOVA was performed to compare extent of relative uptake of saponins; different letters above bars indicate significant differences in the extent of uptake (α<0.05).

5.3 Specific Aim 3

5.3.1 Development of pocket-type doughs

The first approach to the development of saponin-rich baked goods focused on the introduction of saponin-rich ingredients (soy flour and soy milk powder) in the formulation of pocket-type flat doughs (Table 2; section 4.3.1).

TGA thermograms of raw doughs (Figure 28) showed increased water binding in soy formulations: first peak shifted from 40 to 60°C due to the different carbohydrate composition of soy formulations (containing less starch and more oligosaccharides;

Kinsella 1979). Water loss from proteins occurred at higher temperature (120°C) in all products but in a higher extent at 10% and 20% soy (Figure 30) possibly due to their high amount of soy proteins which had high water binding capacity than wheat gluten

(Kinsella 1979; Nordqvist 2010). The highest soy addition (26%) did not follow this trend suggesting that excessive amount of wheat substitution with soy no longer improves dough hydration, possibly due to a destabilization of the dough.

Figure 28 - TGA thermograms of raw doughs with different % of soy blend added

The baked products were characterized by a thin shape (130x45x3mm, weight ~5 g) and consisted of two thin layers (1mm each) separated by a 3mm air pocket. Soy addition in the amount of 10% significantly reduced hardness of baked dough (25 N;

Figure 31; Table 8) yet no further softening was observed with increased soy addition (20 and 26%; Figure 31; Table 8). Along with hardness, chewiness significantly decreased from about 500 N to about 200 N at 10% soy and at higher soy addition (20 and 26%) an additional decrease in chewiness was observed giving soft and less chewy characteristics to the soy products. Nonetheless, springiness and cohesiveness of soy containing

79 products were approximately twice as high as in wheat dough (Figure 29; Table 8) suggesting those products more rubbery and tough (Bourne 2002).

10 Control 10% Compressive load (kgf) load Compressive 20% 26% 0 1 5 10 15 Time (sec)

Figure 29 - TPA of pocket-type baked formulations (Table 1, section 4.3.1)

The moisture contents of conventionally baked products were significantly higher in soy formulations (Table 8) as was previously observed in bread (Vittadini and

Vodovotz 2003) due to the hydrophilic characteristics of soy (Kinsella 1979). The amount of “freezable” water of soy containing products was approximately twice that of wheat dough (15.0 vs. 7.09 g water/100 g sample; Table 8, Figure 30). The amount of

“freezable” water in baked products was influenced by the amount of soy and water added to each formula. Incorporation of the soy blend in doughs required higher amount of water in formulations to completely hydrate the ingredients due to the highly hydrophilic soy proteins (Kinsella 1979).

Figure 30 - DSC thermograms of pocket-type product formulations (Table 1; section 4.3.1)

NMR was used to study the state of water from a molecular perspective. Figure

31 shows that for longer pulse trains in the T2 sequence, 3 peaks are observed in the spectrum as has been previously observed (Lodi et al. 2007). Peak 1 has the same shift as that observed in the T1 experiment and is assigned to water.

PeakPeak 1 Length of CPMG

1 ms

5 ms

Peak 2 2

10 ms PeakPeak 3

25 ms

1414 1212 1010 88 66 4 4 22 0 --2 --4 FrequencyFrequency ( ppm(ppm))

Figure 31 – 1H spectra (obtained using CPMG) used in the calculation of the spin-spin relaxation time of fresh doughs. Peaks 1, 2 and 3 were attributed to water, carbohydrate and lipid protons, respectively (Lodi et al. 2007)

Soy addition significantly affected both relaxation times T1 and T2 in baked products. Spin - lattice relaxation time (T1) ranged from 402 ms for the wheat product to

272 ms of 26% (most soy added) dough (Table 8). The shorter T1 value, the more solid like the 1H are, therefore addition of soy decreased the 1H mobility in the conventionally

1 baked product. Spin – spin relaxation time (T2) describes the mobility of H spins relative to each other (Doona and Baik 2007). T2 were longer in soy products than in the wheat dough (about 2.5 versus about 2 ms, p value <0.001, Table 8) depicting higher water mobility in soy products; similar findings about T2 were previously described for soy bread (Lodi et al. 2007). Previous study on gels (Choi and Kerr 2003) showed that wheat starch concentration is inversely related with water mobility. Gels with less wheat starch were characterized by a longer T2 as seen in this study on soy doughs: higher soy amount was related to longer transverse relaxation time.

High T2 indicates less swelling of starch granules and absorption of water (Wang et al. 2004) and it has been attributed to higher moisture content and increase in free volume therefore higher mobility of 1H being likely associated with molecules such as gluten, starch and glucose (Assifaoui et al. 2006; Doona and Baik 2007).

Parameter Control 10% 20% 26% p value

Hardness (N) 25 ± 5.2a 5.3 ± 2.0b 4.1 ± 0.1b 3.0 ± 0.5b <0.001

Chewiness (N) 517 ± 145a 167 ± 55b 32 ± 1b 100 ± 24b 0.001

Springiness (mm) 1.9 ± 0.3a 3.0 ± 0.2ab 2.3 ± .3b 3.2 ± 0.4ab 0.057

Cohesiveness (ratio) 0.36 ± .04a 0.53 ± .02ab 0.56 ± .09b 0.58 ± .06b 0.016

Moisture Content 30.0 ± 1.9a 32.7 ± 1.9ab 35.8 ± 0.6bc 38.5 ± 0.8c 0.002 (g water/100 g sample)

Freezable Water 7.09 ± 1.1a 15.0 ± 2.5b 15.4 ± 0.1b 14.9 ± 1.0b 0.001 (g water/100 g sample)

1 a ab ab b H T1 (ms) 402 ± 40 313 ± 7 347 ± 6 272 ± 2 0.044

1 a b b b H T2 (ms) 2.0 ± 0.0 2.6 ± 0.02 2.7 ± 0.0 2.6 ± .1 <0.001

Table 8 - Physicochemical characterization of baked pocket-type products (Table 2; 4.3.1). Different letters refer to statistically different samples (α 0.05)

The higher lipid content of soy formulations (soy blend consisted of 9% lipids) possibly plasticized gluten and prevented starch swelling by emulsifying the amylose- gluten network thus softening baked doughs (Ghiasi et al. 1984). Polar lipids such as

83 monoglycerides prevent amylose leaching from starch granules (Eliasson and

Gudmundsson 1996). Lipid functionality is also related to its effect on the stability of the gas cells. In this respect, the positive influence of the polar lipids is attributed to their ability to form lipid monolayers at the gas/liquid interphase of the gas cells, thus increasing the gas retention of the dough (Gan et al. 1995). For breads formulated with flour containing low gluten, the increase in lipids and emulsifier contents decreased the firmness and increased the specific volume of breads (Ozmutlu et al. 2001). 1H NMR relaxation studies confirmed these findings by showing shorter T1 and longer T2 in baked formulations compared to the wheat product (Table 8) thus suggesting the development of a more solid-like product with the addition of soy.

5.3.2 Development of saponin-rich breads

Addition of soy saponins extract to soy bread

The second approach in the utilization of extracts and isolates of saponins as a means to increase saponin content of baked goods.

Soy bread was modified by addition of soy saponin extract in the amount of 1% w/w of the original formulation (Table 3; section 4.3.2). Soy bread was prepared using a patent-pending formulation and baked in 100g-pup loaves pans. Enrichment of soy bread was achieved by adding 1% of soy saponins extract (kindly donated by Dr. Kerem) to the original formula. Physicochemical properties of the so obtained product were tested by

TPA, TGA and DSC at day 0 (2 hours after bread was produced) and after accelerated storage (7 days at +4°C, temperature at which the highest staling rate occurs, Chinachoti and Vodovotz 2000).

Soy saponins significantly reduced firmness, hardness and chewiness of soy bread after storage. Hardness of stored bread containing saponins was only 2.1 ± 0.04 N compared to the 6.1 ± 0.8 N of the control (Figure 32). Saponins also reduced chewiness to a third of the original value. No significant effects were observed in moisture content and “freezable”/unfreezable water ratio (Figure 33). These results show that soy saponins used to increase the nutritional value of soy bread can improve its physicochemical properties after storage. Further studies are needed to investigate the changes in the water status.

Force (N) STD SAP 6.0

day 0 Firmness (N) 1.7 ± 0.1 b 1.4 ± 0 b 4.0 Hardness (N) 1.7 ± 0.1 b 1.4 ± 0.01 b

Chewiness 1.0 ± 0.2 c 0.9 ±0.0 c

2.0 day 7 Firmness (N) 5.8 ± 0.7 a 2.0 ± 0.1 b Hardness (N) 6.1 ± 0.8 a 2.1 ± 0.04 b Chewiness 3.7 ± 0.4 a 1.7 ± 0.1 b

0 0.0 5 10 15Time (sec.)20

Figure 32 - Texture Profile Analysis of fresh and stale soy bread (STD) and soy saponins containing soy bread (SAP), α 0.05

0.0

-0.2

5050 a -0.4 a 40 a a 40 30 b -0.6 30 b Heat Flow Flow (W/g) Heat 20 b b 20 10

-0.8 FW MC (g water/100g sample)water/100g(g MC 100 STD day 0 SAP day 0 STD day 7 SAP day 7

MC (gwater/100gMC sample) 0 -1.0 STD day 0 SAP day 0 STD day 7 SAP day 7 -50 0 50 100 150 Temperature (°C)

Figure 33 - Thermogram of fresh soy bread (STD) and soy saponins containing soy bread (SAP) and their state of water: moisture content (MC) and "freezable" water

Addition of chickpea saponins isolate to wheat bread

Wheat bread was fortified by addition of chickpea saponins isolate in the amounts of 1 and 3% w/w of the formulation. Control bread was produced by following the official method (10-10B AACC, 2000). Saponins fortified breads were formulated by substituting part of the wheat flour with the chickpea extract in the amounts of 1% and

3% w/w (Table 4; section 4.3.3). Saponin content and profile were determined by LC-MS

(as described in section 4.1.2) and results are listed in Table 9. As expected, the chickpea isolate did not contain type A saponins. Total saponin content of 1% CP and 3% CP breads were well below that of soy bread (6.3 and 20 µg/g vs. 605 µg/g; Tables 7 and 9) and consisted mainly of βg saponin.

Saponin MW 1% CP 3% CP (g/mol) (µg/g bread) (µg/g bread) A type Ab 1437 nd nd Ac 1421 nd nd Ad 1407 nd nd Af 1275 nd nd Ah 1245 nd nd B type Ba 959 nd nd Bb 943 1.0 ± 0.2 1.9 ± 0.4 Bb‟ 797 nd nd Bc 913 nd nd Bc‟ 767 nd nd E type Bd E 957 nd nd Be E 941 0.14 ± 0.03 0.40 ± 0.03 DDMP type αg 1085 nd nd βg 1069 5.1 ± 1.4 18 ± 1 βa 1039 nd nd γa 893 nd nd γg 923 nd nd

Total 6.3 ± 1.6 20 ± 1

Table 9 – Saponin profile and composition of the chickpea isolate fortified wheat breads (Table 4; section 4.3.2)

Preliminary test trials on bread formulation have shown increased sweetness associated to increasing addition of chickpea saponin isolate. Therefore, sugar content was reduced in chickpea containing formulations (Table 4). Physicochemical properties of the different formulated breads were assessed by TPA, TGA and DSC at day 0.

Addition of the chickpea saponin isolate to the formulation of wheat bread resulted in significantly harder and chewier texture, proportional to the amount of isolate

88 added and double compared to the control at 3% addition (Table 10). Interestingly, addition of 3% chickpea isolate resulted in higher springiness (8.63 vs. 8.19 mm; Table

10) thus depicting a more elastic structure of the breads. Moreover, cohesiveness significantly decreased at 3% addition (p value 0.020; Table 10) depicting a less tough texture. A significant increase of the specific loaf volume was observed in the isolate containing breads, independent of the amount added (5.1 vs. 4.4 cc/g, p value 0.005;

Table 10).

Changes in color were observed for L* (1% addition resulted in brighter crust) and a* (3% addition resulted in higher red component than 1%; Table 10). Overall, chickpea isolate addition resulted in minimal changes of color, independent of the amount added (ΔE ~3; Table 10). Changes in moisture distribution were observed.

Although moisture content did not significantly differ among formulations (p value

0.069; Table 10) the amount of “freezable” water was significantly lower in the isolate containing breads, independently of the amount added (~20.5 vs. 23.8 g water/100 g sample; p value <0.001; Table 10).

Parameter Control 1% CP 3% CP p value

Hardness (N) 0.9 ± 0.0a 1.6 ± 0.0b 2.3 ± 0.2c <0.001

Chewiness (N) 6.5 ± 0.3a 11 ± 0.5b 14 ± 0.7c <0.001

Springiness (mm) 8.19 ± 0.23a 8.36 ± 0.12ab 8.63 ± 0.04b 0.032

Cohesiveness (ratio) 0.66 ± 0.02a 0.65 ± 0.04a 0.58 ± 0.03b 0.020

Specific Loaf Volume 4.4 ± 0.1a 5.1 ± 0.1b 5.1 ± 0.1b 0.005 (cc/g)

Color L* 58 ± 1a 61 ± 1b 56 ± 1c 0.001

Color a* 11.9 ± 0.8ab 10.5 ± 0.6a 12.9 ± 0.4b 0.008

Color b* 34.3 ± 1.2a 33.4 ± 0.6a 35.4 ± 0.7a 0.860

Color ΔE Reference 3.2 ± 1.3a 2.8 ± 1.5a 0.839

Moisture Content 36.6 ± 0.8a 36.7 ± 0.7a 33.5 ± 1.9a 0.069 (g water/100 g sample)

Freezable Water 23.8 ± 0.4a 20.3 ± 0.1b 20.9 ± 0.5b <0.001 (g water/100 g sample)

Table 10 - Physicochemical characterization of the chickpea saponin fortified wheat breads (Table 4; section 4.3.2). Different letters refer to statistically different samples (α 0.05)

Chickpea saponin isolate, kindly donated by Dr. Kerem, was obtained by extraction from chickpeas (Cicer arietinum L.) in 70% aqueous ethanol. This extraction procedure, applied also to soybean-derived material, yields a product consisting of saponins, isoflavones, soluble fiber and other carbohydrates (predominantly sucrose,

90 raffinose, and stachyose) (Dobbins 2002). Therefore, it is likely that the chickpea saponin isolate had a comparable composition.

Soluble fiber present in the isolate was possibly responsible for the firmer texture as stated by others (Bonafaccia et al. 2000; Wang et al. 2002). Significant increase in hardness was observed upon addition of soluble fiber >1% bread (3% substitution of bread formulation with an inulin extract containing 47% soluble fiber; Wang et al. 2002).

The soluble fiber content of the chickpea saponin isolate used in our experiments will need to be assessed, but considering its low saponin content (~0.01%) it was estimated to be >90%.

Soluble fiber, as well as saponins, may have contributed to a better plasticization of the gluten-starch network (as shown by the lower amounts of “freezable” water) thus resulting in more elastic structure (higher springiness) and therefore higher specific loaf volume (Table 10). Lower freezable water suggested both gluten-starch plasticization and water absorption by either water soluble fiber (oligosaccharides) and/or saponins.

Similarly, Bonafaccia et al. (2000) found that breads made with ingredients rich in soluble fibers resulted in firmer texture and higher volume. Furthermore, Nilufer et al.

(2008) found that addition of soy fiber at low levels (2% bread) enhanced loaf volume while high levels resulted in decreased loaf volume due to a reduction of dough gas retention ability.

Fortification of wheat bread with the chickpea saponin isolate resulted in two different effects on texture: crumb hardening, increased specific loaf volume (Figure 34).

While crumb hardening is a negative textural change, increased specific loaf volume is a positive outcome. Therefore, addition of 1% chickpea saponin isolate resulted in

91 acceptable texture, differently from 3% isolate addition (same volume, higher hardness;

Figure 34).

3.0 6

2.5 5

2.0 4

1.5 3

Hardness (N) Hardness 1.0 2

Specific Volume (cc/g) Volume Specific 0.5 1

0.0 0 Control 1% CP 3% CP

Figure 34 - Hardness (blue dots) and specific volume (red dots) of the chickpea saponin fortified wheat breads (Table 4; section 4.3.2)

5.3.3 Development of breads containing saponin-rich ingredients

The ultimate goal of this research was to develop saponin-rich baked goods by using ingredients that are naturally rich in these compounds, rather than using extracts. In order to achieve this goal, soy bread was reformulated by partially substituting the soy blend with a chickpea protein isolate (Table 11). The addition of a chickpea ingredient was designed to increase the amount of DDMP type saponins (specifically βg) and to test the feasibility of chickpea as a matrix to deliver saponins in leavened baked goods.

Addition of 2/3 chickpea protein isolate resulted in doubling of the hardness compared to the soy bread control (13 vs. 7.6 N, p value < 0.001; Table 11). Chewiness was significantly higher only for 2/3 CPP compared to the other two breads (82 vs. 55 N; p value 0.002; Table 11). The highest addition of chickpea protein isolate was characterized by a lower specific loaf volume compared to the other two breads (1.7 vs.

2.0 cc/g; p value 0.022; Table 11). Chickpea addition significantly affected crust color, resulting in a brighter and less yellow color, regardless of the amount added, as shown by the high values of ΔE (~24, p value 0.071; Table 11). No significant difference was observed in moisture content and macroscopic state of water.

Parameter SOY 1/3 CPP 2/3 CPP p value

Hardness (N) 7.6 ± 0.5a 9.1 ± 0.5b 13 ± 0.1c <0.001

Chewiness (N) 55 ± 7a 56 ± 2a 82 ± 3b 0.002

Springiness (mm) 8.2 ± 0.2a 7.8 ± 0.3a 7.8 ± 0.1a 0.160

Cohesiveness (ratio) 0.67 ± 0.04a 0.56 ± 0.04a 0.59 ± 0.3a 0.087

Specific Loaf Volume 2.0 ± 0.1a 1.9 ± 0.1a 1.7 ± 0.0b 0.022 (cc/g)

Color L* 35 ± 3a 52 ± 2b 52 ± 1b <0.001

Color a* 18 ± 2a 20 ± 1a 18 ± 1a 0.269

Color b* 21 ± 4a 1.9 ± 1.1b 5.9 ± 1.1b 0.001

Color ΔE Reference 26 ± 2a 23 ± 1a 0.071

Moisture Content 45.4 ± 1.3a 45.9 ± 1.2a 45.8 ± 0.8a 0.901 (g water/100 g sample)

Freezable Water 32.3 ± 1.1a 31.2 ± 2.1a 32.1 ± 2.8a 0.868 (g water/100 g sample)

Table 11 - Physicochemical characterization of the breads formulated with different amounts of soy blend and chickpea protein isolate (Table 5; section 4.3.3). Different letters refer to statistically different samples (α 0.05)

Changes in texture may be explained by the different composition of the two saponin-rich ingredients: while soy blend contains 9% w/w fat, the chickpea protein isolate contains <1% fat. The lower fat content of the chickpea breads possibly explains the firm and chewy texture and the lower specific loaf volume. Polar lipids such as mono- and di-glycerides, as well as emulsifiers are known to enhance gluten-starch

94 plasticization (Ghiasi et al. 1984; Eliasson and Gudmundsson 1996) resulting in softer texture and higher specific volume of breads (Ozmutlu 2001). Nonetheless, analysis of the materials showed that the chickpea protein isolate contained smaller concentration of saponins (Tables 7, 8, 9; section 5.2.2) thus resulting in a decrease of emulsifiers (mainly saponins type A) thus less plasticization of the gluten-starch system (Ozmutlu 2001).

Reformulation of soy bread with the chickpea protein isolate deleteriously affected textural quality of the bread at 2/3 substitution, resulting in higher chewiness and lower specific loaf volume (Table 11; Figure 35). Lower contents of polar lipids and saponins were probably responsible for such textural changes.

100 6

5 80

4 60

Chewiness(N) 2

20 (cc/g) Volume Specific 1

0 0 SOY 1/3 CPP 2/3 CPP

Figure 35 - Chewiness (blue dots) and specific volume (red dots) of the breads formulated with different amounts of soy blend and chickpea protein isolate (Table 5; section 4.3.3)

Chapter 6

Conclusions

Aim 1: To develop rapid and sensitive methods for separation, identification and quantification of soy and chickpea saponins in plant materials and bread

Several food ingredients were screened for saponin profile and composition by

HPLC-PDA. Soy flour contained 17 saponins, all the soy saponins currently known, representing types A, B, E, and DDMP with a majority of DDMP type (~65% total saponins), followed by type A (~25%) and minor amounts of types B and E (~10%).

Prevalent type in chickpea flour was βg (~95%) with minor amounts of Bb (~5%).

Protein isolates contained 3 times more saponins than the respective flours, as expected, yet characterized by a much lower contribution of DDMP as shown for soy (DDMP from

~65 to 40%) and chickpea (βg from ~95 to 65%). Such changes correlated with a higher content of type B thus indicating that DDMP conversion occurred during protein isolation. Semi-preparatory HPLC was used to isolate saponins from flours with an optimized method that yielded high amounts of saponin isolates (soy mixture, chickpea mixture, type A, type B, DDMP; purity >60%) and highly pure standard (βg; purity

>95%). The methods developed allowed a complete and accurate saponins characterization of food and biological samples.

Aim 2: To determine saponins stability, bioaccessibility and intestinal cell uptake using the coupled model in vitro digestion/Caco-2 human intestinal cells

Three breads made with different amounts of soy and chickpea ingredients were tested. Saponins type A and B resulted stable during bread making while type E and

DDMP degraded significantly (recovery in bread ~50% ingredients). Chickpea addition resulted in lower total saponin content and different profile: less type A, similar amount of type B, and more DDMP. Furthermore, chickpea containing breads resulted in lower recovery of type E and DDMP saponins (half than for soy bread). The matrix effect observed in the chickpea containing breads was attributed to their low content of fat, which might act as inhibitor of saponins hydrolysis and reduction.

The coupled model in vitro digestion/Caco-2 cells provided unique information on saponins bioaccessibility. Saponins were stable during digestion (total recovery

>90%) but changes in the profile were observed. Type A and DDMP saponins were partially degraded (recovery ~60 and 80%, respectively). DDMP saponins possibly converted to type B and E as shown by the recoveries of these groups (~120%).

Partitioning of saponins in the filtered aqueous phase (% bread) was ~30% for type A,

60% for type B and DDMP and ~90% for type E. Experiments on bile effect on micellarization rate revealed that bile was essential for the bioaccessibility of type

DDMP, not essential for type B and E (partitioning did not significantly decreased with removal of bile), and deleterious for type A. The higher hydrophobicity of DDMP and the ability of type A saponins to form insoluble complexes with bile may explain these results. The intestinal uptake of soy saponins determined with Caco-2 cells culture was low (~1% medium, regardless of the food matrix) similarly to what described in the

98 literature for Bb. Type A saponins were not detected in most of the cell samples likely due to their low concentration in the aqueous fraction. Extent of uptake of saponins Bb‟,

Bc‟ and Be E (approximately 2-3%) was greater than that of other saponins identified in cells suggesting more efficient uptake or type B saponins oxidation and/or hydrolysis of type Bb into intestinal cells.

Aim 3: To develop functional baked goods with elevated saponin content

Introduction of saponin-rich ingredients (soy flour and soy milk powder) in the formulation of pocket-type flat doughs increased water binding in soy doughs. TGA showed a first peak shift from 40 to 60°C due to the different carbohydrate composition of soy formulations (containing less starch and more oligosaccharides) and higher water binding to proteins at 10% and 20% soy addition. Baked soy products were softer and less chewy, yet rubbery and tough, depicting insufficient plasticization of the gluten-

1 starch network (high amount of “freezable” water and small reduction of H T2 relaxation time). Polar lipids content and hygroscopic nature of soy proteins were possibly responsible for such changes.

Enrichment of soy bread with soy saponins extract (1% w/w) significantly improved texture of the stored product. Firmness, hardness and chewiness of soy bread after accelerated storage (7 days at +4°C) decreased by a factor of 3 in the saponin- containing bread, thus suggesting anti-staling activity of soy saponins. Fortification of wheat bread with chickpea saponins isolate (1 and 3% w/w) increased bread sweetness, thus sugar content was reduced in chickpea formulations. Addition of the chickpea

99 saponin isolate to the formulation of wheat bread resulted in significantly harder and chewier texture, proportionally to the amount of isolate introduced and being double than the control at 3% addition. A significant increase of the specific loaf volume was observed in the isolate containing breads, independent of the amount added (5.1 vs. 4.4 cc/g, p value 0.005; Table 12). Soluble fiber present in the isolate was possibly responsible of the firmer texture. Soluble fiber, as well as saponins, may have contributed to a better plasticization of the gluten-starch system (as shown by the lower amounts of

“freezable” water) thus resulting in more elastic structure (higher springiness) and therefore higher specific loaf volume.

Soy bread has been reformulated by partially substituting the soy blend with a chickpea protein isolate (Table 13). The introduction of a chickpea ingredient was meant to increase the amount of DDMP type saponins (specifically βg) and to test the feasibility of chickpea as a matrix to deliver saponins in leavened baked goods. Reformulation of soy bread with the chickpea protein isolate resulted in yellower crust and deleteriously affected textural quality of the bread at 2/3 substitution, resulting in doubled hardness and chewiness and lower specific loaf volume. Lower contents of polar lipids and saponins were probably responsible for such textural changes.

100

Chapter 7

Future Studies

This study provided unique and complete information on saponins composition and profile of several food ingredients. The rapid and accurate LC-MS method developed will allow future characterization of food and biological samples.

Further investigation on saponins health-mediated activities, as well as in vivo study on animals, will be needed. Investigation of interactions of ingested saponins with the gut epithelium and saponin metabolism by gut microbiota (with subsequent release of saponin metabolites) is merited. Furthermore, the research group of Dr. Kerem at the

Hebrew University of Jerusalem is currently investigating saponins effect on cholesterol metabolism in hamsters fed with saponins-rich chows.

Saponins showed to positively affect bread texture and volume when added as isolates. Nonetheless, saponins resulted in tough and rubbery breads when added as ingredient (especially chickpea, containing lower amount of lipids than the soy blend).

Further investigation is needed to determine how the different carriers of saponins affect organoleptical quality of baked goods by performing sensory analysis.

101

References

AACC Method 10-10B. Optimized Straight-Dough Bread-Making Method. 1983.

AACC Method 74-09.01. Measurement of Bread Firmness by Universal Testing Machine. 2008.

Aldin E, Reitmeier HA, and Murphy P. 2006. Bitterness of soy extracts containing isoflavones and saponins. Journal of Food Science 71(3): S211-S215.

American Heart Association. 2010. URL: http://www.americanheart.org/downloadable/heart/1265665152970DS- 3241%20HeartStrokeUpdate_2010.pdf. Accessed on July 13 2010.

Assifaoui A, Champion D, Chiotelli E, Verel A. 2006. Characterization of water mobility in biscuit dough using a low field 1H NMR technique. Carbohydrate Polymers 64 (2): 197-204.

Baik M-Y and Chinachoti P. 2000. Moisture redistribution and phase transitions during bread staling. Cereal Chemistry 77 (4): 484-488.

Bangham AD and Horne RW. 1962. Action of saponins on biological cell membranes. Nature 196: 952–953.

Berhow MA, Cantrell CL. Duval SM, Dobbins TA, Maynes J, and Vaughn SF. 2002. Analysis and quantitative determination of group B saponins in processed soybean products. Phytochemical Analysis 13: 343-348.

Berhow MA, Kong SB, Vermillion KE, and Duval SM. 2006. Complete quantification of group A and B soyasaponins in soybeans. Journal of Agricultural and Food Chemistry 54 (6): 2035-2044.

102

Bohn, T. et al. Supplementation of test meals with fat-free phytosterol products can reduce cholesterol micellarization during simulated digestion and cholesterol accumulation by Caco-2 cells. J. Agr. Food Chem. 2007, 55(2), 267-272.

Bonafaccia G, Valli G, Francisi R, Mair V, Skrabanja V, and Kreft I. 2002. Characteristics of spelt wheat products and nutritional value of spelt wheat-based products. Food Chemistry 68(4): 437-441.

Bourne M. 2002. Food texture and viscosity. Concept and measurement. Food Science and Technology. International Series. 2nd ed. San Diego: Elsevier Science Imprint. 427 p.

Brady PL and Mayer SM. 1985. Correlations of sensory and instrumental measures of bread texture. Cereal Chemistry 62: 70-72.

Cannon GS and Honary LAT. Soybean based transformer oil and transmission line fluid. U.S. Patent, number 5,958,851, Sep. 28, 1999.

Carlson E. 2009. Saponins: bioactivity and potential impact on intestinal health. Master Thesis. The Ohio State University.

Carr HY and Purcell EM. 1954. Effects of diffusion on free precession in nuclear magnetic resonance experiments. Physical Review 94: 630–638.

Charalampopoulos D, Wang R, Pandiella SS, and Webb C. 2002. Applications of cereals and cereal components in functional foods: a review. International Journal of Food Microbiology 79 (1-2): 131-141.

Chinachoti P and Vodovotz Y. 2000. Bread staling. Contemporary food science series. Boca Raton, Florida: CRC Press 177 p.

Chitchumroonchokchai C, Schwartz SJ and Failla ML. 2004. Assessment of lutein bioavailability from meals and a supplement using simulate digestion and Caco-2 human intestinal cells. Journal of Nutrition 134 (9): 2280-2286.

103

Choi SG and Kerr WL. 2003. Water mobility and textural properties of native and hydroxypropylated wheat starch gels. Carbohydrate Polymers 51 (1): 1-8.

Clydesdale F. 2004. Functional foods: opportunities and challenges. Food Technology 58 (12): 1-52.

Cornejo F and Chinachoti P. 2003. NMR in foods: opportunity and challenges. Proceedings of Magnetic Resonance in Food Science: Latest Developments; 2002 September 4-6; Paris, France. Cambridge, UK: The Royal Society of Chemistry, 2003.

Dalluge JJ, Eliason E, and Frazer S. 2003. Journal of Agricultural and Food Chemistry 51 (12): 3520-3524.

Decroos K, Vincken J-P, Heng L, Bakker R, Gruppen H, and Verstraete W. 2005. Simultaneous quantification of differently glycosylated, acetylated, and 2,3- dihydro-2,5-dihydroxy-6-methyl-4H-pyran-4-one-conjugated soyasaponins using reversed-phase high-performance liquid chromatography with evaporative light scattering detection. Journal of Chromatography A 1072: 185-193.

Derome AE. 1987. Modern NMR techniques for chemistry research. Pergamon Press, Oxford, NY, Seoul, Tokyo.

Dobbins T. 2002. Process for isolating saponins from soybean-derived materials. Mar. 12, 2002. United States Patent US 6,335,816 B1.

Doona CJ, Baik MY. 2007. Molecular mobility in model dough systems studied by time- domain nuclear magnetic resonance spectroscopy. Journal of Cereal Science 45 (3): 257–262.

Doxastakis G, Zafiriadis I, Irakli M, Marlani H, Tananaki C. 2002. Lupin, soya and triticale addition to wheat flour doughs and their effect on rheological properties. Food Chemistry 77 (2): 219-227.

Drewnowski A, Gomez-Carneros C. 2000. Bitter taste, phytonutrients, and the consumer: a review. The American Journal of Clinical Nutrition 72(6): 1424-1435.

104

Eliasson AC, Gudmundsson M. 1996. Starch: Physicochemical and functional aspects. In A.-C. Eliasson (Ed.), New York, NY: Marcel Dekker. Carbohydrates in food. 431–503 p.

Engelsen SB, Jensen MK, Pedersen HT, Nørgaard L, Munck L. 2001. NMR-baking and multivariate prediction of instrumental textural parameters in bread. Journal of Cereal Science 33 (1): 56-69.

Erdman JW. 2000. Soy protein and cardiovascular disease. A statement for healthcare professional from the nutrition committee of the AHA. Circulation 102: 2555- 2559.

Fairbanks VF. 1994. Iron in medicine and nutrition. In: Shils ME, Olson JA, and Shike M, editors. Modern Nutrition in Health and Disease. 8th ed. Baltimore, MD: Lean and Febiger. p 185-213.

Fernandez ML and Berry JW. 1989. Rheological properties of flour and sensory characteristics of bread made from germinated chickpea. International Journal of Food Science & Technology 24 (1): 103-110.

Fessas D and Schiraldi A. 1998. Texture and staling of wheat bread crumb: effects of water extractable proteins and „pentosans‟. Thermochimica Acta 323 (1-2): 17- 26.

Fessas D and Schiraldi A. 2001. Water properties in wheat flour dough I: classic thermogravimetry approach. Food Chemistry 72 (2): 237-244.

Fessas D, Signorelli M, Pagani A, Mariotti M, Iametti S, and Schiraldi A. 2008. Guidelines for buckwheat enriched bread. Journal of Thermal Analysis and Calorimetry 91 (1): 9-16.

Figuerola FE, Estévez AM, and Castillo E. 1987. Supplementation of wheat flour with chickpea (Cicer arietinum) flour. I. Preparation of flours and their properties for bread making. Archivos Latinoamericanos Nutrición 37 (2): 378-387.

Francis G, Kerem Z, Makkar HPS, and Becker, K. 2002. The biological action of saponins in animal systems: a review. British Journal of Nutrition 88: 587-605.

105

Friend CP, Serna-Saldivar SO, Waniska RD, and Rooney LW. 1992. Increasing the fiber content of wheat tortillas. Cereal Foods World 37:325–328.

Gallagher E, Gormley TR, and Arendt EK. 2004. Recent advances in the formulation of gluten-free cereal-based products. Trends in Food Science & Technology 15 (3- 4): 143-152.

Gan Z, Ellis PR., and Schofield JD. 1995. Mini review: Gas cell stabilisation and gas retention in wheat bread dough. Journal of Cereal Science 21: 215–30.

Garrett DA, Failla ML, and Sarama RJ. 1999. Development of an in vitro digestion method to assess carotenoid bioavailability from meals. Journal of Agricultural and Food Chemistry 47: 4301-4309.

Gee JM, Price KR, Ridout CL, Johnson IT, and Fenwick GR. 1989. Effects of some purified saponins on transmural potential difference in mammalian small intestine. Toxicology In Vitro 3: 85–90.

Gestetner B, Birk Y, and Bondi A. 1966. Soya bean saponins-VI: composition of carbohydrate and aglycone moieties of soya bean saponin extract and of its fractions. Phtochemistry 1966, 5 (4): 799-802.

Glauert AM, Dingle JT, and Lucy JA. 1962. Action of saponin on biological membranes. Nature 196: 953–955.

Gu L, Tao G, Gu W, and Prior R. 2002. Determination of soyasaponins in soy with LC- MS following structural unification by partial alkaline degradation. Journal of Agricultural and Food Chemistry 50 (24): 6951-6959.

Güçlü-Üstündağ Ö and Mazza G. 2007. Saponins: properties, applications and processing. Critical Reviews in Food Science and Nutrition 47 (3): 231-258.

106

Harwood HJ Jr, Chandler CE, Pellarin LD, Bangerter FW, Wilkins RW, Long CA, Cosgrove PG, Malinow MR, Marcetta CA, Pettini JL, Savoy YE, and Mayne JT. 1993. Pharmacologic consequences of cholesterol absorption inhibition: alteration in cholesterol metabolism and reduction in plasma cholesterol concentration induced by the synthetic saponin b-tigogenin cellobioside (CP- 88818; tiqueside). Journal of Lipid Research 34: 377–395.

Heng L, Vincken JP, Hoppe K, van Koningsveld GA, Decroos K, Gruppen H, van Boekel MAJS, and Voragen AGJ. 2006. Stability of pea DDMP saponin and the mechanism of its decomposition. Food Chemistry 99: 326-334.

Hu J, Lee SO, Hendrich S, and Murphy PA. 2002. Quantification of the group B soyasaponins by high-performance liquid chromatography. Journal of Agricultural and Food Chemistry 50: 2587–2594.

Hu J, Reddy MB, Hendrich S, and Murphy PA. 2004. Soyasaponin I and sapogenol B have limited absorption by Caco-2 intestinal cells and limited bioavailability in women. Journal of Nutrition 134: 1867-1873.

Hwang DC and Damodaran S. Selective precipitation of fat globule membranes of cheese whey by saponin and bile salt. 1994. Journal of Agriculture and Food Chemistry 42 (9): 1872-1878.

Ikedo S, Shimoyamada M, and Watanabe K. 1996. Interaction between bovine serum albumin and saponin as studied by heat stability and protease digestion. Journal of Agricultural and Food Chemistry 44: 792-795.

Jones DB and Divine JP. 1944. The protein nutritional value of soybean, peanut, and cottonseed flours and their value as supplements to wheat flour. Journal of Nutrition 28 (1): 41-49.

Kang JH et al. 2005. Soybean saponins suppress the release of proinflammatory mediators by LPS-simulated peritoneal macrophages. Cancer Letters 230 (2): 219-227.

Kensil CR. 1996. Saponins as vaccine adjuvants. Critical Reviews in Therapeutic Drug Carrier Systems 13: 1–55.

107

Kerem Z, German-Shashoua H, and Yarden O. 2005. Microwave-assisted extraction of bioactive saponins from chickpea (Cicer arietinum L). Journal of the Science of Food and Agriculture 85: 406-412.

Kerem Z, Vodovotz Y, Bonfil DJ, Schwartz SJ, and Failla M. 2006. Do saponins present in model systems and legume bread modulate cholesterol absorption in vitro and in vivo? BARD Research Proposal.

Kerwin SM. 2004. Soy saponins and the anticancer effects of soybean and soy-based foods. Current Medicinal Chemistry – Anticancer Agents 4 (3): 263-272.

Kim HY, Yu R, Kim JS, Kim YK, and Sung MK. 2004. Antiproliferative crude soy saponin extract modulates the expression of IκBα, protein kinase C, and cyclooxygenase-2 in human colon cancer cells. Cancer Letters 210 (1): 1-6.

Kinsella JE. 1979. Functional properties of soy proteins. Journal of the American Oil Chemists’ Society 56 (3): 244-58.

Kjellin M and Johansson I. 2010. Surfactants from renewable sources. 1st ed. Chippenham, Wiltshire, UK: John Wiley & Sons, Ltd. 320 p.

Kudou S, Tonomura M, Tsukamoto C, Uchida T, Sakabe T, Tamura N, and Okubo K. 1993. Bioscience Biotechnology Biochemistry 57 (4): 546-550.

Kuroda M, Mimaki Y, Hasegawa F, Yokosuka A, Sashida Y, and Sakagami H. 2001. Steroidal glycosides from the bulbs of Camassia leichtlinii and their cytotoxic activities. Chemical and Pharmaceutical Bulletin 49: 726–731.

Lee SO, Simons AL, Murphy PA, and Hendrich S. 2005. Soyasaponins lowered plasma cholesterol and increased fecal bile acids in female golden Syrian hamsters. Experimental Biology & Medicine 230: 472-478.

Lodi A, Tiziani S, and Vodovotz Y. 2007. Molecular changes in soy and wheat breads during storage as probed by nuclear magnetic resonance (NMR). Journal of Agricultural and Food Chemistry 55 (14): 5850-5857.

108

Lodi A and Vodovotz Y. 2008. Physical properties and water state changes during storage in soy bread with and without almond. Food Chemistry 110 (3): 554-561.

Lucas EA, Khalil DA, Daggy BP, and Arjmandi BH. 2001. Ethanol-extracted soy protein isolate does not modulate serum cholesterol in golden Syrian hamsters: a model of postmenopausal hypercholesterolemia. Journal of Nutrition 131: 211-214.

MacDonald RS, Guo J, Copeland J, Browning JD Jr., Sleper D, Rottinghaus GE, and Berhow M. 2005. Environmental Influences on Isoflavones and Saponins in Soybeans and Their Role in Colon Cancer. Journal of Nutrition 135: 1239-1242.

Matsuda H, Li Y, Yamahara J, and Yoshikawa M. 1999. Inhibition of gastric emptying by triterpene saponin, momordin Ic, in mice: Roles of blood glucose, capsaicin- sensitive sensory nerves, and central nervous system. Journal of Pharmacology and Experimental Therapeutics 289: 729–734.

Matsuura M. 2001. Saponins in garlic as modifiers of the risk of cardiovascular disease. Journal of Nutrition 131: 1000S–1005S.

Merz-Demlow BE, Duncan AM, Wangen KE, Xu X, Carr TP, Phipps WR, and Kurzer MS. 2000. Soy isoflavones improve plasma lipids in normocholesterolemic, premenopausal women. The American Journal of Clinical Nutrition 71 (6): 1462- 1469.

Milner JA. 2000. Functional foods: the US perspective. American Journal of Clinical Nutrition 71 (6): 1654S-1659S.

Mimaki Y, Yokosuka A, Kuroda M, and Sashida Y. 2001. Cytotoxic activities and structur-cytotoxic relationships of steroidal saponins. Biological and Pharmaceutical Bulletin 24: 1286–1289.

Mondal A and Datta AK. 2008. Bread baking – a review. Journal of Food Engineering 86: 465-474.

Morrissey JP and Osbourn AE. 1999. Fungal resistance to plant antibiotics as a mechanism of pathogenesis. Microbiological and Molecular Biological Reviews 63: 708–724.

109

Murphy PA, Hu J, Barua K, and Hauck C.C. 2008. Group B saponins in soy products in the U.S. Department of Agriculture-Iowa State University isoflavone database and their comparison with isoflavone contents. Journal of Agricultural and Food Chemistry 56 (18): 8534-8540.

Nilufer D, Boyacioglu D and Vodovotz Y. 2008. Functionality of soymilk powder and its components in fresh soy bread. Journal of Food Science 73 (4): 275-281.

Nordqvist P, Khabbaz F, Malmström E. 2010. Comparing bond strength and water resistance of alkali-modified soy protein isolate and wheat gluten adhesives. International Journal of Adhesion and Adhesives 30 (2): 72-79

. Oakenfull DG. 1981. Saponins in food-A review. Food Chemistry 7 (1): 19-40.

Oakenfull DG and Sidhu GS. 1990. Could saponins be a useful treatment for hypercholesterolemia? European Journal of Clinical Nutrition 44: 79–88.

Olivares M, Pizarro F, Pineda O, Name JJ, Hertrampf E, and Walter T. 1997. Milk inhibits and ascorbic acid favors ferrous bis-glycine chelate bioavailability in humans. Journal of Nutrition 127(7): 1407-1411.

Onning G, Wang Q, Westrom BR, Asp NG & Karlsson SW. 1996. Influence of oat saponins on intestinal permeability in vitro and in vivo in the rat. British Journal of Nutrition 76: 141–151.

Otsuki M. 2000. Pathophysiological role of cholecystokinin in humans. Journal of Gastroenterology and Hepatology 15 (1): 71-83.

Ozmutlu O, Sumnu G, and Sahin S. 2001. Effects of different formulations on the quality of microwave baked breads. European Food Research and Technology 213 (1): 38-42.

Petit PR, Sauvaire Y, Ponsin G, Manteghetti M, Fave A, and Ribes G. 1993. Effects of a fenugreek seed extract on feeding behavior in the rat: metabolic endocrine correlates. Pharmacology Biochemistry and Behaviour 45: 369–374.

110

Potter SM, Jimenez-Flores R, Pollack J, Lone TA and Berber-Jimenez MD. 1993. Protein saponin interaction and its influence on blood lipids. Journal of Agricultural and Food Chemistry 41: 1287–1291.

Potter SM. 1998. Soy protein and cardiovascular disease: the impact of bioactive components in soy. Nutrition reviews 56 (8): 231-235.

Price KR, Griffiths NM, Curl CR, and Fenwick GR. 1985. Undesirable sensory properties of the dried pea (pisum sativum). The role of saponins. Food Chemistry 17:105–115.

Price KR, Fenwick GR, and Jurzysta M. 1986. Soyasapogenols-separation, analysis and interconversions. Journal of the Science of Food and Agriculture 37(10): 1027- 1034.

Price K.R, Johnson IT, and Fenwick GR. 1987. The chemistry and biological significance of saponins in foods and feeding stuffs. Critical Reviews in Food Science and Nutrition 26: 27–135.

Ridout C, Price KR, Dupont MS, Parker ML, and Fenwick GR. 1991. Quinoa saponins- analysis and preliminary investigations into the effects of reduction by processing. Journal of the Science of Food and Agriculture 54 (2): 165-176.

Riguera R. 1997. Isolating bioactive compounds from marine organisms. Journal of Marine Biotechnology 5: 187–193.

Rosen MJ. 2004. Surfactants and interfacial phenomena. Third edition. Hoboken, New Jersey: John Wiley and sons, Inc. 444 p.

Rudakewich M, Ba F, and Benishin CG. 2001. Neurotrophic and neuroprotective actions of Ginsenosides Rb (1) and Rg (1). Planta Medica 67: 533–537.

Sablani SS, Marcotte M, and Baik OD. 1998. Modelling of simultaneous heat and water transport in the baking process. Lebensmittel Wissenschaft und Technologie 31: 201-209.

111

Sagratini G, Zuo Y, Caprioli G, Cristalli G, Giardinà D, Maggi F, Molin L, Ricciutelli M, Traldi P, and Vittori S. 2009. Quantification of soyasaponins I and βg in Italian lentil seeds by solid-phase extraction (SPE) and high-performance liquid chromatography-mass spectrometry (HPLC-MS). Journal of Agricultural and Food Chemistry 57 (23): 11226-11233.

Schiraldi A, Piazza L, and Riva M. 1996. Bread staling: a calorimetric approach. Cereal Chemistry 73 (1): 32-39.

Seow CC and Teo CH. 2006. Staling of starch-based products: a comparative study by firmness and pulsed NMR measurements. Starch 48 (3): 90-93.

Serventi L, Carini E, Curti E, and Vittadini E. 2009. Effect of formulation on physicochemical properties of nutritionally enhanced tortillas. Journal of the Science of Food and Agriculture 89 (1): 73-79.

Setchell KDR and Cassidy A. 1999. Dietary isoflavones: biological effects and relevance to human health. The Journal of Nutrition 129: 758-767.

Setchell KDR, Brown, NM, Desai, P, Zimmer-Nechemias L, Wolfe BE, Brashear WT, Kirschner AS, Cassidy A, and Heubi JE. 2001. Bioavailability of pure isoflavones in healthy humans and analysis of commercial soy isoflavone supplements. Journal of Nutrition 131: 1362S-1375S.

Shiraiwa M, Kudo S, Shimoyamada M, Harada K, and Okubo K. 1991. Composition and structure of “group A saponin” in soybean seeds. Agricultural and Biological Chemistry 55 (2): 315-322.

Southon S, Johnson IT, Gee JM, and Price KR. 1988. The effect of Gypsophylla saponins in the diet on mineral status and plasma cholesterol concentration in the rat. British Journal of Nutrition 59: 49–55.

Stampfli L and Nersten B. 1995. Emulsifiers in bread baking. Food Chemistry 52 (4): 353-360.

112

Vittadini E and Vodovotz Y. 2003. Changes in the physicochemical properties of wheat- and soy-containing breads during storage as studied by thermal analyses. Journal of Food Science. 68 (6): 2022-2027.

Vodovotz, Y., Ballard, C. Composition and process for making high soy protein- containing bakery products. US Patent 7,592,028 B2. 2009.

Volz FE, Forbes RM, Nelson, WL, and Loosli JK. 1945. Journal of Nutrition 4: 269-275.

Wang J, Rosell CM, and Benedito de Barber C. 2002. Effect of the addition of different fibres on wheat dough performance and bread quality. Food Chemistry 79(2): 221-226.

Wang X, Choi SG, Kerr WL. 2004. Water dynamics in white bread and starch gels as affected by water and gluten content. Swiss Society of Food Science and Technology 37 (3): 377-84.

Williams PC and Singh U. 1987. In The Chickpea, eds. Saxena MC and Singh KB (CAB International, Wallingford, UK), pp. 329-356.

Yamsaengsung R, Schoenlechner R, and Berghofer E. 2010. The effects of chickpea on the functional properties of white and whole wheat bread. International Journal of Food Science & Technology 45: 610-620.

Yáñez E, Ballester D, Aguayo M, and Wulf H. 1982. Enrichment of bread with soy flour. Archivos latinoamericanos de nutrición 32 (2): 417-428.

Yoshikawa M, Murakami T, Kishi A, Kageura T, and Matsuda H. 2001. Medicinal flowers. III. Marigold (1): hypoglycaemic, gastric emptying inhibitory, and gastroprotective principles and new oleanane-type triterpene oligoglycosides, calendasaponins A, B, C, and D, from Egyptian Calendula officinalis. Chemical and Pharmaceutical Bulletin 49: 863–870.

Yoshiki Y and Okubo K. 1995. Active oxygen scavenging activity of DDMP (2,3- dihydro-2, 5-dihydroxy-6-methyl-4H-pyran-4-one) saponin in soybean seed. Bioscience Biotechnology and Biochemistry 59: 1556–1557.

113

Yoshiki Y, Kudou S, and Okubo K. 1998. Relationship between chemical structures and biological activities of triterpenoids saponins from soybean (Review). Bioscience Biotechnology and Biochemistry 62: 2291–2299.

Zhang YC, Albrecht D, Bomser J, Schwartz SJ, and Vodovotz Y. 2003. Isoflavone profile and biological activity of soy bread. Journal of Agricultural and Food Chemistry 51 (26): 7611-7616.

114