Evaluating the Mechanism of Ascorbic Acid Bleaching of and Proposed Ways to Mitigate the Interaction

DISSERTATION

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

By

Jacob E. Farr, B.S.

Graduate Program in Food Science and Technology

The Ohio State University

2018

Dissertation Committee:

M. Mónica Giusti, Ph.D., Advisor

Lynn Knipe, Ph.D.

Luis Rodriguez-Saona, Ph.D.

Christopher Simons, Ph.D.

Copyrighted by

Jacob E. Farr

2018

Abstract

Anthocyanins are a class of polyphenolic compounds which produce a wide range of hues in nature. Their use as a natural colorant in foods has been hindered by ascorbic acid bleaching, producing faded color expression and unappealing products. Ascorbic acid is a robust antioxidant in many food systems and is also a nutrient essential to diet.

Therefore, finding ways to stabilize the pigment in response to ascorbic acid is essential in aiding the food industry’s transition to naturally-colored foods. Mitigation strategies must be investigated and solutions proposed for this deleterious interaction.

Anthocyanins possess great structural diversity and are considered highly reactive molecules. It was hypothesized that pigment reactions such as and formation of could be strategies for reducing color loss caused by ascorbic acid. In addition, the effect of pigment structure may sterically hinder the site of reaction. These hypotheses rely on reducing accessibility to Carbon-4, the suspected site of condensation between the and ascorbic acid. The objective of these studies was to evaluate means of stabilizing the anthocyanin and its color against ascorbic acid fortification.

Copigmentation with green tea catechins initially enhanced color expression

(decreased Lightness, increased chroma and maximum absorbance) but provided limited protection against ascorbic acid-induced degradation of anthocyanins. The strength of

ii interaction between green tea catechins and anthocyanins was considered too weak to reduce accessibility of ascorbic acid to the anthocyanin chromophore.

Structural modification of the pigment was then investigated. Pyranoanthocyanins are derived pigments that contain a fourth ring attached at Carbon-4. The Carbon-4 site was assumed as the primary, but not singular, place of reactivity and that stabilization of color and pigment would be greater compared to anthocyanins. Carboxy- color expression was found to be 8-13x more stable in the presence of ascorbic acid than the anthocyanin counterpart, not due to enhanced pigment stability, but because of the formation of colored degradation products between the derived pigment and ascorbic acid.

The enhanced color stability of pyranoanthocyanins against ascorbic acid bleaching spurred investigation into how anthocyanin structural patterns influence the formation of pyranoanthocyanins. Mono-, di-, and tri-glycosylated C3 derivatives, and C3, C5 glycosylation were compared for the rate of formation. Di- glycosylated anthocyanins with 1→6 glycosyl linkages had the highest formation of pyranoanthocyanins, followed by mono-glycosylated, the tri-saccharide, and 1→2 disaccharides with the lowest yield. The difference in 1→6 and 1→2 disaccharide pyranoanthocyanin formation was unexpected and attributed to conformational differences of glycosyl branching patterns.

iii Further research into the glycosyl-conformational properties of different cyanidin derivatives was completed to evaluate the differences in kinetics of degradation as well bleaching behaviors for C3 substituted anthocyanins. The 1→2 disaccharides exhibited higher λmax, and enhanced resistance to bleaching and hydration compared to the 1→6 disaccharides; however, their stability was generally lower than their monosaccharide counterparts. The 1→6 disaccharides exhibited similar λmax, slightly improved resistance to hydration and bleaching, and significantly improved half-lives in most pHs. These findings are contrary to assumptions of greater number of glycosyl units equating to enhanced stability.

With the evaluation of bleaching, measurement of color parameters was essential.

ColorBySpectra was developed to convert absorbance spectra data to commonly used color spaces including CIEL*c*h*, CIELab, HunterLab, and tristimulus values. Instead of single-sample measurements, this allowed for efficient measurement using more advanced instrumentation such as a high-throughput plate reader. This had the benefit of enhancing efficiency, more precise color reproduction, and measurement of color being independent of specialized instrumentation.

Stabilization of anthocyanins by formation of pyranoanthocyanins and selection of pigments with certain structural characteristics were found to limit the color loss due to bleaching; whereas green tea catechin copigmentation had limited benefit. Manipulation of stereochemistry was found to be an underlying theme in all studies. iv Acknowledgments

With no hesitation, my first acknowledgment is owed to my advisor, Dr. M.

Mónica Giusti. As a second year undergraduate student enrolled in her course, I knew she possessed a passion and enthusiasm for our field that is rarely matched, and that I wanted to perform research in her lab. Monica is a leader who inspires. She epitomizes the word tireless and I have gained much more than just the expertise of knowledge in our field.

Monica knew what I was capable of when I did not. At every point in my graduate career when I was presented with a shortcut or I had chosen to do the lesser, Monica knew what to say to encourage me to do more than what was ever asked of me. She gifted me with permission to pursue an unorthodox education path that is truly mine, and I am honored to have had the opportunity to work in her group.

I must also acknowledge my excellent advisory committee members Dr. Lynn

Knipe, Dr. Luis Rodriguez-Saona, and Dr. Christopher Simons. My choice of committee members came only after enrolling in every course offered by Dr. Rodriguez-Saona and

Dr. Simons: I knew they were the type of people I want on my team. With Dr. Knipe, it was after studying abroad in China and witnessing the true diligence he has with every personal interaction and lecture. These three committee members have consistently provided the optimal challenge and many temporary failures, inevitably to be followed with personal success: this is the greatest gift an educator can provide.

v My last professional acknowledgement is owed to the many lab members, and now life-long friends, I have made along the way. One of the greatest surprises from my time at Ohio State is that people graduate and leave, and unfortunately these people have also become some of your greatest friends. I would like to thank the following people for making the journey fun at every turn: Ronald Bangcuyo, Olivia Geoghegan, Peipei Tang,

Dr. Gregory Sigurdson, and Kevin Wong,. Special thanks is owed to Dr. Gregory

Sigurdson. His commitment to my academic success has been unquestionable and he will make a great professor.

I would also like to thank my wife, Dr. Jessica Barnum Farr for the dedication and sacrifice she has made for me to reach this point. She has given me so much more to look forward to than simply career success. I am incredibly lucky to have a partner to both travel the world and also share in the most ordinary experiences of living. I will always look to the end of my day when I can take a long walk with you and our dogs. My parents are also most worthy of acknowledgements. My mother, Susan Farr, may be one of the greatest examples of deriving purpose from serving others. She is a fearless community leader and teacher who continuously sacrifices for the benefit of others. My father, James

Farr, taught me early on the benefits of working hard, possessing integrity and responsibility, while also having fun at every opportune moment. It is with great pride I become the first doctor in our family.

vi VITA

1992...... Born, Fort Wayne, Indiana

2011...... Paulding High School

2014 ...... B.S., Food Science and Technology, The

Ohio State University

2014 to present ...... Graduate Research Associate, Department

of Food Science and Technology, The Ohio

State University

Field of Study

Major: Food Science and Technology

Minor: Computer Science and Engineering

Publications

Farr JE, Sigurdson GT, Giusti MM. Influence of cyanidin glycosylation patterns on carboxypyranoanthocyanin formation. Food Chem 2018;259:261–9. doi:10.1016/j.foodchem.2018.03.117.

Farr JE, Giusti MM. Investigating the Interaction of Ascorbic Acid with Anthocyanins and Pyranoanthocyanins. Molecules 2018;23:744. doi:10.3390/molecules23040744.

vii Table of Contents

Abstract ...... ii

Acknowledgments...... v

VITA ...... vii

Table of Contents ...... viii

List of Tables ...... xiii

List of Figures ...... xv

Chapter 1. Introduction ...... 1

Chapter 2. Review of Literature ...... 5

2.1. Color in Foods...... 5

2.1.1. Its Role and Historical Use ...... 5

2.1.2. Safety and Regulations ...... 7

2.1.3. Natural Pigments in Foods ...... 8

2.2. The Anthocyanin ...... 11

2.2.1. Structure ...... 11

2.2.2. Equilibrium ...... 14

2.2.3. Stability ...... 15

2.2.4. Potential Health Benefits ...... 16

2.3. Reactions of Anthocyanins ...... 16

viii 2.3.1. Bleaching ...... 16

2.3.2. Copigmentation ...... 19

2.3.3. Pyranoanthocyanins ...... 23

2.3.4. Self-Association ...... 25

2.3.5. Polymerization ...... 26

2.3.6. Metal Complexation...... 26

2.4. Ascorbic Acid in Foods ...... 27

Chapter 3. The role of green tea catechins in modifying the bleaching of anthocyanins

by ascorbic acid...... 29

3.1. Abstract ...... 29

3.2. Introduction ...... 30

3.3. Materials and Methods ...... 33

3.3.1. Materials ...... 33

3.3.2. Methods...... 33

3.4. Results & Discussion ...... 39

3.4.1. Initial Color Enhancement and Spectra Changes...... 39

3.4.2. Copigmentation effect on anthocyanin stability ...... 41

3.4.3. Change in Ascorbic Acid levels ...... 44

3.4.4. Change in chroma and Lightness ...... 46

ix 3.5. Conclusions ...... 48

Chapter 4. Investigating the interaction of ascorbic acid with anthocyanins and

pyranoanthocyanins ...... 49

4.1. Abstract ...... 49

4.2. Introduction ...... 50

4.3. Materials and Methods ...... 54

4.3.1. Materials ...... 54

4.3.2. Methods...... 54

4.4. Results and Discussion ...... 60

4.4.1. UV-Vis Spectrophotometry of solutions ...... 60

4.4.2. Kinetics of degradation ...... 62

4.4.3. Colorimetry ...... 64

4.4.4. HPLC and MS/MS Evaluation...... 68

4.5. Conclusions ...... 73

Chapter 5. Influence of cyanidin glycosylation patterns on carboxypyranoanthocyanin

formation ...... 74

5.1. Abstract ...... 74

5.2. Introduction ...... 75

5.3. Materials and Methods ...... 78 x 5.3.1. Materials ...... 78

5.3.2. Methods...... 80

5.4. Results & Discussion ...... 87

5.4.1. Kinetics of pyranoanthocyanin formation ...... 87

5.4.2. Changes in Spectra during Reaction with Pyruvic Acid ...... 93

5.4.3. Changes in Color during Reaction with Pyruvic Acid...... 97

5.5. Conclusion ...... 100

Chapter 6. Stereochemistry of C3-Glycosidic Attachments Affects the Reactivity of

Cyanidin Derivatives ...... 102

6.1. Abstract ...... 102

6.2. Introduction ...... 103

6.3. Materials and Methods ...... 106

6.3.1. Materials ...... 106

6.3.2. Methods...... 107

6.4. Results ...... 113

6.4.1. Initial vis-max differences at different pH ...... 113

6.4.2. Absorbance Retention after Equilibration at Different pH ...... 115

6.4.3. pH: stability & half lives ...... 117

6.4.4. Bisulfite Bleaching...... 119 xi 6.4.5. Ascorbic Acid Bleaching ...... 122

6.5. Conclusion ...... 126

Chapter 7. ColorBySpectra: an application to automate conversion of spectral data to

color spaces ...... 127

7.1. Abstract ...... 127

7.2. Introduction and Background ...... 128

7.2.1. Background Information on Color ...... 128

7.2.2. Problem and Target Resolution ...... 134

7.3. Software Architecture, Framework, and Implementation ...... 135

7.4. Software Functionalities ...... 138

7.5. Results ...... 138

7.5.1. Experimental Details ...... 139

7.5.2. Data Generated...... 139

7.6. Illustrative Examples ...... 141

7.7. Conclusions ...... 141

Chapter 8. Conclusions and Future Work ...... 142

References ...... 145

xii List of Tables

Table 1. Kinetic Parameters of anthocyanins in response to ascorbic acid and green tea

catechin addition and change in color (E) for blackberry, black carrot, and red

cabbage anthocyanins. Reaction rate and half-life calculated as first-order kinetics

from HPLC...... 42

Table 2. Reaction rates and half-life (t½) of solutions colored with different pigments

stored at 25 °C in the dark, modeled with first-order kinetics. Calculations are

based on the changes in Absorbance at the vis-max of the solution over time...... 63

Table 3. Day 0 and 5 colorimetric values (CIEL*c*h*) and total color change (ΔE) of

chokeberry, cyanidin-3-galactoside and 5-Carboxypyranocyanidin-3-galactoside

for all AA levels over time. Numbers are means of 3 replications, followed by

(standard deviations)...... 67

Table 4. λmax (in nm) of glycosylated cyanidin derivatives in pH 1-9. Values reported

are means (n = 3) and (standard deviation). Different superscript letters indicate

significant differences between anthocyanins in the same pH...... 115

Table 5. % absorbance retention of glycosylated cyanidin derivatives in pH 1-9. %

Absorbance retention is defined as Absorbance in pHn / Absorbance in pH1 at

respective λmax ×100, after 15-30 min equilibration. Values reported are means

(n = 3) and (standard deviation) ...... 116

xiii Table 6. Half-lives (hr) of glycosylated cyanidin derivatives in pH 1-9, stored at 25 ˚C in

dark. Values reported are means (n = 3) and (standard deviation). Different

superscript letters indicate significant differences between anthocyanins in the

same pH. R2 ≥ 0.90...... 119

Table 7. Bleaching kinetics of cyanidin-3-glycosylates in response to ascorbic acid at pH

3 over a week (25 C, dark). First-order kinetics applied...... 125

xiv List of Figures

Figure 1. The four classes of plant pigments commonly found in dietary sources along

with the source ...... 10

Figure 2. The basic chemical structure of (aglycones) ...... 13

Figure 3. Equilibrium forms of anthocyanidins. Colorless forms (left) expressed at pH 3-

6. The flavylium cation (center) most predominant at pH <3 and the quinonoidal

bases (right) preferred at pH >6 ...... 14

Figure 4. Proposed bleaching mechanism of anthocyanins in response to bisulfite.

Condensation occurring primarily at Carbon-4 (bottom arrow) and to a lesser

extent at Carbon-2 (upper arrow) ...... 18

Figure 5. Intermolecular copigmentation showing planar stacking with gallic acid (left)

and intramolecular copigmentation showing folding over chromophore with

coumaric acid acyl group ...... 20

Figure 6. Primary equilibrium of vitamin C with the redox reaction between the vitamers

ascorbic and dehydroascorbic acid ...... 27

Figure 7. Initial changes in spectra and color characteristics of blackberry, black carrot,

and red cabbage anthocyanins in response to green tea catechins...... 40

Figure 8. Spectral changes of blackberry, black carrot, and red cabbage anthocyanins

with ascorbic acid and green tea catechins added over a 12 day period...... 44

xv Figure 9. Ascorbic acid content of mixtures of anthocyanin extracts and green tea

catechins over a 12 day period at 25 C ...... 46

Figure 10. Change in Lightness (L*) of blackberry, black carrot, and red cabbage in

response to green tea catechin and ascorbic acid addition ...... 47

Figure 11. Change in Chroma (c*) of blackberry, black carrot, and red cabbage in

response to green tea catechins and ascorbic acid addition ...... 48

Figure 12. Formation of pyranoanthocyanin from cyanidin and pyruvic acid by

heterocyclic addition ...... 53

Figure 13. Spectral absorbance changes in response to 500 mg/L AA for chokeberry,

cyanidin-3-galactoside and 5-Carboxypyranocyanidin-3-galactoside colored

solutions over a 5 day period ...... 61

Figure 14. Reaction rates of 5-Carboxypyranocyanidin-3-galactoside, Cyanidin-3-

galactoside, and chokeberry plotted against ascorbic acid level (0-1000 mg/L).

Calculations are based on the changes in Absorbance at the vis-max of the solution

over time...... 64

Figure 15. Colorimetric changes (CIEL*c*h*) of change of solutions colored with

chokeberry, cyanidin-3-galactoside and 5-Carboxypyranocyanidin-3-galactoside

from day 0 to day 5 for all AA levels over time. Error bars represent standard

deviation ...... 66

xvi Figure 16. HPLC profiles (470-520 nm) for 5-Carboxypyranocyanidin-3-galactoside,

Cyanidin-3-galactoside, and chokeberry with 1000 mg/L AA added on day 0 and

1...... 70

Figure 17. Anthocyanin and pyranoanthocyanin aglycone structure along with sugar

substitutions of selected cyanidin isolates.#→# showing glycosidic linkage of

substitutions. Φ and Ψ angles shown exemplifying rotation of 1→2 glycosidic

bonds and added ω rotation in 1→6 glycosidic bonds. am/z data from Q3 Scan of

MS-MS, cyanidin (287 m/z) and carboxypyranoanthocyanidin aglycone (355 m/z)

found for all isolates and newly formed pyranoanthocyanins. Formed

Pyranoanthocyanin parent m/z was +68 by addition of fourth ring from the

anthocyanin Q3 scan. bPurity described as A.U.C. (260-700 nm) of isolate peaks

to profile...... 80

Figure 18. Pyranoanthocyanin yield (top) for anthocyanins (500 µM) subjected to pyruvic

acid treatment (x100 molar ratio) and anthocyanin survival (bottom) of control

treatments at pH 2.5 acidified water at day 7, 14, and 31. Yield defined as

(AUC500-520nm of PACN at tn / AUC500-520nm of ACN at t0)*100. Survival defined

as (AUC500-520nm of ACN at tn / AUC500-520nm of ACN at t0)*100. Standard

deviation reported as error bars...... 88

Figure 19. Pyranoanthocyanin formation over time as monitored by HPLC. Anthocyanin

(ACN) and pyranoanthocyanin (PACN) peaks labeled. #→# showing glycosidic xvii bond of each substitution of C3 for anthocyanin. *Cy3rut at day 14 and 31 ran

under alternative HPLC conditions, See Section 2.2.3...... 90

Figure 20. Changes in spectral characteristics of anthocyanin isolates (500 µM) treated

with pyruvic acid (x100 molar ratio) in pH 2.5 acidified water, stored over a 31

day period at 25 °C. Standard deviation represented as errors bars and (#)...... 94

Figure 21. Representative UV-Vis absorbance spectra (absorbance standardized at

respective λvis-max) of an anthocyanin and pyranoanthocyanin (cy3xylglugal and

carboxycy3xylglugal), λvis-max (nm), and color characteristics (CIEL*c*h*) of

isolated anthocyanins and their respective pyranoanthocyanins obtained from

HPLC-PDA detector...... 96

Figure 22. Color parameters for CIE Lightness, Chroma and Hue angle for anthocyanins

(500 µM) subjected to pyruvic acid treatment (x100 molar ratio) in pH 2.5

acidified water after 31 days of storage at 25 °C. Standard deviation represented

as errors bars and (#)...... 99

Figure 23. Cyanidin aglycone structure, different glycosyl moieties, branching patterns

(linkage), and purity of isolated pigments. Purity defined as the % area under

curve of the target anthocyanin to any other peaks present in the PDA

chromatogram (260-700 nm) ...... 107

Figure 24. Bleaching of cyanidin isolates with differing number of glycosylations and

patterns in response to bisulfite addition at pH 3. Isolate Name reported alongside xviii % of bleaching in response to 50 ppm (orange arrow) and 200 ppm bisulfite (blue

arrow) ...... 120

Figure 25. Common illuminants selected for ColorBySpectra and their tristimulus values

and chromaticity coordinates of the light source ...... 130

Figure 26. Calculation of Tristimulus values used in calculating Color Spaces ...... 131

Figure 27. Visual depiction of CIEL*c*h* and CIEL*a*b* color spaces ...... 133

Figure 28. Equations used in calculating color parameters ...... 134

Figure 29. Implemented strategy for parsing input spectral data and producing outputted

color parameters ...... 136

Figure 30. Visual comparison of color data generated using ColorBySpectra to object

images and colorimeter software reproduction ...... 140

xix

Chapter 1. Introduction

Color contributes greatly to the organoleptic characteristics of food items.

Manufacturers utilize color to associate product identity, mask natural variations, and to offset color loss due to processing. Consumers utilize it as a tool in evaluating the safety, nutritional contents, flavor expectations, and overall quality of products. The use of synthetic dyes in coloring foods is widespread due to their cost, tinctorial strength, and stability in a wide range of applications; however, consumers have expressed concern regarding their associated safety implications. Many food companies have announced ambitious plans of eliminating synthetic dyes and lakes in their product lines with plans to transition to natural sources of pigments. Naturally, manufacturers have viewed anthocyanins as an alternative coloring agent.

Anthocyanins are a class of polyphenols found in a wide range of fruits and vegetables common to diet. They exhibit hues which include blue, purple, orange, red and even green. They are structurally diverse molecules which contain different substitutions on the B-ring (-H, -OH, -OCH3), are glycosylated with different type, sites, and number of sugars, and frequently contain the addition of aliphatic or aromatic acids extending from the glycosylation. They have a pH-dependent equilibrium and shift between colored and colorless forms. Their stability is highly influenced by many environmental factors including light, pH, enzymes, oxygen, and bleaching agents. The 1 latter is of particular concern for food due to color fading, resulting in unappealing products. Ascorbic acid is a bleaching agent which has been widely adopted for use in foods as an antioxidant and accelerates the loss of color from anthocyanins.

The objective of these studies was to evaluate reactions of anthocyanins and the role of anthocyanin structure in bleaching with ascorbic acid. Previous studies have proposed condensation of ascorbic acid at Carbon-4 of the anthocyanin. Our hypothesis was reactions and structural modifications of anthocyanins may modify accessibility to

Carbon-4 in ways that reduce bleaching of the pigment.

The first chapter evaluates copigmentation with green tea catechins as a means of reducing ascorbic acid bleaching. Copigmentation is a - and hydrophobic interaction that results in planar stacking of a colorless (or slightly colored) compound and the anthocyanin, enhancing color production. It was thought this planar stacking would reduce accessibility to Carbon-4 and provide a feasible solution for the food industry to stabilize color. Copigmentation with green tea catechins has been previously found to enhance color of anthocyanins. Blackberry, black carrot, and red cabbage were selected due to being different substituted cyanidin derivatives. This allowed for comparison of copigmentation with green tea catechins response as well as bleaching behaviors.

The following chapter looked at comparing the reactivity of pyranoanthocyanins to anthocyanins in response to ascorbic acid. Pyranoanthocyanins are derived pigments first reported in red wines. Polar enolizable groups, often small yeast metabolites, react with the anthocyanin and form a fourth ring extending from Carbon-4 and Carbon-5.

Their color is hypsochromically shifted and they do not have colorless equilibrium forms

2 like anthocyanins. The covalent occupation of the suspected site of interaction between anthocyanin and ascorbic acid was thought as possibly an effective way to reduce interaction.

In order for pyranoanthocyanins to be considered a feasible colorant for foods containing ascorbic acid, understanding how anthocyanin substitution patterns influences the formation of pyranoanthocyanins was investigated.. The role of anthocyanin glycosylation in pyranoanthocyanin formation was studied. Nine anthocyanins which differed by number, type, and site of glycosyl attachment were evaluated for carboxypyranoanthocyanin (pyranoanthocyanins resulting from pyruvic acid addition) yield. Substitution at C5 has previously been reported as blocking pyranoanthocyanin formation, yet research into other structural characteristics of anthocyanins and how they may influence pyranoanthocyanin formation has not been reported. This data could aid color manufacturers in their pursuit of forming these derived pigments with more stable color expression for use in foods.

The next chapter investigated how different Carbon-3 substituted cyanidin derivatives react in response to ascorbic acid and bisulfite bleaching as well as kinetics of degradation over pH 1-9. It was thought that conformational properties of the glycosyl units would modify the anthocyanin stability and bleaching behaviors. With bleaching being suspected to occur primarily at Carbon-4, the spatial occupation of the glycosyl moieties would be altered by both the number and bond type of the glycosyl moieties off

Carbon-3 of the anthocyanin. In addition, chemical properties such as electron

3 delocalization and intramolecular hydrogen bonding may further extend to improved stability over the pH ranges evaluated.

With color being a critical component of all research, software development in the domain was pursued. ColorBySpectra was developed using the Java programming language for creation of a scientific tool that converts raw absorbance spectral data to color spaces. These color spaces are essential for color communication and reproduction, yet many researchers rely on specialized measurement instruments (colorimeter) when absorbance data alone is enough to yield color space data for many applications. The software gives users the flexibility to collect absorbance data on the instrument of their choice, selection of the top three illuminants used, and the flexibility to collect in nanometer intervals of their choice. The inputted absorbance data then follows the mathematical models for conversion to the most commonly used color spaces. This tool has been open-sourced for the benefit of color scientists all around the world.

4

Chapter 2. Review of Literature 2.1. Color in Foods

2.1.1. Its Role and Historical Use

Color serves several important functions in food. It aids consumers in assessing safety, quality, flavor, and nutritional value of foods. For food manufacturers, it can provide functionality in standardizing raw materials that are naturally variable, can subsidize color that might be lost due to processing, and can be used to define the identity of a brand (Gregory T. Sigurdson et al. 2017; Belitz et al. 2009; Schwartz et al. 2017).

Color can be naturally present such as the red coloration of or added as is the case with many confectionary products (Belitz et al. 2009; Martins et al. 2016).

The earliest known use of a food colorant dates back to early Egyptians who colored wines as early as 1500 BC (Burrows, J.D. 2009). The use of colorants from both plant and mineral sources has been observed in several other early societies such as

Native Americans in North, Central, and South America for a broad range of uses including artwork, textiles, and food (Moerman 2010; Rowe et al. 2007). A millennia and several centuries later, the first discovery of a synthetic dye was found by William Perkin who synthesized mauve in the year 1856 (Hons et al. 1994). By the early 20th century, many of the foods available in the United States had added synthetic dyes (Gregory T.

Sigurdson et al. 2017). Confectionaries, cereals, dairy, juices, and beverages have been the largest use cases for synthetic dyes (Belitz et al. 2009). 5 In the past couple decades, consumers have expressed preference for naturally colored foods with no use of synthetic dyes. Many large companies in the food industry have declared goals or timelines to fully eliminate synthetic dyes across their product lines. These companies include Kraft, Mars, and most recently Dunkin’ Donuts (Hunt

2015; Mudd 2016; National Public Radio 2015; Domonoske 2018). The shift to natural has not been seamless though. General Mills, for example, transitioned to using and purple carrot to color Trix cereal. After receiving a wave of complaints from consumers, they reversed their decision and have now went back to offering a synthetically-dyed variant (Dewey 2017). Challenges remain for other members of industry to address this desire from consumers.

Color as a visual tool in sensory perception is quite remarkable. It can convince consumers to discount or ignore other sensory stimuli such as olfaction and rely solely on visual cues. Morret and others (2001) have reported that by coloring a white wine red, panelists ignored the sense of odor and used descriptors only associated with red wines such as chicory and berry. Stillman (1993) has reported that panelists struggle to identify raspberry and orange fruit-flavored beverages when made colorless and that correct coloration was critical in identifying both flavors. As shown by the infamous example of

Chrystal Pepsi, a colorless cola, consumers reject products that do not satisfy visual expectations (Murray 2017). It is no wonder why coloring is so important to the overall success of a food product.

6 2.1.2. Safety and Regulations

More recently, concern has arisen around the use of synthetic dyes and their association with hyperactivity disorders (McCann et al. 2007). The implication that certain artificial colorants are capable of causing hyperactivity and other behavioral disorders was first reported as early as 1973, yet a scientific consensus has not been reached around the topic (Arnold et al. 2012). There have also been associations with synthetic dyes and hyperallergenicity and mutagenicity (Kobylewski & Jacobson 2010).

A study published in 2007 resulted in what has been one of the biggest decisions in recent history for use of colorants in foods; the authors claimed that artificial colors (specifically azo-based dyes) and sodium benzoate increased hyperactivity in children (McCann et al.

2007). While the study did not form a direct relationship between the dyes and hyperactivity due to the additive mixture, the regulatory organizations of both the

European Union (EU) and United States reacted to these reported findings (Lehto et al.

2017).

The European Food Safety Authority, acting on behalf of the EU, implemented changes that requires any food product colored with synthetic dyes to contain the label

“may have an adverse effect on activity and attention in children (Lehto et al. 2017).”

The United States’ Food and Drug Administration received a petition requesting for the

FD&C synthetic dyes to be revoked on the same grounds, but the committee responsible for the decision ultimately refused to yield stating “that a causal relationship between exposure to color additives and hyperactivity in children in the general population has not been established (Cheeseman & Ph 2011).” These decisions along with other trends such

7 as clean labeling and preference for natural and organic have been the main drivers in consumers’ preference for naturally-colored foods (Martins et al. 2016).

In the United States, colorants fall under one of the two classifications: exempt from certification or subject to certification (Harp & Barrows 2015). There are currently nine color additives used in foods that are subject to certification commonly referred to as

FD&C color additives. Every batch of these FD&C colorants must be approved by the

Food and Drug Administration prior to use with strict criteria such as the amount of heavy metals that are tolerable (FDA 2018a). Colorants that are classified as exempt from certification include sources such as minerals, skin extract, fruit and vegetable juice, and caramel (FDA 2016). Any company interested in commercializing or utilizing a certain colorant can petition the Food and Drug Administration by submission of a dossier which includes information regarding safety, tolerance limits, applications, and labeling of the proposed colorant (FDA 2018b).

2.1.3. Natural Pigments in Foods

Pigments used to color foods may come from natural sources including plants, animals, and minerals. These all fall under exempt from certification status, but that classification does not imply natural. There are many challenges arising as a result of the shift from synthetic to naturally colored foods. Synthetic dyes typically exhibit superior stability, greater tinctorial strength, lower costs, and provide a wider array of possible hues (Gregory T. Sigurdson et al. 2017).

Further categorizing the plant pigments, there are several classes: chlorophyll

(green), carotenoids (red to yellow), anthocyanins (red, purple, blue), and betalains (red

8 to yellow), seen in Figure 1. These pigments exhibit a wide range of color expressions and functionality which is dependent on their application in foods, making the task of replacing their synthetic counterparts even more challenging. Chlorophylls are lipophilic compounds containing a chromophore which is a porphyrin ring with a magnesium atom in the center (Humphrey 2006). They also include an aliphatic phytol tail extending from the porphyrin ring. The use of Copper and Zinc ions have been used to replace the magnesium atom which enhances and stabilizes the green color expression (Schwartz &

Lorenzo 1990). Their use as a natural colorant has yet to be widely adopted due to the poor stability of the pigment and green hues not being highly sought after.

Carotenoids are another class of lipid-soluble pigments present in sources such as carrots, tomatoes, fungi, and even insects (Mangels et al. 2018). Carotenoids are further categorized into xanthophylls or carotenes. Xanthophylls contain polyunsaturated hydrocarbons with oxygen functional groups while carotenes are identical but lacking oxygen species (Eldahshan et al. 2013). The high degree of unsaturation means they are highly susceptible to degradation caused by light (Pesek & Warthesen 1987). Carotenoids can contain zero to two rings which contribute to the color expression. Carotenoids have found success as use as a food colorant with exempt from certification status with pigments such as β-Carotene being used in different commercial products.

Anthocyanins are the most structurally diverse of plant pigments used to color foods. They are polyphenols found in fruits and vegetables that are composed of a

C6C3C6 backbone usually containing a sugar moiety. They have an equilibrium that is pH dependent and can result in the same pigment having different color expressions in

9 different applications, a challenge to their application. Anthocyanins are susceptible light, heat, pH, oxygen, and bleaching agents (Markakis & Jurd 1974). Their use as a natural colorant has been accepted by the food industry most commonly in the form of fruit and vegetable juices and extracts as well as grape-skin extracts. These pigments may also contain the addition of aliphatic or aromatic acyl groups, resulting in enhanced pigment stability. Acylated sources such as black carrot juice have found commercial success as a colorant for this reason. Anthocyanins can interact with a wide range of other plant materials which contributes to the incredible color diversity of the pigment in comparison to the other mentioned plant pigments (Eugster & Märki‐Fischer 1991).

Betalains are indole (nitrogen-containing heterocyclic) compounds containing either an amine conjugation (betaxanthin) or cyclo-Dopa (betacyanin) (Sigurdson et al.

2017). They have a yellow or red color expression and are found in sources such as beets,

Figure 1. The four classes of plant pigments commonly found in dietary sources along with the source

10 cactuses, and even the stems of swiss chard (Khan 2016). These pigments are highly susceptible to heat but more stable at near neutral pH (5-7), a range where anthocyanins degrade rapidly and have little color expression (Cai et al. 2005).

2.2. The Anthocyanin

2.2.1. Structure

There are currently over 700 reported anthocyanin structures (Andersen &

Jordheim 2010), and likely many more yet to be discovered. This great variety in structure contributes to their diverse color and chemical properties, making them so fascinating to investigate. Anthocyanins are composed of an aglycone structure, 2- phenylbenzopyrylium salt, referred to as an (Schwartz et al. 2017). The aglycone structure is composed of a C6-C3-C6 backbone, A, B, and C ring. The A and C ring are fused, with a charged oxygen species present in the C ring, and a Carbon-Carbon bond connecting the B and C ring. From the B-ring, there are two sites that allow one aglycone structure to differ from another (3ʹ and 5ʹ). There are six primary aglycones found in food materials that differ by substitution with –H, -OH, and –OCH3, Figure 2.

These substitution patterns have influence on the color expression as well as stability and reactivity of the pigment (Schwartz et al. 2017).

As the substitutions on the aglycone structure become more electrophilic, the λmax increases more towards the blue region of the visible spectrum (Cabrita et al. 2000).

Inversely, greater electrophilic substitution on B-ring of the aglycone structure leads to higher rates of degradation in most food systems. For metal-anthocyanin complexes to 11 form, two or more hydroxyl substitutions must be present on the B ring ( and cyanidin) (Sigurdson et al. 2016). The electrophilicity also plays a role in copigmentation, likely a result of enhanced hydrogen bonding. The anthocyanidin is highly-unstable and the molecule is often modified within plant vacuoles through enzymatic activity and the phenylpropanoid pathway (Shi & Xie 2014).

Formation of anthocyanins (aglycone + sugar moiety) occurs following in vivo synthesis of the anthocyanidin by addition of a sugar moiety. This occurs primarily at C3, but can be found at other sites such as C4, C5, and C7 (Andersen & Jordheim 2010; Zhao et al. 2014). Mono-glycosylated anthocyanins are most prevalent; however, di- and tri- saccharidic anthocyanins are also naturally present (Harborne & Hall 1964). In addition to sugar moieties, acyl groups attached through ester bonds to the sugar are also another means of structural modification. Aliphatic and aromatic acids, predominantly hydroxycinnamic acids, add additional variation to the structural possibilities of anthocyanins, as well as modifying color expression and generally enhancing stability

(Zhao et al. 2017).

12 R1

OH 3ʹ 2ʹ 4ʹ B 1ʹ 5ʹ HO O 6ʹ 8 1 7 9 2 R2 A C 6 10 3 5 4 OH

OH

Figure 2. The basic chemical structure of anthocyanidins (aglycones) a λvis-max reported as anthocyanidin-3-glucoside in pH 1 aqueous solution by Cabrita et al (2000)

13 2.2.2. Equilibrium

Anthocyanins exist in several equilibrium states, dependent upon pH. These forms

are both colored and colorless. At low pH (3), the anthocyanin exists predominantly as a

flavylium cation, Figure 3. This is generally considered the most stable form and is

responsible for the red hues often seen in fruits and vegetables. Most foods are slightly to

moderately acidic; this presents challenges for the application of these pigments as a

colorant that has a pH-dependent response. The flavylium form is in equilibrium with the

hemiacetal forms (chalcone and carbinol pseudobase) which is colorless or slightly

yellow and is predominant between pHs 3 to 6. The equilibrium between flavylium and

colorless forms is often denoted as the pKa of hydration (pKh). B-ring substitution,

glycosyl moieties, as well acyl groups all modify these two equilibriums and generally

Figure 3. Equilibrium forms of anthocyanidins. Colorless forms (left) expressed at pH 3-6. The flavylium cation (center) most predominant at pH <3 and the quinonoidal bases (right) preferred at pH >6

14 increase the pKh (Cabrita et al., 2000; Giusti & Wrolstad, 2003; Stintzing, Stintzing,

Carle, Frei, & Wrolstad, 2002). At higher pH ranges (>6), quinonoidal bases are preferred. The flavylium form deprotonates and loses its positive charge. These forms exhibit both blue and purple hues but have extremely short half-lives.

2.2.3. Stability

Anthocyanin stability remains a challenge for their use as a food colorant. Many factors including heat, light, enzyme, pH, oxygen, and bleaching agents are all capable of increasing pigment degradation (Markakis & Jurd 1974). Of particular concern to the stability of juices and beverages is the presence of bleaching agents. Bleaching results in the rapid loss of color. These agents include peroxides, sulfites, and ascorbic acid – all common additives used in food processing with the sulfites and peroxides serving as antimicrobial agents and ascorbic acid serving as an antioxidant (Ozkan et al. 2002; Starr

& Francis 1968; Jurd 1964).

In addition to these environmental factors, the pigment structure itself is also important in modifying stability. Generally, acylated anthocyanins are regarded as more stable as their non-acylated counterparts (Giusti & Wrolstad 1996). This is likely due to these acyl groups folding above or below the chromophore, sterically hindering the anthocyanin’s more reactive sites. Anthocyanins which have been isolated from plant materials are even more susceptible to degradation (Giusti & Wrolstad 2005). The stabilizing effect of other compounds in the plant material is quite common and has been reported with quercetin, chlorogenic acid and catechins, and many additional compounds.

15 2.2.4. Potential Health Benefits

The relationship between anthocyanins in dietary consumption and potential health benefits have been studied extensively, yet major findings continue to be published. Anthocyanins have previously been thought to have very poor bioavailability with less than 2% absorption of the micronutrient and its conjugated derivatives (Kay et al. 2005; Manach 2004). More recently, the bioavailability has been reported as being significantly underestimated in the past. Ferrars et al. (2014) reported 12.4% recovery in urine and breath of a cyanidin-3-glucoside dose. Development of more comprehensive approaches of monitoring metabolites have likely played a role for this revised value.

Consumption of anthocyanins and anthocyanin-rich foods has been investigated in regards to cancer risk. Intake of this micronutrient has been reported as being capable of inhibiting certain tumor types in both animal models as well as cell cultures.

Anthocyanins have been reported to act by inhibition of tumor initiation and reducing development of already formed tumors such as breast, prostate, liver cancer (Li et al.

2017). Beyond anti-cancer properties, anthocyanins have exhibited a diverse range of biological effects including anti-inflammation, neuroprotection, cardiovascular disease prevention, anti-obesity, and antidiabetic activity (Li et al. 2017). These biological activities have likely been a driver for their choice as a natural colorant of foods.

2.3. Reactions of Anthocyanins

2.3.1. Bleaching

A major challenge specific to the use of anthocyanins as a food colorant is their interaction with bleaching agents. Bleaching of anthocyanins is the rapid loss of pigment 16 color with an increase in lightness and decrease in chroma. The prevalence of bleaching agents is particularly challenging to the juice and beverage industry, one of the largest markets for food colorants, which heavily relies upon fortification with ascorbic acid. It is thought to be caused by highly electrophilic species condenses with anthocyanins primarily at the nucleophilic C4 position and to a lesser extent at C2 of the anthocyanin structure, seen in Figure 4 with bisulfite addition (Poei-Langston & Wrolstad 1981;

Berké et al. 1998). Bisulfite bleaching has been investigated using nuclear magnetic resonance: the largest shifts in NMR data were found to correspond to primarily to C4 of the anthocyanin and conjugation loss of the flavylium ion (Berké et al. 1998). While it has generally been thought that the loss of color associated with bleaching is a result of an opened C-ring, similar to the chalcone form, these same authors propose that the C- ring remained closed and loss of aromaticity as driving the color loss.

Ascorbic acid bleaching is less understood in comparison to bisulfite bleaching.

Several authors have postulated that formation of hydrogen peroxide from ascorbic acid leads to the color loss exhibited versus direct condensation of the principle micronutrient

(Markakis 1982; Talcott et al. 2003). Further evidence that supports this is the comparative rate of bleaching; bisulfite bleaches anthocyanins within minutes, whereas equivalent levels of ascorbic acid take significantly longer to induce color loss.

17 The challenges of ascorbic acid bleaching are of particular interest because the

compound is well accepted by consumers because it is also Vitamin C, and for

manufacturers serves the dual-purpose of being a highly useful antioxidant for many food

systems (Varvara et al. 2016). Ascorbic acid has the capability of scavenging oxygen. It

has been repeatedly been reported that elevated levels of dissolved oxygen lead to higher

rates of degradation for ascorbic acid (Yuan & Chen 1998; Eison-Perchonok & Downes

1982; Khan & Martell 1967). Due to this reason, juice and beverage manufacturers often

add in significantly higher levels than labeled to guarantee the amount by the expiration

date, further contributing to the deleterious loss of anthocyanin color.

OH

OH

OH SO3Na HO

HO O OH OH O OH OH OH OH

HO O OH NaHSO + 3 HO OH O OH O OH OH HO O

OH HO O OH O OH OH SO3Na OH

Figure 4. Proposed bleaching mechanism of anthocyanins in response to bisulfite. Condensation occurring primarily at Carbon-4 (bottom arrow) and to a lesser extent at Carbon-2 (upper arrow)

Polymeric anthocyanins have been found to be less susceptible to bleaching,

likely a result of the C4-C8 bridge that often connects these pigments. In fact, addition of

sulfites to anthocyanins is a common way to evaluate the polymeric anthocyanin content 18 of a material (Giusti & Wrolstad, 2001). Pigments containing both a C3 and C5 substitution have also been found more resistant to bleaching reactions, likely a result of greater steric hindrance around the C4 site (Garcia-Viguera & Bridle 1999). In addition, pyranoanthocyanins which possess covalent occupation of C4 have been found to reduce color loss from bisulfite addition as compared to their anthocyanin counterparts (He et al.

2010; Oliveira et al. 2006; Gómez-Alonso et al. 2012).

2.3.2. Copigmentation

There is incredible diversity in the range of color expressions and hues that anthocyanins are capable of producing. One of the many factors responsible for this is copigmentation. This interaction is a unique phenomenon where a colorless, or slightly- colored (concentration dependent), compound enhances the color produced by the anthocyanin. The color enhancement is denoted by a hyperchromic shift, and sometimes a bathochromic shift. This is thought to be the result of enhanced π- π interaction as well as hydrophobic interactions between pigment and copigment (Boulton 2001; Trouillas et al. 2016). This complex interaction and form of color modification was first observed by

German Chemist R. Willstätter (1913), but has yet to be fully understood. Researchers were shocked to find the same cyanin pigment in both the blue cornflower as well as the rose (Goto & Kondo 1991). From the many hues that wine possesses to the great diversity in rose color, copigmentation is an important factor contributing to the wide swathe of colors from anthocyanins (Boulton 2001; Eugster & Märki‐Fischer 1991) .

Copigments, also referred to as cofactors, include phenolic acids, flavan-3-ols, amino acids, and additional classes of compounds (Boulton 2001). These cofactors are

19 present alongside the plant material and are often derived from the same phenylpropanoid pathway responsible for anthocyanin production. These groups can either free in solution, known as inter-molecular copigmentation, or covalently attached, called intra-molecular copigmentation. Figure 5 shows the difference between the two forms of copigmentation.

Figure 5. Intermolecular copigmentation showing planar stacking with gallic acid (left) and intramolecular copigmentation showing folding over chromophore with coumaric acid acyl group (right)

Intramolecular copigmentation is more common, but not limited to, vegetable sources of anthocyanins. Generally, intramolecular copigmentation is induced by acyl groups that are covalently linked from the glycosidic substitution extending off C3 or C5 of the pigment (Zhao et al. 2017). This results in pigments with greater enhancement of color expression, as seen with higher molar absorptivities, and stability; This is likely the result of the covalent linkage inducing greater steric hindrance and π-π interaction with the chromophore. Aromatic acyl groups such as hydroxycinnamic acids tend to enhance folding above or below the chromophore more so than aliphatic acyl groups like succinic and acetyl acyl groups (Zhao et al. 2017; Dangles, Saito & Brouillard 1993).

20 Intermolecular copigmentation is the result of cofactors, being free in solution, interacting with the anthocyanins, resulting in a similar delocalization of electrons in the chromophore and enhanced color expression. While intermolecular copigmentation is typically seen with organic acid cofactors, intermolecular copigmentation can be observed with a wide range of plant molecules including flavan-3-ols, caffeine, amino acids, and organic acids (Boulton 2001). Intramolecular copigmentation typically has an elevated response in color enhancement as compared to intermolecular copigmentation, likely the result of continuous movement of the anthocyanin and cofactor in solution whereas intra- is covalently bound (Sigurdson et al. 2017). Intermolecular copigmentation color enhancement is a dose-dependent interaction meaning elevated levels of cofactor can result in even greater color enhancement, revealing a linear relationship (Zhang et al. 2009).

There are several factors that determine the influence of a copigment in intermolecular copigmentation. The initial anthocyanin concentration, copigment concentration, pH, aglycone, temperature, and substitution pattern all play a role into the extent of copigmentation observed. Copigmentation reactions are thought to require a pigment concentration of 35 μM or above in order to be observed (Jurd 1967). As temperature is elevated, copigmentation effects are lost (Brouillard et al. 1989).

Copigmentation-induced color enhancement is greatly dependent upon pH.

Greater color enhancement is seen at mildly acidic pH ranges, often around and above the pKh between the colored flavylium and colorless form (Li et al. 2017). In fact, many studies have observed how the presence of intra and intermolecular copigmentation are

21 capable of shifting the pKh upwards (Dangles, Saito & Brouillard 1993; Davies et al.

1993; Lambert et al. 2011). At pH levels lower than an anthocyanin molecule’s pKh, the copigmentation effect is often lost. This might provide a basis as cofactors contributing to enhanced color by possibly modifying the pKh of the anthocyanin structure.

Catechin Copigmentation

Catechins are a common copigmentation agent of anthocyanins and are even building blocks in proanthocyanidins. They are naturally present in teas, wines, and chocolates. Catechins are a class of that are structurally similar to anthocyanins with the major difference being the loss of aromaticity in the C ring. This similarity likely extends to the ability of the two free molecules to stack in solution and induce intermolecular copigmentation. Catechins have two chiral centers (C2 and C3), leading to the epi- and gallo- prefixes in their names. They can contain substitutions such as gallic acid acyl bond off C3, similar to anthocyanins with sugar moieties. The most predominant catechin is green tea is epigallocatechingallate (EGCG). Catechins, overall, make up to 30% of the dry weight of green tea leaves (Graham 1992).

It’s ability to function as a copigment has been previously described. It is more capable of inducing color enhancement (hyperchromic shift) more so than a bathochromic effect (higher λmax). Chen and Hrazdina (1981) have reported catechins capable of enhancing color by 35%, as compared to myricetin (172%) and quercetin

(149%). Cai and others (1990) have reported similar results with a 44% increase for epigallocatechingallate. Surprisingly, the effect of diastereomers also greatly alters the extent of copigmentation. Brouillard and others (1991) have reported that catechin can 22 enhance color by 71%, whereas epicatechin was only 8%. The free rotation and torsion of the bond between the C and B ring as well as the acylation off C3 are thought to lower the extent to which catechins can function as a copigment, especially compared to other, more planar flavonoids like myricetin and quercetin.

2.3.3. Pyranoanthocyanins

The word pyran means a six-membered, non-aromatic ring containing an oxygen.

For a pyranoanthocyanin, it results in the addition of a fourth ring by heterocyclic addition at C4 and C5 of the anthocyanin. These pigments were first observed in red wines, but have since been reported in low levels in red and (Somers

1971; Andersen et al. 2004; Fossen & Andersen 2003). In addition, they have been observed in fruit juices, thought to the result of pasteurization in the presence of hydroxycinnamic acids (Rein et al. 2005). The cofactors capable of yielding a pyranoanthocyanin must contain a polar-enolizable group which allows for the SN2 heterocyclic addition to occur. These include small yeast metabolites (pyruvic acid, acetaldehyde), flavanols, and hydroxycinnamic acids. In addition, a combination of these can occur in polymeric pyranoanthocyanins where vinyl bridging can lead to significantly larger, more complex structures. The addition of the fourth ring significantly alters the stability and color characteristics of this class of derived pigments.

Pyranoanthocyanins have several unique color characteristics as compared to their parent compound. They are typically hypsochromically shifted and contain hues that have orange-like color expression. Anthocyanins are subject to extreme color changes, pH dependent, and can range from red (low pH), to colorless (mildly acidic, low pH), to even

23 purples, blues and greens (alkaline pH). This is caused by the shifts in equilibrium,

Figure 3. Pyranoanthocyanins exhibit a unique equilibrium that has yet to be fully understood. NMR analysis of the equilibrium has been investigated by de Freitas and others (2009) and they have reported that the Vitisin B pyranoanthocyanin (pyrano- -3-glucoside) to have an equilibrium form that has no hemiacetal state. Instead, proton transfer is thought to directly occur between the flavylium and quinoidal form.

This has the impact that the pigment is less susceptible to color loss associated with anthocyanins in the hemiacetal form. The pyranoanthocyanin is less susceptible to discoloration as seen with anthocyanins. Many foods are mildly acidic, meaning the pyranoanthocyanin could be an advantageous colorant for systems where anthocyanins have a lower tinctorial strength.

In terms of stability, pyranoanthocyanins have been reported as being significantly or completely more resistant to hydration as compared to anthocyanins

(Oliveira et al. 2009). The covalent linkage at C4 blocks the preferred site of nucleophilic attack of the pigment. This also has implications for stability against bleaching agents.

Several studies have specifically investigated pyranoanthocyanin resistant to bleaching induced by sulfites. Sulfites have likely been selected for investigation because of their use in the wine industry as an antibacterial agent. As mentioned in section 1.2, anthocyanins are highly susceptible to bleaching; whereas, pyranoanthocyanins have stability against this reaction.

The addition of a fourth ring blocks the C4 thought to be the site of condensation observed between anthocyanin and electrophilic bleaching agent. Pyranoanthocyanins

24 have generally been observed as being more resistant to this, but several specific observations have been noted. Carboxypyranoanthocyanins have been reported as experiencing significantly less bleaching caused by sulfite addition (He et al. 2010;

Oliveira et al. 2006). These authors report that carboxypyranoanthocyanin derivatives of malvidin-3-glucoside and cyanidin-3-glucoside have a minimum of a 3x fold enhancement in resistance to bleaching. Mateus and others (2010) also report that oligomeric pyranoanthocyanins, containing pyrano-malvidin-3-glucoside and a vinyl bridge to a catechin or epicatechin unit, were completely resistant to bleaching up to 250 ppm sulfite. It is possible that intramolecular interactions and the folding of catechin above or below the chromophore could be aiding in blocking C2 of anthocyanins, going from significantly more resistant to completely as compared to anthocyanins.

2.3.4. Self-Association

Anthocyanin self-association is similar to intermolecular copigmentation reactions but differs in that the cofactor is another anthocyanin. It was first discovered by

Stewart and Norris (Asen et al. 1972) when increasing concentrations were found to deviate from Beer-Lambert’s law, resulting in higher than expected absorbance values.

This interaction results in stabilization of the molecule through planar stacking and enhanced hydrophobic interaction (Cavalcanti et al. 2011). The addition of sodium chloride has been found to promote self-association of anthocyanins (Hoshino et al.

1980): it is likely that the metal ion is forming complexes with the anthocyanin B-ring’s of several anthocyanin molecules through metal-anthocyanin complexation, thus promoting the vertical stacking of pigments.

25 2.3.5. Polymerization

Many anthocyanin sources contain oligomeric and polymeric anthocyanins, commonly called proanthocyanidins. These derived compounds can be found in many sources including fruits, fruit juices, legumes, seeds, and wines (Santos-Buelga &

Scalbert 2000). These compounds are the result of condensation between C4-C8 and C6-

C8 (C-C preferentially and C-O) of anthocyanins with additional anthocyanins or other flavan-3-ols. Catechins have been repeatedly mentioned as polymerizing with anthocyanins in sources such as , cranberries, and chokeberries (Yokota et al.

2013; Gu et al. 2002; Kulling & Rawel 2008). These compounds are broadly categorized by the degree of polymerization, the connecting sites, and the monomeric components

(Yokota et al. 2013; Santos-Buelga & Scalbert 2000). Their analysis remains challenging due to the structural diversity: for this reason it is thought that proanthocyanidin consumption in diet is underestimated (Santos-Buelga & Scalbert 2000). This class of compounds has been reported as being more stable than their monomeric counterpart, anthocyanins, as well as still expressing dark-red, purple hues (Gu et al. 2002; Rodríguez-

Pérez et al. 2015).

2.3.6. Metal Complexation

Anthocyanins may interact with certain metals to form blue coloration at mildly acidic pH ranges. In order for this reaction to occur, more than one free hydroxyl group must be present on the B-ring, meaning the reaction takes place with only cyanidin and delphinidin aglycone structures. The effect of metals is two-fold. One, the multivalent metal ion competes for hydrogen ions with the flavylium cation, resulting in

26 deprotonation of the pigment to the quinonoidal base. Two, the complexation of a metal ion and anthocyanin stoichiometrically favors multiple anthocyanin molecules per metal ion, causing enhanced stacking and self-association of the pigment, inducing a hyperchromic effect. This phenomena requires a multivalent metal ion. Trivalent ions such as Cr3+, Al3+, Fe3+, Ga3+ result in greater bathochromic and hyperchromic shifts at pH 3 to 8 (Sigurdson et al. 2016). Certain bivalent metals such as lead and tin have also produced limited responses in metal-anthocyanin complexation.

2.4. Ascorbic Acid in Foods

Ascorbic acid addition in foods is widespread and usually owed to its functionality as both an antioxidant and stabilizing agent. It is a water-soluble vitamin, meaning frequent consumption through diet is required. Severe vitamin C deficiency

(scurvy) is rather rare due its fortification in many foods, but the World Health

Organization estimates mild and marginal deficiencies to be significantly higher (Allen et al. 2006). Ascorbic acid is essential due to its function in collagen and hormone production: Therefore is has been viewed as functional food additive by the food industry

Figure 6. Primary equilibrium of vitamin C with the redox reaction between the vitamers ascorbic and dehydroascorbic acid and widely accepted by consumers (Varvara et al. 2016).

27 In solution, vitamin C primarily exists in two vitamers forms, ascorbic (reduced form) and dehydroascorbic acid (oxidized form), Figure 6. Ascorbic acid is generally considered the more stable vitamers and degradation is thought to occur primarily with reduced form. Vitamin C is highly reactive; it scavenges oxygen with conversion to water, reduces certain metal ions at levels as low as 1 μM and oxidizes, and readily reduces radicals (Tsao et al. 2009; Jansson et al. 2004; Nursten 2005; Varvara et al.

2016).

Ascorbic acid is generally recognizes as safe the Food and Drug administration and is commonly added to fruit and vegetable products to protect freshness and color.

This is accomplished though inhibition of polyphenol-oxidase. Ascorbic acid can prevent the formation of o-quinones by poly-phenoloxidase, thus preventing the formation of the substrate required for polymerization and brown-colored melanoidin formation (Amiot et al. 1992). The antioxidant capacity of ascorbic acid is also utilized in other food systems.

Salts of ascorbic acid are required in cured meats such as bacon to effectively reduce the formation of the carcinogen nitrosamine (Epley et al. 1992). The esterified form of ascorbic acid, ascorbyl palmitate is an approved additive that is capable of reducing auto- oxidation of oils and has synergistic capacity in combination with tocopherols (vitamin

E) (Yanishlieva & Marinova 2001). Ascorbic acid is a versatile molecule that is heavily relied on in food preservation with ubiquitous use in many food systems. For this reason, its stabilization in response to anthocyanins warrants thorough investigation.

28

Chapter 3. The role of green tea catechins in modifying the bleaching of anthocyanins by ascorbic acid 3.1. Abstract

Bleaching of anthocyanins by ascorbic acid is a major hurdle in the shift to natural colorants for the juice and beverage industry. Ascorbic acid if a common fortifying agent in foods due to its nutritive and antioxidant roles. Green tea catechins have been previously found to enhance anthocyanin color through copigmentation. The hypothesis was that addition of green tea catechins could reduce the interaction of anthocyanin and ascorbic acid and prevent fading of color. Blackberry, black carrot, and red cabbage were semi-purified and diluted to 50 M in pH 3 citrate buffer. Green tea catechins (1.25 mM gallic acid equivalency) and ascorbic acid (250 mg/L) were added to the pigment solutions. Samples were stored for 12 days in the dark at 25 C. HPLC was utilized to quantify changes of anthocyanin and ascorbic acid concentration, and absorbance spectra

(380-700 nm) of solutions were collected and converted to color parameters (CIE

Lightness and chroma). The addition of catechins was found to initially enhance absorbance for all three sources (6.17-0.21%). Chroma increased and lightness decreased.

All pigment sources degraded rapidly in the presence of ascorbic acid with t1/2 reduced by

90-97% versus the control with red cabbage being least susceptible to ascorbic acid addition. The inclusion of green tea catechins did not mitigate or lower the effect of

29 ascorbic acid on loss of anthocyanin or color changes. An increase in Lightness (8.4-

15.3) and decrease in Chroma (20.5 to 28.3) were observed for all treatments containing ascorbic acid. Ascorbic acid was most stable as a control and addition of anthocyanins or catechins accelerated loss of the vitamin (8 vs 4-6% remaining at day 12). The interaction of green tea catechins and anthocyanins through copigmentation was assumed too weak to reduce the impact of ascorbic acid bleaching and pigment degradation.

3.2. Introduction

Anthocyanins are pigments in fruit and vegetable sources that are capable of producing a wide array of colors. Interest in their use as a food colorant has been driven in recent years by several trends including concern over the use of synthetic dyes associated with hyperallergenicity and more consumer-friendly labeling (Arnold et al.

2012; Martins et al. 2016). Their stability in food applications is limited by a multitude of factors including degradation in response to light, heat, enzymes, pH, and bleaching agents (Castañeda-Ovando et al. 2009). The latter is of special concern for the juice and beverage industry which relies heavily on the use of both fruit and vegetable juices and extracts as a colorant as well as ascorbic acid, a bleaching agent, to stabilize drink applications.

Bleaching of anthocyanins results in rapid color loss that yields unattractive colors and lowered product quality. It is thought to occur as a result of condensation between an electrophilic species (sulfites, ascorbic acid, hydrogen peroxide) and the nucleophilic Carbon-4 (C4) of anthocyanins, resulting in loss of aromaticity of the C-ring 30 as well as possible opening (Poei-Langston & Wrolstad 1981; Ozkan et al. 2002).

Bisulfite bleaching has been most extensively studied as a bleaching agent as it is a common preservative in the food industry and wines. Nuclear mangnetic resonance investigation has elucidated the structure of byproducts between anthocyanins and bisulfites showing the two sites where electrophilic attack is most likely as C4 and to a lesser extent C2 (Berké et al. 1998). In instances where C4 is inaccessible such as pyranoanthocyanins, C2 is thought to be a possible reaction site for bleaching (Oliveira et al. 2006). Several studies have investigated how structural modification of the anthocyanin can modify reactivity with bleaching agents. Substitution on C4 of flavylium salts with both phenyl and methyl group instead of typical hydrogen attachment have been reported as being more resistant to bisulfite bleaching. The same authors observed malvidin-3,5-diglucoside as having superior resistance to bleaching over malvidin-3- glucoside, likely a result of steric hindrance reducing the ability of the electrophile to access C4 (Garcia-Viguera & Bridle 1999). Recent work has led to the possibility of using copigmentation as a means to reduce bleaching.

Copigmentation is a phenomena where a colorless compound enhances the color expression of an anthocyanin. Enhanced electron delocalization from pi-pi interaction is thought to result in a greater molar absorptivity (Trouillas et al. 2016). Compounds such as flavanoids, aromatic acids, and amino acids have all been reported as having copigmentation effects (Boulton 2001). Intramolecular copigmentation is where the colorless compound is covalently attached and intermolecular copigmentation involves a cofactor that is free in solution interacting to enhance color (Giusti & Wallace 2013).

31 Previous research has found catechins capable of enhancing color with a hyperchromic shift and color enhancement up to 35% (Brouillard et al. 1991; Asen et al. 1972; Dangles,

Saito & Brouillard 1993). Catechins are found in polyphenol-rich sources such as tea, chocolate, and (Manach 2004).

Recent evidence has also been published that establishes the interaction between the bleaching agents ascorbic acid and sulfites and catechins. Similar to anthocyanins, catechins have an equilibrium and a quinone form. Bisulfites have been found to greatly enhance the stability of catechins in model wines (Danilewicz & Wallbridge 2010).

Ascorbic acid has also been reported as having a stabilizing effect on catechins from green tea. The authors suggest that ascorbic acid is critical in reducing oxygen levels that aid in the formation of quinone catechin forms (Z.-Y. Chen et al. 1998). Instead of experiencing color loss caused by interaction between ascorbic acid and anthocyanin, catechin copigmentation could serve as a means to subsidize color expression lost while also reducing interaction between the two subtrates. The objective of this study was to observe if anthocyanin reactivity and color expression could be favorably enhanced by addition of catechins from green tea. This system was thought to serve as a potential means of aiding the food industry in transitioning to naturally-colored juices and beverages subject to ascorbic acid-induced color loss.

32 3.3. Materials and Methods

3.3.1. Materials

Anthocyanins were obtained from the three following sources: blackberry (Rubus sp.), red cabbage (Brassica oleracea), and black carrot (Daucus carota subsp. Sativus).

Blackberries were obtained from a local grocer (Columbus, OH, USA) and red cabbage and black carrot were obtained as commercial powders from Mars, Inc. (McLean, VA,

USA). Green tea catechins were obtained from loose leaf green tea also purchased at a local grocer. Analytical grade L-ascorbic acid (>99.99%) and formic acid (>98%) were obtained from Sigma-Aldrich (St. Louis, MO, USA). Acetic acid (99.5%) was purchased from Acros Organics (Fisher Scientific, New Jersey, USA). Analytical grade gallic acid was purchased from MP Biomedicals (Santa Ana, CA, USA). HPLC grade water and methanol were purchased from Fisher Scientific (New Jersey, USA).

3.3.2. Methods

Anthocyanin and Green Tea Catechin extraction and semi-purification

Blackberry, black carrot, and red cabbage were subjected to semi-purification prior to use in the model system. The fresh blackberry sample was frozen with liquid nitrogen and blended into a homogenous powder. It was then solubilized in 70% acetone:

30% acidified water (0.01% v/v HCl). The resulting slurry was filtered with Whatmann

#4 paper under vaccuum and the filtrate was mixed with twice the volume of chloroform.

The solution was gently mixed in a separatory funnel and refrigerated overnight. The following morning the aqueous phase was collected and the acetone was removed on a

33 Buchi Rotovap at 35 C under vaccuum. The remaining solution (0.01% v/v HCl in water) was stored until C18 semi-purification.

The frozen powders of black carrot and red cabbage were hydrated with acidified water (0.01% v/v HCl). The three pigments in aquous solution were then subjected to

C18 semi-purification. The solutions were loaded on an activated Waters Sep-pak C18 cartridge (Milford, MA, USA). With the pigment bound, two volumes of acidified water

(0.01% v/v HCl) were passed through the column for removal of polar components. The pigments were then recovered in acidified methanol (0.01% v/v HCl). The solvent was then removed with the Buchi Rotovap at 35 C under vaccuum. These procedures are described in detail by Rodriguez-Saona and Wrolstad (2001).The dried pigment was resolubilized in acidified water (0.01% v/v HCl) and frozen until further use. These methods were adopted from Rodríguez-Saona et al. (2001).

The methods for extraction of catechins from green tea were adopted from Roach et al. (2011). Fifty grams of loose-leaf green tea was mixed with half a liter of reverse osmosis water and was held at 80 C for 30 minutes. The tea leaves were removed and the solution was filtered through Whatmann #4 paper under vaccuum. The filtrate was then subjected to the same C18 semi-purification procudure that was used for the anthocyanins. Catechins, being structurally similar were bound and removed from the column in a similar fashion. The catechins were dried on a Buchi Rotovap at 35 C under vaccuum, resolubilized in acidified water (0.01% v/v HCl), and frozen until further use.

34 Anthocyanin quantification

Changes in anthocyanin concentration were monitored through the use of ultra- high performance liquid chromatography. A standard curve with cyanidin-3-glucoside

(Sigma Aldrich, St. Louis, Missouri, U.S.A.) was constructed with the AUC at 520 nm of the anthocyanin. A 2040C Nexera-i uHPLC instrument (Shimadzu, Maryland, U.S.A.), with a binary gradient, was used for both the standard curve and samples. A Pinnacle DB

(Restek Corporation, Bellefonte, PA, U.S.A.) C18 Column (1.9 µm particle size, 50 x 2.1 mm length) with a 0.4 mL/min flow rate was used with the column oven set at 40 C.

Solvent A was 4.5% formic acid and solvent B was 100% acetonitrile. The ramp included holding solvent B at 0% for the first minute followed with a ramp to 20% by 10 minutes.

Anthocyanin quantification of samples was recorded at day 0, 1, 3, 5 and 12.

Sample preparation

The semi-purified blackberry, black carrot, and red cabbage extracts were diluted to 55.6 uM in 0.1 M citrate buffer at pH 3 using the Total Monomeric Anthocyanin assay

(Giusti & Wrolstad 2005). The three sources were quantified in terms of cyanidin-3- glucoside equivalency, the same pigment also used for the HPLC standard curve. A 1.8 mL aliquot of the diluted pigment was added to each HPLC vial. Citric acid is prevelant in fruit juices and was selected to better model that matrix. pH 3 is also a relevant pH for many juices and beverages as well. A block design was used with the following treatments for each pigment source: ACN, ACN+AA, ACN+GTC, and ACN+AA+GTC.

This allowed for observation of the effect of green tea catechins and ascorbic acid as well as their interaction with one another in the presence of anthocyanins. In addition, AA and 35 AA+GTC treatments were prepared without anthocyanins. AA stock solution (5 g/L) was prepared and 100 uL aliquots were added to AA treatments to achieve a final concentration of 250 mg/L, mimicking 200% the Daily Value of vitamin C. Catechins were added at a 1:25 molar ratio (ACN:GTC) with levels from the FC test. Samples were all brought to the same final volume of 2 mL with additional pH 3 citrate buffer, resulting in a final concentration of 50 M ACN, 250 mg/L AA, and 1.25 mM GTC. Prior to addition to the HPLC vials, all solutions were filtered with a Phenex RC 0.45 µm, 15 mm membrane syringe filter (Phenomenex, CA, U.S.A.). All treatments were formulated in triplicate.

Ascorbic acid quantification

Total ascorbic acid was monitored through reverse phase HPLC coupled with a

PDA detector. The instrument included the following modules: LC-20AD pumps, CBM-

20A communication module, SIL-20A HT autosampler, CTO-20A column oven, and

SPD-M20A Photodiode Array detector (Shimadzu, MD, U.S.A.). Separation of ascorbic acid was achieved using a 0.5 mL flow rate on a Varian Pursuit C18 column (150 x 4.6 mm)(Palo Alto, CA, U.S.A). An isocratic solvent system consisting of 5% MeOH and

95% acidified water (0.1% acetic acid v/v) was used with methods adapted from Sawant et al. (2010). The vitamin had a retention time of three minutes. To account for ascorbic acid equilibrium forms, the standard as well as samples were reduced to ascorbic acid prior to injection. Total ascorbic acid includes both ascorbic and dehydroascorbic acid.

Ascorbic acid contains a chromophore and has a λmax at ~240 nm. Dehydroascorbic acid

(DHA) is the oxidized equililbrium form of ascorbic acid and does not have an 36 observable chromophore in the UV or Visible absorbance ranges, so reducing it to AA allowed for quantification of the vitamer. A standard curve was constructed using analytical grade AA which was mixed in 3% meta-phosphoric acid (MPA) solution adjusted to pH 1.9. Meta-phosphoric acid has been reported as stabilizing ascorbic acid in solution (Chebrolu et al. 2012). One mL of 1 g/L AA was then mixed with 1 mL of a 100 mM tris-2-carboxyethylphosphine (TCEP) in 3% meta-phosphoric acid solution, a reducing agent capable of converting dehydroascorbic acid to ascorbic acid, thus allowing for quantification of both vitamers. For total ascorbic acid quantification of samples, a 50 uL aliquot was removed from sealed HPLC vials using an SGE syringe

(Restek Corporation, Bellefonte, PA, U.S.A.). This was then mixed with with 450 uL of a

11.1 mM TCEP solution containing 3%. Solutions were allowed to equilibrate for fifteen minutes prior to injection on the HPLC system. Ascorbic acid was quantified on day 0, 1,

3, 5, and 10.

Catechin quantification

Quantification of phenolics in green tea catechins was performed using the Folin-

Ciocalteau (FC) method (Andrew L. 2001). A standard curve of gallic acid (50 to 500 mg/L) was constructed. A 20 uL aliquot of the green tea catechins and gallic acid standards were mixed with 1.58 mL reverse osmosis water, followed with 100 uL of FC reagent. This was stored for 5 minutes in the dark and then 300 uL of a saturated sodium carbonate solution was mixed in. The samples were stored for two hours in the dark and absorbance at 765 nm was collected on SpectraMax 190 Microplate Reader (Molecular

Devices, Sunnyvale, CA, U.S.A.) 37 pH Monitoring

The pH of all samples were evaluated using a S220 SevenCompact pH meter

(Mettler Toledo, Columbus, OH, U.S.A.). The pH was recorded on days 0 and 12 and were found to be within 0.1 of the 3.0 target at both the first and last time point.

UV-vis spectrophotometry and colorimetry of samples

Spectra was evaluated to monitor changes in λmax, maximum absorbance, and color. A 200 uL aliquot was removed from the sealed vials with the same syringe mentioned in ascorbic acid quantification. Spectra data was collected from 380 to 780 nm in 1 nm intervals. The aliquot was placed in a 96 polylysine well plate and data was collected with a plate reader. With the spectral data, conversion to color values was achieved using ColorBySpectra (Farr & Giusti 2017). These color calculations were made with the following settings: regular transmission, D65 illuminant, and 10° observer angle. Color values are reported in the CIEL*c*h* scale (CIE 2004).

Statistical analysis of data

Several statistical tests were performed to analyze the significance of data. Means and standard deviations were organized using Microsoft Excel (Redmond, Washington,

U.S.A). One-Way ANOVA was performed within block on the rate of anthocyanin degradation between treatments (ACN, ACN+AA, ACN+GTC, ACN+AA+GTC). T-tests were used to find the significance in initial enhancement of maximum absorbance induced by the addition of green tea catechins (ACN versus ACN+GTC). One-way

ANOVA was also performed to analyze the differences in the rate of ascorbic acid

38 degradation for all samples containing the nutrient. ANOVA and T-tests were performed using SPSS (IBM, North Castle, NY, U.S.A.).

3.4. Results & Discussion

3.4.1. Initial Color Enhancement and Spectra Changes

Based on previous research, it was expected that green tea catechins would be capable of enhancing anthocyanin color through copigmentation (Hsiao 2014; Lambert et al. 2011; Boulton 2001). Green tea catechin copigmentation enhanced maximum absorbance for all three sources. The effect of copigmentation was greatest for blackberry with 6.17% enhancement in maximum absorbance as well as a 3 nm bathochromic shift

(512 to 515 nm), Figure 7. This color enhancement was statistically significant (p-value

<0.001); whereas, the black carrot color enhancement of 0.21% and red cabbage of

3.28% were not statistically significant. Black carrot with green tea catechins had a bathochromic shift of 1 nm and red cabbage max remained unchanged at 529 nm.

Blackberry likely was most suspectible to copigmentation as a result of it’s relatively simple anthocyanin profile. The primary anthocyanin in blackberry, cyanidin-3- glucoside, likely presents minimal steric hindrance to the chromophore, allowing greater

- interaction between anthocyanin and catechin. Black carrot anthocyanins contain primarily di- and tri-glycosylated pigments that are primarily acylated with hydroxycinnamic acids. Red cabbage anthocyanins, predominantly having both 3- and 5- substitutions, also likely played a role in reduced copigmentation effect. These spectral shifts were also reflected in color data as well. 39

0.7 1.4 0.6 Blackberry Black Carrot Red Cabbage 0.6 1.2 0.5 ACN+GTC 0.5 1

ACN 0.4

e

e

e

c

c

c

n n

0.4 0.8 n

a

a

a

b

b

b

r r

r 0.3

o

o

o

s

s

s

b b

0.3 0.6 b

A A A 0.2 0.2 0.4

0.1 0.2 0.1

0 0 0 380 480 580 680 380 480 580 680 380 480 580 680 Wavelength (nm) Wavelength (nm) Wavelength (nm) Anthocyanin Anthocyanin + Green Tea Catechin Pigment ΔE L* c* h* L* c* h* Blackberry 82.6(0.1) 37.0(0.2) 14.1(0.0) 79.9(0.1) 39.7(0.2) 11.3(0.1) 2.4 Black Carrot 65.7(0.0) 60.7(0.1) 10.1(0.1) 64.3(0.5) 60.8(0.0) 8.9(0.6) 1.3 Red Cabbage 81.0(0.0) 39.9(0.1) 346(0.0) 80.6(0.9) 39.6(0.3) 347(0.1) 1

Figure 7. Initial changes in spectra and color characteristics of blackberry, black carrot, and red cabbage anthocyanins in response to green tea catechins.

Copigmentation coincides with an increase in chroma and decrease in Lightness,

Figure 7. For chroma, the following changes were observed between the anthocyanin

(ACN) and the anthocyanin with catechin (ACN+GTC) treatments: blackberry, 37.0

→39.7 (2.7); black carrot, 60.7 → 60.8 (0.1); red cabbage, 39.9 → 39.6 (-0.3).

Because blackberry experienced, the greatest color enhancement, it was also expected it

would have the greatest increase in Chroma. Black carrot and red cabbage, having little

color enhancement, also saw a negligible shift in chroma from catechin addition. For

change in lightness, the following was observed between ACN and ACN+GTC

treatments: blackberry, 82.6 → 79.9 (-2.7); black carrot, 65.7 → 64.3 (-1.4); and red

cabbage, 81.0 → 79.9 (-1.05). The changes in lightness again fell in line with the

observed color enhancement observations.

40 3.4.2. Copigmentation effect on anthocyanin stability

Copigmentation has been reported by several authors as having capacity to stabilize anthocyanins for longer periods of time (Boulton 2001; Trouillas et al. 2016;

Rein 2005). It was hypothesized that addition of green tea catechins to the three pigment sources evaluated would have potential in stabilizing the anthocyanins. The hypothesis was that - and hydrophobic interaction could reduce the accessibility between anthocyanins and ascorbic acid. Due to the nature of the copigmentation reaction enhancing color, HPLC quantification of anthocyanins was chosen as the method of monitoring the pigment concentration over time.

The addition of green tea catechins appeared to not have the desired impact of enhanced stability. Green tea catechins did not stabilize anthocyanins alone or in combination with ascorbic acid. Upon addition of green tea catechin, the half-life of blackberry decreased by 61.4%; black carrot, 63.4%; and red cabbage, 75.2% as compared to only anthocyanin, Table 1. The extent of pigment loss was unexpected as compared to other work; however, differences in assay choice may have played an important factor. Previous work regarding green tea catechin copigmentation within our lab has found more limited degradation of anthocyanins in response to catechin addition, but the total monomeric anthocyanin method was used. This assay presents challenges to evaluating copigmentation reactions because copigmentation has been reported as modifying the pKh (pK of hydration to colorless forms) of anthocyanins and enhancing color expression in mildly acidic pH; likely resulting in modified extent of hydration at pH 4.5 used in the assay. It is for this reason that HPLC quantification may present a

41 greater extent of pigment loss as compared to other forms of anthocyanin measurement.

Table 1. Kinetic Parameters of anthocyanins (ACN) in response to ascorbic acid (AA) and green tea catechin (GTC) addition and change in color (E) for blackberry, black carrot, and red cabbage anthocyanins. Reaction rate and half-life calculated as first-order kinetics from HPLC

k half-life E Treatment R2 (days-1) (days) (day 0 vs day 12) ACN 7.77E-03 89.1 0.97 1.1 ACN+AA 1.36E-01 5.1 0.99 11.1 Blackberry ACN+GTC 2.01E-02 34.4 1.00 3.1 ACN+AA+GTC 1.31E-01 5.3 0.99 22.8 ACN 1.95E-03 355.2 0.70 0.7 ACN+AA 6.82E-02 10.2 0.77 12.6 Black Carrot ACN+GTC 5.34E-03 129.8 0.81 1.8 ACN+AA+GTC 6.06E-02 11.4 0.80 12.8 ACN 6.46E-03 107.3 0.62 2.3 ACN+AA 6.53E-02 10.6 0.93 10.8 Red Cabbage ACN+GTC 2.60E-02 26.6 0.97 3.9 ACN+AA+GTC 7.89E-02 8.8 0.95 17.7

As expected, the addition of ascorbic acid resulted in extensive degradation of anthocyanins amongst all souces; however, the impact varied between sources, Table 1.

Upon addition of ascorbic acid, blackberry saw a reduction of 94.3%; black carrot,

97.1%; red cabbage, 90.1% as compared to the anthocyanin only treatment. Black carrot was not expected to be most susceptible. Due to the simple anthocyanin structure in blackberry, greater than 90% cyanidin-3-glucoside, it was thought that it would be most susceptible to ascorbic acid-induced degradation as compared to its control. Possible reasons for the difference could be attributed to the stereochemistry of the acylated black carrot anthocyanin pigments. The heavily acylated profile leads to intramolecular copigmentation, the planar stacking of the acyl group parallel to the chromophore. This

42 ordered conformation may have increased accessibility to Carbon-4 more so than the mono-glycosylated blackberry anthocyanins. In addition, these materials likely consist of other phenols which differentiate the response to bleaching.

The enhanced resistance of ascorbic acid-induced degradation of red cabbage anthocyanins was expected. Red cabbage contains C3 and C3,C5 substituted anthocyanins. The C3,C5 substituted anthocyanins generally exhibit lowered stability compared to the C3-only anthocyanins, but previous research has exemplified their enhanced resistance to ascorbic acid attack (Garcia-Viguera & Bridle 1999). The suspected site of condensation, C4, being in between a glycosylated C3 and C5-site likely restricted accessibility and led to an improved half-life in response to ascorbic acid.

The ACN+AA+GTC treatment had a mixed outcome correlated to the original extent of copigmentation observed. The half-life with added ascorbic acid and catechin

(ACN+AA+GTC) compared to just ACN+AA was the following: blackberry, 5.3 versus

5.1 days; black carrot, 11.4 versus 10.2 days; and red cabbage, 8.8 versus 10.6 days. The addition of catechins to anthocyanins in the presence of ascorbic acid slightly improved the half-life of blackberry and black carrot, and reduced the half-life of red cabbage.

Catechin addition alone was not beneficial for pigment stability and had limited capacity in reducing anthocyanin-ascorbic acid reactivity.

43 The rapid loss in maximum absorbance, associated with the bleaching of the

pigment, can been seen in Figure 8. Pigments saw dramatic decreases in maximum

absorbance over the 12 day period in the presence of ascorbic acid. For the ACN+AA

treatment, blackberry maximum absorbance dropped from 0.56 to 0.11, black carrot from

1.20 to 0.53, and red cabbage decreased from 0.50 to 0.30. Comparing the changes in

spectra for ACN+AA and ACN+AA+GTC treatment shows how catechins accelerated

the loss of absorbance more so than hindering it. For all treatments, the ACN+AA+GTC

treatment had a lower maximum absorbance than the ACN+AA treatment. For

blackberry, this occurred by day 3, black carrot was day 1, and red cabbage was day 3.

0.7 1.4 0.6 Blackberry Black Carrot Red Cabbage 0.6 1.2 Day 0 ACN 0.5 ACN+GTC 0.5 Day 1 1

0.4 e

c Day 3

n 0.4 0.8

a b

r Day 12 0.3 o

s 0.3 0.6 b

A 0.2 0.2 0.4

0.1 0.2 0.1

0 0 0 400 450 500 550 600 400 450 500 550 600 400 450 500 550 600 Wavelength (nm) Figure 8. Spectral changes of blackberry, black carrot, and red cabbage anthocyanins with ascorbic acid and green tea catechins added over a 12 day period.

3.4.3. Change in ascorbic acid levels

Because anthocyanin and ascorbic acid are thought to condense, evaluating

changes in the nutrient level was thought to provide insight to the extent of the reaction.

Ascorbic acid degraded extensively by day 12 regardless of the presence of catechins or

pigments, yet differences were observed between treatments, Figure 9. Ascorbic acid

44 appeared most stable as a control (ascorbic acid only) with only 20% loss at day 3 versus

43% for AA+GTC and 50-59% loss for ACN+AA+GTC treatments. Chen et al. (1998) have previously reported on the ability of ascorbic acid in stabilizing green tea catechins.

They postulated that ascorbic acid reduces catechin’s tendency to form semiquinone radicals and epimerize; it is possible that ascorbic acid is lower in the presence of catechins because it is reducing catechin radicals and degrading in the process.

The addition of the pigments contributed to faster loss of ascorbic acid which was expected. It was thought that the loss of ascorbic acid would mirror results from anthocyanin kinetics but this was not observed. The blackberry and ascorbic acid combination had higher remaining ascorbic acid content for days 3, 5 and 12 than black carrot and red cabbage treatments even though it experienced greater loss of anthocyanins, Figure 9. The effect of both anthocyanin and green tea catechins together on ascorbic acid stability appeared additive in that having both present reduced ascorbic acid stability further. For example, by day 5 the AA+GTC combination treatment had

24% ascorbic acid remaining and the ACN+AA treatments had 21-24% remaining. The treatments with both pigments and green tea catechins had 16-18% ascorbic acid remaining at day 5.

45

100% Day 1 Day 3 Day 5 Day 12

80%

60%

40%

20%

Total Ascorbic Ascorbic Remaining Acid Total 0%

Figure 9. Ascorbic acid content of mixtures of anthocyanin extracts and green tea catechins over a 12 day period at 25 C

3.4.4. Change in chroma and Lightness

The loss in color quality of anthocyanin bleaching is a limiting factor in their adoption by the food industry. Bleaching reactions coincide with an increase in Lightness and decrease in Chroma. The change in CIEL*c*h* were monitored by conversion of spectra absorbance data. For all samples, ascorbic acid induced an increase in Lightness,

Figure 10. The increase in lightness for the ACN+AA treatment was the following: blackberry, 11.9; black carrot, 13.2; and red cabbage, 8.4. The red cabbage anthocyanins likely saw the smallest change in L* due to the pigment’s enhanced resistance to degradation caused by ascorbic acid. The addition of catechins, while initially decreasing

L*, had no appreciable impact on the color stability as compared to the control. The differences between ACN and ACN+GTC Lightness was no more than 1.1 *L by day 12.

The initial decrease in Lightness observed for the ACN+AA+GTC treatment had all but

46 reversed by day 12 for the three sources. For black carrot and blackberry, the ACN+AA

and ACN+AA+GTC treatment were less than 0.9 L* apart and for red cabbage, 3.9 at

day 12. While catechin addition was effective in lowering Lightness initially with all

three sources, it was not an effective means for maintainenance of the original color

characteristics.

100 Blackberry Black Carrot Red Cabbage

90

)

*

L

(

s s

e 80

n t

h ACN+AA+GTC

g i

L 70 ACN+AA ACN ACN+GTC 60 0 5 10 0 5 10 0 5 10 Time (days) Figure 10. Change in Lightness (L*) of blackberry, black carrot, and red cabbage in response to green tea catechin and ascorbic acid addition

Decreases in chroma (c*) coincided with anthocyanin bleaching reactions, Figure

11. The three pigments all saw a decrease for ascorbic acid treatments. The ACN+AA

treatment saw the following decreases in chroma: blackberry, -26.8; black carrot, -24.3;

red cabbage, -20.5. Red cabbage, again, showed less extensive changes attributed to the

enhanced pigment stability. The addition of catechins (ACN+GTC) treatment ultimately

resulted in lowered Chroma compared to the control treatment by day 12, lowering the

value by 0.8 to 3.6.

47

65 Blackberry ACN Red Cabbage 55 ACN+GTC

) ACN+AA

* 45 C

( ACN+AA+GTC

a 35

m

o r

h 25 C

15 Black Carrot 5 0 5 10 0 5 10 0 5 10 Time (days) Figure 11. Change in Chroma (c*) of blackberry, black carrot, and red cabbage in response to green tea catechins and ascorbic acid addition

3.5. Conclusions

Copigmentation is capable of enhancing color and stabilizing anthocyanins;

however, the addition of green tea catechins to blackberry, black carrot, and red cabbage

anthocyanins neither stabilized the pigment or protected it against ascorbic acid-induced

degradation. The addition of catechins did initially enhance the maximum absorbance of

the three pigment sources, resulting in decreased Lightness and increased chroma color

characteristics. This improvement in color was rapidly reversed upon addition of ascorbic

acid. Samples that included ascorbic acid resulted in accelerated loss of anthocyanins that

was not improved by addition of green tea catechins. It is possible that copigmentation of

green tea catechins was too weak to prevent interaction between anthocyanins and

ascorbic acid.

48

Chapter 4. Investigating the interaction of ascorbic acid with anthocyanins and pyranoanthocyanins

Published in Molecules.

Farr JE, Giusti MM. Investigating the Interaction of Ascorbic Acid with Anthocyanins and Pyranoanthocyanins. Molecules 2018;23:744. doi:10.3390/molecules23040744.

4.1. Abstract

Juices colored by anthocyanins experience color loss related to fortification with ascorbic acid (AA), thought to be the result of condensation at Carbon-4 of anthocyanins.

To further understand this mechanism, pyranoanthocyanins, having a fourth-ring covalently occupying Carbon-4, were synthesized to compare its reactivity with AA against that of anthocyanins. Pyranoanthocyanins were synthesized by combining chokeberry anthocyanins with pyruvic acid. AA (250-1000 mg/L) was added to either chokeberry, cyanidin-3-galactoside, or 5-Carboxypyranocyanidin-3-galactoside. Samples were stored in the dark for five days at 25 C and spectra (380-700 nm), color (CIE-

L*c*h*), and composition changes (HPLC) were monitored. Extensive bleaching occurred for cyanidin-3-galactoside and chokeberry colored solutions, with a decrease in half-lives from 22.8 to 0.3 days for Cyanidin-3-galactoside when 1000mg/L AA was added. 5-Carboxypyranocyanidin-3-galactoside solution better maintained color expression with limited loss in absorbance, due to formation of colored degradation

49 products (vis-max = 477 to 487 nm), and half-life decrease from 40.8 to 2.7 days, an 8-13 fold improvement compared to anthocyanins. This suggested alternative sites of reactivity with AA. Carbon-4 may be the preferred site for AA-pigment interactions, but was not the only location. With Carbon-4 blocked, 5-Carboxypyranocyanidin-3- galactoside reacted with AA to form new pigments and reduce bleaching.

4.2. Introduction

Consumers commonly use color to make assessments on acceptance and liking, implied flavor, safety and overall quality of food products (Sharma et al. 2010). Synthetic colorants have been used to correct for natural variation of food items, mask imperfections as well as offer alternative product identities (Sharma et al. 2010). The innate stability of synthetic colorants over natural pigments has been a driver for their selection in coloring food products. Recently, this trend has begun to reverse as consumers have expressed concerns over the safety of synthetic colorants and preference for colorants from natural sources (Kobylewski & Jacobson 2010; Martins et al. 2016).

Anthocyanins are widely viewed as a natural alternative due to their wide spectrum of hue expression; however, their application has been limited due to stability (Sigurdson et al. 2017).

Anthocyanins (ACN) are a class of water-soluble polyphenols found in many fruits and vegetables. Their color properties are greatly influenced by the substitution patterns on the aglycone structure as well as pH environment (Sigurdson et al. 2017).

Warm hues including reds are observed at low pH but shift expression to vibrant purple- 50 blues in more alkaline conditions. Their stability is influenced by many factors including pH, heat, enzymes, light, as well as certain bleaching agents including sulfites, hydrogen peroxide, and vitamin C (ascorbic acid, AA) (Eiro & Heinonen 2002). The latter is of significance for the food industry with widespread use of AA as both a fortifying agent as well as an antioxidant in many food and beverage systems (Varvara et al. 2016).

It has long been known that the presence of AA in anthocyanin-colored solutions can accelerate degradation and loss of color (Sondheimer & Kertesz 1953). Bisulfites, hydrogen peroxide, as well as ascorbic acid are electrophilic compounds and are thought to attack the same nucleophilic sites of the anthocyanin. For ascorbic acid, it has been postulated to cause mutual and irreversible destruction of both the pigment and micronutrient (Shriner & Moffett 1941). This is differentiated from bisulfite bleaching which is reversible and pH dependent (Giusti & Wrolstad 2001). This presents a major hurdle for the food industry to use ACN-based colorants specifically in juices and beverages which are often fortified with vitamin C. Previous research (Poei-Langston &

Wrolstad 1981; Garcia-Viguera & Bridle 1999) has proposed that anthocyanin bleaching is the result of condensation of ascorbic acid, as well as other bleaching agents, at

Carbon-4 (C4) of the anthocyanin (Figure 12); with this site being the most susceptible to electrophilic attack. However, there is also NMR evidence suggesting alternative sites of bisulfite addition such as Carbon-2 (C2) (Berké et al. 1998). The proposed condensation is thought to result in the loss of conjugation in the C-ring, therefore lacking the original color expression of the pigment.

51 Previous work has found that anthocyanins with both 3- and 5-substitions increase pigment stability against ascorbic acid compared to just 3-substitution, likely a result of further restricting access to C4 in between. Viguera and Bridle reported that Malvidin-

3,5-diglucoside experienced slower color loss as compared to Malvidin-3-glucoside. The same authors reported direct substitution of the C4, with phenyl and methyl groups, enhanced their stability against ascorbic acid color loss versus typical –H substitution

(Garcia-Viguera & Bridle 1999). Copigmentation of grape anthocyanins with rosemary polyphenols has also been shown to have a protective effect on the pigment: It is possible that the - interaction can limit accessibility to the chromophore (Brenes et al. 2005).

Another means by which interaction between anthocyanins and ascorbic acid could be investigated is by evaluation and comparison to pyranoanthocyanins.

Pyranoanthocyanins (Figure 12) are formed by anthocyanins undergoing heterocyclic addition of a polar carboxyl-containing compound such as pyruvic acid, acetaldehyde, or catechins which are often byproducts from yeast fermentation (De

Freitas & Mateus 2011). This results in the formation of a fourth ring (D) that covalently occupies C4 and C5 of the pigment. Pyranoanthocyanins (PACN) are found in aged wines and have been reported in red and strawberries (De Freitas & Mateus 2011;

Fossen & Andersen 2003; Andersen et al. 2004). Previous research on pyranoanthocyanin stability has shown their enhanced resistance to bisulfite bleaching

(He et al. 2010; Oliveira et al. 2006; Gómez-Alonso et al. 2012) but not to ascorbic acid.

Carboxy-pyranoanthocyanins, resulting from the reaction of Malvidin-3-glucoside from grape with pyruvic acid, showed enhanced stability against bisulfite bleaching (up to 250

52 ppm) compared to anthocyanins (He et al. 2010). Oligomeric pyranoanthocyanins were shown to exhibit complete resistance to bisulfite bleaching (up to 250 ppm) for 2 days

(He et al. 2010). It is possible the oligomeric structures also block other potential sites of reaction such as C2. Acetyl-pyranoanthocyanins, synthesized with acetaldehyde, have been shown to not only overcome bisulfite bleaching, but were reported to experience a hyperchromic shift in response to bisulfite at up to 200 ppm, an unexpected response to a common bleaching agent (Gómez-Alonso et al. 2012).

Figure 12. Formation of pyranoanthocyanin from cyanidin and pyruvic acid by heterocyclic addition

The objective of this study was to compare the reactivity of anthocyanins and pyranoanthocyanins, upon the addition of AA. If anthocyanin bleaching with AA occurs only at site C4, pyranoanthocyanins, with C4 unavailable, would not undergo bleaching.

Furthermore, no changes would occur on the PACN pigments as a results of AA addition.

Our hypothesis was that C4 is not the only site of reactivity. Comparison of the reactivity of these pigments in response to ascorbic acid will aid in further understanding the deleterious interaction of this micronutrient and pigment and could allow for better selection of anthocyanin sources in applications with ascorbic acid. 53 4.3. Materials and Methods

4.3.1. Materials

Powdered chokeberry fruit was provided by Artemis Inc. (Fort Wayne, Indiana,

U.S.). Lab grade pyruvic acid used for the synthesis of pyranoanthocyanins was purchased from Sigma Aldrich (St. Louis, Missouri, U.S.). Analytical grade ascorbic acid

(99% L-ascorbic acid) was purchased from Sigma Aldrich (St. Louis, Missouri, U.S.).

HPLC grade acetonitrile and water were obtained from Fisher Scientific (Hampton, New

Hampshire, U.S.) and HPLC grade formic acid from Sigma Aldrich (St. Louis, Missouri,

U.S.).

4.3.2. Methods

Anthocyanin semi-purification (SPE)

Chokeberry powder was mixed with water acidified with 0.01% HCl prior to purification. The solution was loaded onto a Waters Sep- pak C18 cartridge for solid phase extraction (SPE). The column was then washed with acidified water (0.01% HCl) to remove of sugars and acids then followed with ethyl acetate for removal of the more non-polar phenolics. Pigments were recovered form the cartridge with methanol acidified with 0.01% HCl, and the solvent was removed by rotary evaporation (40 C, under vaccuum). Pigments were then solubilized and stored in acidified water for future use.

This was the only preparatory step for chokeberry treatments.

54 Pyranoanthocyanin synthesis

Pyrananthocyanins were synthesized from the semi-purified chokeberry by addition of pyruvic acid. The extract (1000 μM cyanidin-3-glucoside equivalent) was added to a pH 2.6 citrate buffer that had 0.1% potassium sorbate and 0.1% sodium benzoate to prevent molding. A molar ratio of 1:50 (ACN: pyruvic acid) was followed as previously described (He et al. 2010). The prepared anthocyanin pyruvic acid solution was stored in an incubator in the dark at 35 oC for 10 days (Isotemp, Fisher Scientific,

Waltham, MA, US). After the incubation period ended, cyanidin-3-galactoside and 5-

Carboxypyranocyanidin-3-galactoside, the resulting pyranoanthocyanin from cyanidin-3- galactoside and pyruvic acid, were isolated from the solution using semipreparatory

HPLC.

Anthocyanin and pyranoanthocyanin purification

A reverse phase HPLC system composed of the following modules was used: LC-

6AD pumps, CBM-20A communication module, SIL-20A HT autosampler, CTO-20A column oven, and SPD-M20A Photodiode Array detector (Shimadzu, Maryland, U.S.).

The reverse-phase column selected was a 250 x 21.2 mm Luna pentafluorophenyl column with 5 µm particle size and 100Å pore size (Phenomenex, California, U.S.). Samples were filtered prior to injection with a Phenex RC 0.45 µm, 15 mm membrane syringe filter (Phenomenex, California, U.S.). With a flow rate of 10 mL/min and a run time of

30 minutes, peaks were separated and collected. An isocratic system with the following solvents were used: 11:89 (Solvent A: Solvent B v/v) with Solvent A being 4.5% formic acid in HPLC grade water and Solvent B was HPLC grade acetonitrile. Elution of peaks 55 was monitored at 500 nm. Peaks were manually collected. The two collected peaks were diluted with distilled water and again subjected to SPE semi-purification to remove formic acid and acetonitrile. Rotary evaporation was used to remove methanol (40oC, under vaccuum), and the pigments were stored in 0.01% HCl in acidified water.

Anthocyanin and pyranoanthocyanin purity

Prior to experimentation, pigments were evaluated for purity by using an analytical HPLC only different from the previously listed one by the use of different pumps (LC-20AD, Shimadzu, MD, US). Purified pigments were filtered using the

Phenex RC 0.45 µm membranes. A binary system with 1 mL/min flow rate was used:

Solvent A: 4.5% formic acid in HPLC grade water and Solvent B: HPLC grade acetonitrile. The gradient began with an isocratic flow of 6% solvent B for 17 minutes

(elution of primary anthocyanins), increasing to 15% solvent B by 45 minutes (elution of primary pyranoanthocyanin), to 40% solvent B by 50 minutes (wash). A 10 uL injection volume was loaded onto a Phenomenex Kinetix 5µm EVO C18 100 A. 150 × 4.6 mm column and Phenomenex Ultra UHPLC EVO C18 guard cartridge attached. Purity was expressed in terms of % peak area of targeted pigment as compared to the total area of all peaks present in the max plot (260-700 nm). The isolate of 5-Carboxypyranocyanidin-3- galactoside accounted for 94% of the overall area under the curve (AUC), cyanidin-3- galactoside isolate was 92% AUC while chokeberry ACN purity was 35% AUC.

Chokeberry ACN likely contained other phenols present in the source material.

56

Sample preparation

The semi-purified chokeberry extract, the isolated cyanidin-3-galactoside, and the purified 5-Carboxypyranocyanidin-3-galactoside were diluted in pH 3.0 citrate buffer

(0.1M adjusted with HCl) until an absorbance of 1.0 at their respective vis-max was reached. Levels of AA of 250, 500, and1000 mg/L were added using a concentrated ascorbic acid stock solution, and a control consisting of each pigment with the absence of

AA was maintained. All samples were brought to the same final volume with additional citrate buffer. The pH of all samples were evaluted using a S220 SevenCompact pH meter (Mettler Toledo, Columbus, OH, U.S.) and were found to have a pH of 3.0 ± 0.05.

Samples were stored in the dark at 25 °C in an incubator (listed in 2.2.2). UV-Vis spectrophotometry, colorimetry, and HPLC analyses were conducted over a 5 day period following the addition of ascorbic acid. UV-Vis spectral data was collected every hour for the first 8 hours, and then at 12hr, and daily from that point on for 5 days. Spectra and color were eveluated with this data. HPLC analyses were conducted on days 0, 1, 3 and 5.

All treatments were run in triplicate.

UV-vis spectrophotometry of samples

A SpectraMax 190 Microplate Reader (Molecular Devices, Sunnyvale, California,

U.S.) with a 96-well plate (poly-D-lysine coated polystyrene) were used for the evaluation of absorbance from 380-700 nm, 1 nm intervals. Aliquots (200 uL) of samples were loaded into individual wells, and a blank consisted of the same citrate buffer used.

57

Color analyses of samples

Using UV-Vis spectral data in combination with software written for color conversion, absorbance data was translated to CIE-L*c*h (Farr & Giusti 2017). The calculations for CIE-L*c*h* implemented by the software used CIE relative spectral power distribution for a D65 standard illuminants and 10° observer angle function (CIE

2004).

HPLC monitoring of samples

Prepared solutions were monitored to determine the formation of potential degradation products or profile changes. Using the analytical HPLC system and conditions previously described (2.2.4), chromatograms were monitored with the max plot (260-700 nm), 490 nm (near vis-max of 5-Carboxypyranocyanidin-3-galactoside), and

520 nm (near vis-max of cyanidin-3-galactoside). A max plot of 470-520 nm was later added to standardize changes in AUC for all three pigments and their degradation products.

MS/MS Evaluation of Pigments

With the development of newly formed peaks for the pyranoanthocyanin and ascorbic acid treatment, MS/MS was performed to include additional information on the novel structures. The same HPLC conditions and column mentioned in 3.2.4 were used on a uHPLC (iNexera) system coupled to a tandem MS unit (LCMS 8040)(Shimadzu,

Maryland, USA). The following ionization conditions were used: 1.5 L/min nebulizing 58 gas flow, 230 °C desolvation line temperature, 200 °C heat block temperature, and 15

L/min drying gas flow. Two scans (Q3) were performed, both positive and negative mode from 100-1500 mass unit with an event time of 0.1 seconds. Based on initial scan results, the following events were added (positive mode) and the sample reran: product ion scan,

535 m/z; product ion scan, 519 m/z; product ion scan, 503 m/z; and precursor ion scan,

355 m/z. These additional events all had an event time of 0.1 and a collision energy of -

35.0 kV. To also consider for the possibility of condensation of the pyranoanthocyanin and ascorbic acid, the following single ion monitoring was also added: 675 m/z for 5- carboxypyranocyanidin-3-galactoside (517) + ascorbic acid (176) – H2O (18) and 673 m/z for dehydroscorbic acid (174).

Statistical analysis of data

Data was organized for means and standard deviations using Microsoft Excel

(Redmond, Washington, U.S.). 1-Way ANOVA was performed for each treatment at all time points to determine if a significant change in CIEL*c*h and maximum absorbance occurred One-Way ANOVA was also performed for each pigment (control, 250, 500,

1000 mg/L AA) at each time point to determine if and when which CIEL*c*h and maximum absorbance became significantly different from the control. Software used for

ANOVA tests was SPSS (IBM, North Castle, New York, U.S.).

59 4.4. Results and Discussion

4.4.1. UV-Vis Spectrophotometry of solutions

Anthocyanins degraded quickly in the presence of ascorbic acid, as seen in Figure

13. Chokeberry extract, containing an ACN profile which is ~70% Cyanidin-3- galactoside and with anthocyanins representing ~35% of the total AUC in the max plot, showed greater resistance to bleaching compared to the purified Cyanidin-3-galactoside.

This is likely the result of other chokeberry phenols playing a protective role against AA- induced degradation. Copigmentation as well as antioxidant capacity have both been demonstrated as a way by which additional polyphenols protect anthocyanins (Trouillas et al. 2016; Chung et al. 2016; Choe & Min 2009). Other phenolic constituents of chokeberry include procyanidins, quercetin derivatives, epicatechin, as well as chlorogenic acid (Kulling & Rawel 2008). PACN (5-Carboxypyranocyanidin-3- galactoside) derived from Cyanidin-3-galactoside showed the least change in absorbance over time. Covalently occupying C4 in 5-Carboxypyranocyanidin-3-galactoside was thought to result in less change in absorbance as compared to Cyanidin-3-galactoside and chokeberry, similar to other reports of bleaching of pyranoanthocyanins observed with bisulfites (Oliveira et al. 2006). All pigment solutions with ascorbic acid experienced signficant changes in maximum absorbance over time with p-values of less than 0.01. As

AA levels increased, the loss in absorbance for each pigment over time also increased, revealing a dose-dependent effect of AA. For the 500 mg/L AA treatments over a 5 day period, 5-Carboxypyranocyanidin-3-galactoside saw a reduction of 38% reduction in

60 maximum absorbance, chokeberry a 79% reduction, and Cyandin-3-galactoside an 88% reduction.

Changes in vis-max were also observed over the five-day period. For the 500 mg/L

AA level, the following hypsochromic changes in vis-max occurred: chokeberry, 512 to

511 nm; cyanidin-3-galactoside, 511 to 509 nm, 5-Carboxypyranocyanidin-3-galactoside,

491 to 484 nm. These shifts in vis-max are reflected in Figure 13. The change for 5-

Carboxypyranocyanidin-3-galactoside correlated with the newly developed peaks discovered during HPLC analysis, later discussed, and resulted in the solution being more orange-red. Hypsochromic changes on ACN (chokeberry extract and cyanidin-3- galactoside) vis-max observed over the 5 days of the AA treatment, were less than 5 nm, regardless of the levels of AA and were more likely attributed to pigment degradation.

The PACN experienced hypsochromic shifts as large as 10 nm with less loss in maximum absorbance, with shifts becoming more pronounced as AA levels increased.

Figure 13. Spectral absorbance changes in response to 500 mg/L AA for chokeberry, cyanidin-3- galactoside and 5-Carboxypyranocyanidin-3-galactoside colored solutions over a 5 day period

61 4.4.2. Kinetics of degradation

Degradation kinetics were evaluated in terms of change in maximum absorbance of the solutions at the original vis-max over time. Bleaching has been previously reported as a first-order reaction, typical of ACN degradation, and was modeled as such in determining the reaction rate and half-life (Sondheimer & Kertesz 1952; Ozkan et al.

2002). The decrease in maximum absorbance correlated with an increase in lightness

(L*) as well as the decrease in chroma (c*) for all pigments. The reduction in maximum absorbance did closely follow first-order kinetics for all treatments with an R2 higher than

0.94 for control treatments and 0.96 samples with added ascorbic acid. The kinetics results for each pigment and AA level can be found in Table 2. Without ascorbic acid, 5-

Carboxypyranocyanidin-3-galactoside had the greatest half-life (978 hours), followed by chokeberry extract (858 hours) and then Cyanidin-3-galactoside (546 hours). Addition of ascorbic acid dramatically reduced half-lives for all pigments. With 1000 mg/L AA added, 5-Carboxypyranocyanidin-3-galactoside half-life was 64 hours; chokeberry extract, 24 hours; and Cyanidin-3-galactoside had a half-life of 8 hours, seen in Table 2.

This order of stability was also exhibited across all AA levels. The enhanced stability and extension of half-life for 5-Carboxypyranocyanidin-3-galactoside was more evident upon addition of ascorbic acid. Pyranoanthocyanins exhibited a half-life 8-13x higher than Cyanidin-3-galactoside in the presence of AA. Kinetics data supports the hypothesis that C4 is likely the primary or preferred site for anthocyanin-ascorbic acid interaction. It also reveals that the pyranoanthocyanin is still susceptible to ascorbic acid

62 through alternative mechanisms or sites due to the half-life lowering in response to AA compared to its control.

Table 2. Reaction rates and half-life (t½) of solutions colored with different pigments stored at 25 °C in the dark, modeled with first-order kinetics. Calculations are based on the changes in Absorbance at the vis-max of the solution over time.

Ascorbic Acid k t1/2 Pigment R2 Level (hour-1) (hours) Chokeberry Extract 8.08E-04 858 0.947 Control Cyanidin-3-galactoside 1.27E-03 546 0.957 5-Carboxypyranocyanidin-3-galactoside 7.08E-04 978 0.977 Chokeberry Extract 1.02E-02 68 0.991 250 mg/L AA Cyanidin-3-galactoside 3.18E-02 22 0.996 5-Carboxypyranocyanidin-3-galactoside 2.69E-03 258 0.992 Chokeberry Extract 1.60E-02 43 0.991 500 mg/L AA Cyanidin-3-galactoside 5.90E-02 12 0.998 5-Carboxypyranocyanidin-3-galactoside 4.61E-03 150 0.965 Chokeberry Extract 2.85E-02 24 0.999 1000 mg/L AA Cyanidin-3-galactoside 8.64E-02 8 0.996 5-Carboxypyranocyanidin-3-galactoside 1.08E-02 64 0.998

The relationship between ascorbic acid level and the reaction rates was additionally assessed to observe how each of these pigments responds to ascorbic acid addition. A linear relationship was found and can be seen in Figure 14. The R2 values for these pigments at varying AA levels are the following: chokeberry extract, 0.99;

Cyanidin-3-galactoside, 0.96, and 5-carboxypyranocyanidin-3-galactoside, 0.98.

Linearity is lost to some degree for Cyanidin-3-galactoside with 1000 mg/L AA addition.

The slope could be effectively regarded as how responsive (or deleterious) the change in pigment solution maximum absorbance is upon AA addition, with a higher slope indicating greater susceptibility to AA. The slope of Cyanidin-3-galactoside was 3.1x that

63 of chokeberry extract and by comparison to 5-carboxypyranocyanidin-3-galactoside, cyanidin-3-galactoside was 8.2x and chokeberry extract 2.7x more suspectible.

1.E-01 Cyanidin-3-galactoside 9.E-02 Chokeberry Extract 8.E-02 5-Carboxypyranocyanidin-3-galactoside

7.E-02 e

t 6.E-02

a

R

n

o 5.E-02

i

t

c a

e 4.E-02 R

3.E-02

2.E-02

1.E-02

0.E+00 0 250 500 750 1000 Ascorbic Acid Level (mg/L) Figure 14. Reaction rates of 5-Carboxypyranocyanidin-3-galactoside, Cyanidin-3-galactoside, and chokeberry plotted against ascorbic acid level (0-1000 mg/L). Calculations are based on the changes in Absorbance at the vis-max of the solution over time.

4.4.3. Colorimetry

Lightness

Rapid color loss and extensive bleaching of pigments can be seen with CIE lightness (L*) in Figure 15. Within 48 hours exposed to 1000 mg/L AA, L* increased from 77.2 to 96.4 (∆19.2) for Cyanidin-3-galactoside; chokeberry, 74.4 to 89.4 (∆15.0); and 5-Carboxypyranocyanidin-3-galactoside, 81.7 to 86.6 (∆4.9). The presence of AA resulted in higher lightness over time, and this was dose-dependent. Pyranoanthocyanins

64 showed the least change in L* in reponse to AA, and Cyanidin-3-galactoside the greatest.

An increase in L* represents a lighter color expression and was most evident for chokeberry and cyanidin-3-galactoside.

Chroma

Pigment levels were standardized by absorbance at their respective vis-max; therefore, chroma values were in close agreement at day 0. Chroma, being a measure of color intensity, is useful for determining the extent of bleaching and is reported in Figure

15. The pyranoanthocyanin (5-carboxypyranocyanidin-3-galactoside, PACN) had less change in chroma compared to Cyanidin-3-galactoside and chokeberry. Chroma decreased with increasing AA levels with the exception of Cyanidin-3-galactoside after

48 hours, likely the result of ascorbic acid and pigment browning playing a larger role at those times. The changes in chroma in reponse to AA addition followed: Cyanidin-3- galactoside > chokeberry > 5-Carboxypyranocyanidin-3-galactoside. All pigments

(including controls) experienced a significant change in chroma over 5 days with p- values of less than 0.001.

Hue angle

Synthesis of pyranoanthocyanins results in a pigment with a lower vis-max and higher hue angle compared to the respective anthocyanin, having a more orange-red color expression, as compared to the red color of ACN. This was clearly observed on the initial hue angle values of 5-Carboxypyranocyanidin-3-galactoside (50°), a more orange-red hue than Cyanidin-3-galactoside (16.6°), with a more red color (Figure 15). While the initial

65 hue angle for chokeberry and Cyandin-3-galactoside started much lower and more red

(<20°) than the PACN, the reaction between ACN-AA resulted in a dramatic color shift

toward a yellow coloration . For the 1000 mg/L AA level, large increases in hue angle

were observed from day 0 to 5 for chokeberry (17.6° to 53.8°) and cyanidin-3-galactoside

(18.7° to 77.5°) while the hue angle change was much smaller for 5-

Carboxypyranocyanidin-3-galactoside, changing from 51.0° to 57.1°. Changes in hue

angle in the presence of AA were dose-dependent. The rapid increase in hue angle for

cyanidin-3-galactoside and chokeberry was likely the result of pigment degradation and

fading; whereas, for 5-Carboxypyranocyanidin-3-galactoside which better retained

original chroma and lightness parameters, new pigment formation may explan hue angle

changes.

Figure 15. Colorimetric changes (CIEL*c*h*) of change of solutions colored with chokeberry, cyanidin-3- galactoside and 5-Carboxypyranocyanidin-3-galactoside from day 0 to day 5 for all AA levels over time. Error bars represent standard deviation 66 Total color change (ΔE)

Total color changes (ΔE) were calculated as the color change from day 0 to 5 for each respective treatment, and presented in Table 3. Without AA, chokeberry had the smallest ΔE, followed by 5-Carboxypyranocyanidin-3-galactoside and Cyanidin-3- galactoside. Other phenolics present in chokeberry could have enhanced color retention and stability through mechanisms such as copigmentation or radical scavenging by additional phenolics which would have not been possible with isolated 5-

Carboxypyranocyanidin-3-galactoside and cyanidin-3-galactoside. This possible explanation could be supported by previous work where an abundance of other polyphenols has been reported in chokeberry (Kulling & Rawel 2008). The chokeberry fruit has previously been reported to have 89 mg/kg of quercetin and further supports why greater stability was observed for the chokeberry treatment as compared to purified

Cyanidin-3-galactoside (Häkkinen et al. 1999). For all levels of AA addition, chokeberry and Cyanidin-3-galactoside had a ΔE greater than 29.

Table 3. Day 0 and 5 colorimetric values (CIEL*c*h*) and total color change (ΔE) of chokeberry, cyanidin-3-galactoside and 5-Carboxypyranocyanidin-3-galactoside for all AA levels over time. Numbers are means of 3 replications, followed by (standard deviations).

67 However, 5-Carboxypyranocyanidin-3-galactoside with AA exhibited a signficiantly smaller color shift with ΔE’s ranging from 8.7 to 10.9. This three-fold reduction in ΔE resulted in overall better retention of color in response to AA addition.

4.4.4. HPLC and MS/MS Evaluation

To determine the relationship between spectral and color changes with changes in pigment composition, HPLC analysis was utilized. The initial (day 0) HPLC profile for chokeberry extract revealed the following anthocyanin profile seen in Figure 16:

Cyanidin-3-galactoside (Peak #1, 70%), Cyanidin-3-glucoside (Peak #2, 3%), Cyanidin-

3-arabinoside (Peak #3, 23%), and Cyanidin-3-xyloside (Peak #4, 4%), in line with the expected profile (Kulling & Rawel 2008). Anthocyanins contributed to 35% of the total

AUC in the max plot (260-700nm), mainly due to the presence of other polyphenols. The largest non-anthocyanin peak with a λmax of 322 nm was likely chlorogenic acid, as it is reported as being present in the berry (Kulling & Rawel 2008). After 1 day of exposure to

1 g/L AA, all anthocyanins in the chokeberry extract decreased by 65%. The loss of individual pigments ranged from 64-68%, revealing similar degradation kinetics for all pigments present. This is not surprising considering they are all cyanidin-3- monosaccharides. By day 5, only 4% of the original pigments in the chokeberry extract had survived. The behavior of Cyanidin-3-galactoside was similar to that of chokeberry extract but was thought to experience more rapid bleaching due to the absence of other polyphenols imparting a protective effect. The isolated anthocyanin accounted for 92% of

68 the AUC in the day 0 maxplot. By day 1, Cyanidin-3-galactoside was reduced by 91% and day 5, >99%.

For the pyranoanthocyanin, the isolated structure accounted for 94% of the AUC from the maxplot. By day 1, this peak was reduced 93% and 99% by day 5; However, unlike the anthocyanins, where the pigments degraded into colorless forms, the PACN-

AA interaction resulted in the development of new peaks in the visible range, labeled A,

B, and C in Figure 16. Peak A appears to be entirely newly formed in response to AA and was not present in either the control or PACN+AA day 0 treatment. Peaks B and C were present at low levels in both the control and AA treatment at day 0 and could be colored degradation products. It appears as if AA promotes the formation of these two compounds with peak B having 3.7x the AUC and C 15.9x AUC (470-520 nm) by day 1 as compared to the day 0 control treatment. These newly formed peaks also corroborate the spectra changes observed from the plate reader. The newly formed peaks had the following vis-max: A, 487 nm; B, 486 nm, and C, 477 nm. The formation of these new compounds aligned with both colorimetric data (increase in hue angle) and the spectral shift (hypsochromic response) that was observed with the solutions in response to AA.

69 The formation of the new peaks is likely how the pyranoanthocyanin solution better maintained original color expression even with the rapid loss of the parent compound.

Figure 16. HPLC profiles (470-520 nm) for 5-Carboxypyranocyanidin-3-galactoside, Cyanidin-3- galactoside, and chokeberry with 1000 mg/L AA added on day 0 and 1 The formation of three new chromophores between the PACN-AA interaction could be the result of several different phenomena and additional experiments were performed to include MS/MS data of the new structures. It is thought that AA condensation at Carbon-2 is occurring with the pyranoanthocyanins and resulting in a smaller bleaching effect compared to the Cyanidin-3-galactoside and chokeberry extract, a result of the limited accessibility to Carbon-4 (Berké et al. 1998). For the three peaks produced after AA addition, it is possible that these are the result of interaction with

PACN at alternative sites (not C2 or C4). With the addition of a fourth ring, ascorbic acid could have reacted with the D-ring substitution (carboxylic acid group) and produced 70 colored byproducts. It has previously been reported that acetyl pyranoanthocyanins experience both a hyperchromic and hypsochromic shift in response to up to 200 ppm sulfites, and it was proposed this was the result of sulfite covalent linkage at the acetyl group in D-ring, enhancing the molar absorptivity (Gómez-Alonso et al. 2012).

The MS/MS data generated provided valuable insight to the structures of the newly formed compounds. Peak A was the only structure that was not present in trace amounts in the control treatment. A positive ion scan revealed a parent m/z of 535. This was +18 mass units compared to 5-carboxypyranocyanidin-3-galactoside. A followup product ion scan revealed a daughter ion of 373 m/z, a transition of -162 m.u. from the parent ion. This is a commonly reported transition and represents the loss of galactose from the structure (Giusti et al. n.d.). It is likely that the aglycone structure is being modified and not the sugar substitution. With a parent m/z of 535, direct condensation of ascorbic or dehydroascorbic were ruled out. Ascorbic acid degradation byproducts were also considered. Ascorbic acid has previously been reported as being catalyzed by trace levels of metal (1 μM) to form hydrogen peroxide (Jansson et al. 2004; Khan & Martell

1967; Buettner & Jurkiewicz 1996). It was thought that ascorbic acid could be forming hydrogen peroxide which then could react with the pyranoanthocyanin, but other publications have led us to doubt this hypothesis. Stebbins et al. (Stebbins et al. 2016) reported the formation of 6-hydroxy-cyanidin-3-glucoside from cyanidin-3-glucoside and ascorbic acid, the transitional m/z being +16 mass units. It is possible that the pyranoanthocyanin reacts differently than the anthocyanin mentioned and undergoes

71 rearomatization or protonation after the addition of a 6-hydroxyl group, accounting for the possible difference.

Peak B, under positive ionization, had a parent m/z of 519, and the respective product ion scan revealed a daughter ion at 357 mass units. Peak C had a parent m/z of

503 and the product ion scan revealed a daughter with 341 m.u. The shifts for each of these structures of -162 m.u. from parent to daughter ion was again attributed to the loss of galactose. Of the three new compounds, the structural modification induced by ascorbic acid addtion was isolated to the aglycone. Interestingly, single ion monitoring for condensation products of ascorbic or dehydroascorbic acid with 5- carboxypyranocyanidin-3-galactoside (675 and 673 m/z, respectively) did not appear at all in the chromatogram. This supports that ascorbic and dehydroascorbic acid are not directly condensing with the pyranoanthocyanin.

To test whether the formation of the new peaks was in response to hydrogen peroxide formed as a byproduct of ascorbic acid degradation, the treatment was repeated except with hydrogen peroxide in place of ascorbic acid. The sample was monitored with a 0, 8, and 24 hour injection and and only revealed a reduction in the original pyranoanthocyanin. Surprisingly, the three new peaks were not formed in response to direct H2O2 addition and no MS/MS transitions observed with PACN+AA were found.

With the three peaks absent in both the PDA and MS chromatogram, this theory was

72 rejected. It was thought that ascorbic acid was degrading and byproducts, other than

H2O2, are reacting with the pyranoanthocyanin.

4.5. Conclusions

Cyanidin-3-galactoside degraded rapidly in the presence of ascorbic acid, followed by chokeberry extract. Other phenols in chokeberry extract likely played a protective role against AA mediated pigment bleaching. The 5-Carboxypyranocyanidin-

3-galactoside colored solution exhibited the smallest change in color (ΔE) and limited bleaching in response to ascorbic acid (for 1000 mg/L AA, ΔE of 5.2 versus 27.6 for cyanidin-3-galactoside). The interaction between PACN-AA resulted in the formation of three new chromophores, as revealed by HPLC. NMR work is underway to determine the site of reaction for PACN-AA as well the as ACN-AA reactivity. The fact that PACN, with C4 position blocked, still exhibited limited bleaching further supported the hypothesis that C4 plays an important, but not singular, role for AA mediated bleaching of anthocyanins. The pyranoanthocyanin better maintained absorbance and color expression in the presence of AA, not a result of 5-Carboxypyranocyandin-3-galactoside survival, but by formation of colored byproducts suspected at alternative sites.

73

Chapter 5. Influence of cyanidin glycosylation patterns on carboxypyranoanthocyanin formation

Published in Food Chemistry.

Farr JE, Sigurdson GT, Giusti MM. Influence of cyanidin glycosylation patterns on carboxypyranoanthocyanin formation. Food Chem 2018;259:261–9. doi:10.1016/j.foodchem.2018.03.117.

5.1. Abstract

Anthocyanins can undergo condensation with compounds having enolizable groups to form pyranoanthocyanins. These pigments are less susceptible to degradation and color changes associated with nucleophilic addition common to anthocyanins. This study aimed to evaluate the impact of glycosylation patterns of anthocyanins on carboxypyranoanthocyanin formation. Nine cyanidin derivatives were isolated by semi- preparative HPLC. Pyruvic acid was added to induce pyranoanthocyanin formation; solutions were evaluated by HPLC-MS/MS, spectrophotometry (absorbance 380-700 nm), and colorimetry (CIEL*c*h*) during 31 days storage at 25°C. Cyanidin-3- glycosides with 1→6 disaccharides produced the highest pyranoanthocyanin yield

(~31%), followed by Cyanidin-3-monoglycosides (~20%); 1→2 disaccharides produced the least proportions of pyranoanthocyanins (5-7%). Cyanidin-3-arabinoside converted to

74 pyranoanthocyanins but degraded quickly (3% yield) under these conditions. No pyranoanthocyanins were formed from Cyanidin-3-sophoroside-5-glucoside. Glycosyl bonds were more critical than the size of the substitution alone, further supported by

Cyanidin-3-(glucosyl)-(1→6)-(xylosyl-(1→2)-galactoside) yield (11%).

Pyranoanthocyanins were hypsochromically shifted and had higher hue angles than their respective anthocyanins.

5.2. Introduction

Anthocyanins (ACNs) are a class of polyphenolic pigments present in many fruits and vegetables common to the diet. Interest in their use as food colorants has been increasing due to consumer demand for more natural products and also for their many possible dietary health benefits including cancer chemoprotection, improving memory, and even atherosclerosis prevention (Hou 2003; Andres-Lacueva et al. 2005; Nasri 2012;

Krikorian et al. 2010; Wang et al. 2012). One of the most limiting factors in their use as viable alternatives to synthetic dyes and lakes is their relative instability (Henry 1996).

Many factors including light, oxygen, ascorbic acid, and enzymes induce pigment degradation and result in unappealing food products. Anthocyanins are able to undergo many types of reactions capable of stabilizing the pigment including copigmentation, self-association, polymerization, and metal complexation (Castañeda-Ovando et al.

2009). One of the lesser explored options for enhancing pigment stability would include the formation of pyranoanthocyanins (PACNs), which contain a fourth ring (Sigurdson et al. 2017). 75 This class of pigments was first observed in 1971 in red wines, formed during the winemaking process itself, and have since been reported in low levels in natural sources such as the scales of red onion and strawberries (Andersen, Fossen, Torskangerpoll,

Fossen, & Hauge, 2004; Fossen & Andersen, 2003; Somers, 1971). Pyranoanthocyanins

(PACNs) have also been reported in strawberry and raspberry juices, formed with sinapic and ferulic acid (Rein et al. 2005). Pyranoanthocyanins are generally considered derived pigments, often the result of anthocyanin condensation with yeast metabolites or other compounds (cofactors) containing an enolizable group. The reaction is the result of heterocyclic addition with the anthocyanin at C4 and hydroxyl-substituted C5 sites.

Commonly mentioned cofactors for pyranoanthocyanin formation include pyruvic acid, acetaldehyde, diacetyl, hydroxycinnamic acids, catechins, and even acetone (Lu & Foo

2001). Carboxypyranoanthocyanins are important anthocyanin derivatives in red wine and are the result of condensation between anthocyanins and pyruvic acid (Oliveira et al.

2014). The newly formed fourth ring reduces the potential for nucleophilic addition at

C4, a common cause in many anthocyanin degradation reactions (Rentzsch et al. 2007).

In fact, the carboxypyranoanthocyanin vitisin A was found to not undergo any hydration reactions (Oliveira et al. 2014).

In addition to generally being considered more stable than anthocyanins, pyranoanthocyanins have several unique traits that enhance their potential as viable food colorants (He et al. 2010). Anthocyanins at neutral to mildly acidic pH typically exist primarily in the colorless hemiacetal (carbinol) form (Brouillard 1990). With the addition of a fourth ring, equilibria shift to colorless forms are less extreme over the pH range by

76 pyranoanthocyanins (Oliveira et al. 2006). At slightly acidic pH, carboxypyranoanthocyanins maintained better color saturation (higher Chroma) and showed smaller increases in Lightness as compared to their respective anthocyanins over the same pH range (Oliveira et al. 2006). It has also been reported that anthocyanin-3- glycosides and 5-carboxypyranoanthocyanin 3-glycosides had similar molar absorptivities in pH 1 aqueous buffer or in acidified methanol (∆ε of 2000 or less), suggesting pyranoanthocyanin formation does not result in significant loss of tinctorial strength of pigment (Jordheim et al. 2007). This could prove to be beneficial in food systems which are not highly acidic or could possibly be subjected to a change in pH.

Development of pyranoanthocyanins has previously been researched in regards to winemaking and yeast selection and oxygen levels, in which certain strains were found to produce elevated levels of pyruvic acid and acetaldehyde which resulted in greater vitisin

(a pyranoanthocyanin derived from malvidin) formation (Morata et al. 2006; Quaglieri et al. 2017). Blanco-Vega et al (2011) have previously evaluated pyranoanthocyanin formation in a model wine system with Syrah grape anthocyanins. They reported that the type of acyl group branching from the C3-glucosyl moiety and the B ring substitution did not have appreciable effects of pyranoanthocyanin formation but concluded that further investigation would be needed to confirm this claim. They also reported vitisin-like pyranoanthocyanins formed from yeast metabolites pyruvic acid and acetaldehyde were maximal after only two weeks, whereas the development of hydroxycinnamic acid pyranoanthocyanins were slower to develop over a nine week period (Blanco-Vega et al.

2011). Formation of pyranoanthocyanins derived from hydroxycinnamic acids were

77 greater as the degree of hydroxyl or methoxyl substitution of the acid itself was greater

(Blanco-Vega et al. 2011). At the time of this publication, no known study has directly investigated the relationship between glycosidic substitution patterns and pyranoanthocyanin formation.

With over 700 unique anthocyanins reported in literature (Andersen & Jordheim,

2013), there is a great degree of natural structural diversity which can be an important component in the formation of these derived pigments. The objective of this study was to evaluate how different glycosylation patterns could influence pyranoanthocyanin formation, more specifically focusing on mono-, di-, and tri-glycosylation at C3 of cyanidin and the influence of sugar moiety branching patterns of the disaccharides. It was hypothesized that, generally, as the size of the substitution at C3 grows, pyranoanthocyanin formation would be hindered. Pentosides and monoglycosides, being smaller, would be more favorable than a hexosyl substitution or di or tri-glycosides in pyranoanthocyanin formation, comparatively. C5 substitution would be expected to inhibit any pyranoanthocyanin formation; therefore, an anthocyanin with 3,5- glycosylation was included.

5.3. Materials and Methods

5.3.1. Materials

Several plant materials were used for the isolation of specific anthocyanins.

Blackberry (Rubus sp.) was selected as a source of Cyanidin-3-glucoside (Fan-Chiang &

Wrolstad 2005). Chokeberry ( melanocarpa) was selected for isolation of 78 Cyanidin-3-galactoside and Cyanidin-3-arabinoside (Oszmianski & Sapis 1988). Black carrot (Daucus carota L.) was used as a starting material for Cyanidin-3-

(xylosylglucosyl)galactoside, Cyanidin-3-(xylosyl)galactoside, and Cyanidin-3-

(glucosyl)galactoside (Montilla et al. 2011). Mulberry (Morus nigra) was selected as a source of Cyanidin-3-rutinoside (Du et al. 2008). Red cabbage (Brassica oleracea L.)was used as starting materials for preparation of Cyanidin-3-sophoroside and Cyanidin-3- sophoroside-5-glucoside (Scalzo et al. 2008). The structures of the isolated anthocyanin along with the abbreviations are displayed in Figure 17.

79

Figure 17. Anthocyanin and pyranoanthocyanin aglycone structure along with sugar substitutions of selected cyanidin isolates.#→# showing glycosidic linkage of substitutions. Φ and Ψ angles shown exemplifying rotation of 1→2 glycosidic bonds and added ω rotation in 1→6 glycosidic bonds. am/z data from Q3 Scan of MS-MS, cyanidin (287 m/z) and carboxypyranoanthocyanidin aglycone (355 m/z) found for all isolates and newly formed pyranoanthocyanins. Formed Pyranoanthocyanin parent m/z was +68 by addition of fourth ring from the anthocyanin Q3 scan. bPurity described as A.U.C. (260-700 nm) of isolate peaks to profile.

5.3.2. Methods

Anthocyanin extraction and semi-purification (SPE)

Anthocyanin rich extracts of blackberry and mulberry were prepared from fresh

materials obtained from a local grocery store and harvested from local trees, respectively

(Columbus, OH). Extraction of anthocyanins from plant materials followed procedures

80 described by Rodríguez-Saona & Wrolstad (2001b) using aqueous acetone and partition with chloroform. Black carrot, chokeberry, and red cabbage, obtained in commercial powdered forms, were hydrated in acidified water (0.01% HCl) prior to semi-purification.

To achieve high yields of the cy3xylglugal and cy3xylgal, hydrated black carrot was subjected to saponification (alkaline hydrolysis), using 10% KOH to remove the hydroxycinnamic acid substitution on the acylated pigments. The same procedure was used to obtain cy3soph5glu from acylated pigments in red cabbage. Cy3glugal, not being a part of the original black carrot anthocyanin profile, was formed by subjecting the saponified black carrot anthocyanins to partial acid hydrolysis, using 2N HCl at boiling temperature to induce loss of xylose from cy3xylglugal (Durst & Wrolstad 2005). The same technique was used to form cy3soph from cy3soph5glu from red cabbage, as well.

Instead of the full suggested time of thirty minutes to remove all sugar moieties, the saponified pigments were only exposed to eight minutes of acid boil. The peaks were then isolated by semi-preparative HPLC as described below.

The solutions were then loaded onto a C18 cartridge (Waters Sep-pak) for solid phase extraction (SPE). With the pigment bound to the activated column, two volumes of acidified water (0.01% HCl) were passed through to wash sugars and polar components.

This was followed by addition of two volumes of ethyl acetate to remove less polar phenols. The bound pigment was recovered in acidified methanol (0.01% HCl). The acidified methanol was removed by use of a Buchi rotovap (New Castle, Deleware,

U.S.A.) under vacuum at 35 °C (Rodríguez-Saona et al. 2001). The pigment was resolubilized in acidified water (0.01% HCl) and frozen for further use.

81 Anthocyanin isolation

The anthocyanins listed in the materials section were isolated from the extracts through the use of semi-preparative HPLC. A semi-preparative HPLC that included the following modules was used: LC-6AD pumps, CBM-20A communication module, SIL-

20A HT autosampler, CTO-20A column oven, and SPD-M20A Photodiode Array detector (Shimadzu, Maryland, U.S.). A 250 x 21.2 mm Luna pentafluorophenyl column was use was a with 5 µm particle size and 100Å pore size guard column (Phenomenex,

California, U.S.). Samples were filtered prior to injection with a Phenex RC 0.45 µm, 15 mm membrane syringe filter (Phenomenex, California, U.S.).

The semi-purified plant extracts mentioned in the section “Anthocyanin extraction and semi-purification (SPE)” were used to obtain isolates. A 10 mL/min flow rate was used in combination with a binary gradient for separation for all anthocyanin isolation, and UV-Visible absorbance data was collected from 250-700 nm. Solvent A was 4.5% formic acid in water and solvent B was 100% acetonitrile. The gradient for isolation of black carrot and chokeberry anthocyanins started with holding 10% solvent B for 10 minutes followed by a ramp to 13% solvent B by 25 minutes. Anthocyanins from hydrolyzed red cabbage and blackberry were isolated with a gradient beginning at 10% B and increased to 35% by 50 minutes. For isolation of cy3rut from mulberry, the gradient was 8% B for the first minute and increased to 20% B over 50 minutes.

The anthocyanin isolate fractions collected from semi-preparative HPLC were dilute in a 1:1 ratio with distilled water and concentrated by the same SPE treatment described above and rotovapory drying used for plant extracts. This allowed for the 82 removal of acetonitrile and formic acid. Isolated anthocyanins were tested for purity and then frozen until use in the sample preparation step.

Anthocyanin purity and monitoring of pigments

A uHPLC system (Shimadzu Nexera-i LC-2040C, Maryland, U.S.A.), coupled with tandem MS (Shimadzu LCMS-8040, Maryland, U.S.A.), was used in evaluating initial anthocyanin purity and monitoring changes in anthocyanins with and without pyruvic acid treatment over time. A 0.4 mL/min flow rate was used on a Pinnacle DB

(Restek Corporation, Bellefonte, PA) C18 Column (1.9 µm particle size, 50 x 2.1 mm length). The column oven was set at 40 °C. The following gradient of solvent A (4.5% formic acid) and B (100% acetonitrile) was used: 0% B for the first minute, ramped to

15% B by 10 minutes. This method was effective for separation of the formed pyranoanthocyanin from anthocyanin for all isolates except cy3rut which coeluted with the formed pyranoanthocyanin. This was observed through MS-MS aglycone data. In order to obtain separation of this isolate and pyrananthocyanin derivative, the oven temperature was increased to 60 °C, and the solvent gradient was held isocratic at 2% solvent B for 10 minutes.

Mass spectroscropy was used to tentatively identify parent structures and respective aglycones of isolates. Ionizing conditions from electrospray ionization include the following: 1.5 L/min nebulizing gas flow, 230 °C desolvation line temperature, 200

°C heat block temperature, and 15 L/min drying gas flow. M+H of intact structures was evaluated using a Q1 scan with a range of 100-1000 m/z and event time of 100 millisecond. Both the cyanidin and carboxypyranocyanidin aglycone were evaluted using 83 precursor ion scan. A collision energy of -35.0 V was used in secondary collision event with the cyanidin aglycone scan looking for product ions of 287 m/z and carboxypyranocyanidin of 355 m/z (Tian et al. 2005; Oliveira et al. 2006). The difference between the parent structure of the anthocyanin and the pyranoanthocyanin was +68 m/z.

Purity of the anthocyanin isolates was described as the percent area under the curve for the target anthocyanin as compared to the total area of all peaks in the max plot from PDA data (260-700 nm). Pigment purities are reported in Figure 17. Cy3soph5glu was found to contain an unidentified phenolic, lowering purity to 85%. All others had less than 8% of impurities; however, Cy3xylglugal contained minor amounts of cy3xylgal. Cy3xylgal also contained minor amounts of cy3xylglugal and cy3gal.

Pyranoanthocyanin formation was quantified in the following terms:

퐴푈퐶 표푓 푃퐴퐶푁 푎푡 푡𝑖푚푒 푝표𝑖푛푡 푝푦푟푎푛표푎푛푡ℎ표푐푦푎푛𝑖푛 푦𝑖푒푙푑 (%) = 500−520푛푚 푥 100 퐴푈퐶 표푓 퐴퐶푁 푎푡 푑푎푦 0 500−520푛푚

Quantification of pigments was conducted on day 0, 3, 7, 14, and 31 using a standard curve of cyanidin-3-glucoside (Sigma Aldrich, St. Louis, MO) run under the same conditions used for isolates. Molar absorptivities of anthocyanins and carboxypyranoanthocyanidins have been previously reported as being relatively similar and were thought to provide a basis of confidence in using AUC500-520nm as a quantitive measure of pyranoanthocyanin yield (Jordheim et al. 2007).

Sample preparation

84 Prior to dilution, the concentrations of the isolated pigment extrtacts was determined by the pH differential method for susbsequent dilution to known concentrations (Giusti & Wrolstad 2005). The isolates were then diluted to 500 µM

(expressed as Cyanidin-3-glucoside equivalents) in pH 2.5 deionized water (previously adjusted with HCl) which contained 0.1% potassium sorbate (w/v) and 0.1% sodium benzoate (w/v) to function as preservative agents during storage. Control samples of each isolate were prepared and contained no pyruvic acid. For treatments containing pyruvic acid, a previously diluted pyruvic acid standard (pyruvic acid in the same pH 2.5 water containing preservatives) was added to obtain concentrations 100× the molar ratio of anthocyanins (50 mM of pyruvic acid). All treatments were brought to the same final volume by the addition of more pH 2.5 water. Samples were filtered through a 0.2 µm membrane into 1.5 mL glass HPLC vials. Vials were capped and stored in an incubator in the dark at 25 °C for a 31 day duration (Fisher Scientific, Waltham, MA, US). All treatments were prepared in triplicate.

UV-vis spectrophotometry of samples

Spectra of anthocyanin isolates subjected to pyruvic acid treatment was collected using a SpectraMax 190 microplate reader (Molecular Devices, Sunnyvale, California,

U.S.A). 75 µL sample aliquots were loaded in a 96-well plate (poly-D-lysine coated polystyrene). Samples were evaluated from 380 to 780 nm in 1 nm intervals. The blank consisted of the same acidified pH 2.5 water solution used to dilute pigment samples.

Spectra was collected on day 0, 7, 14, and 31. Spectrophotometric data of the samples was used to correlate with pyranoanthocyanins formation. 85 Colorimetry of samples

Colorimetric data of samples was calculated from spectra of samples collected from the plate reader with ColorBySpectra (Farr & Giusti 2017). Absorbance data was obtained and conversion to color values was achieved using the following settings: regular transmission, D65 illuminant, and 10° observer angle. Values reported are in the

CIE-L*c*h* scale. After storage for 31 days, the mathematical differences in color (ΔE) were were calculated according to the CIE-L*a*b* Delta E2000 equation (Mokrzycki &

Tatol 2012). Colorimetric data of the samples containing pyruvic acid was used to correlate with pyranoanthocyanins formation. Spectral absorbance data of the peaks of the anthocyanins and their respective pyranoanthocyanins were also collected from the

PDA during uHPLC analyses and converted to colorimetric data using ColorBySpectra

(Farr & Giusti 2017) under the settings described above. The spectra was obtained from the time of elution of the target compound and then standardized to an absorbance of 1.0 at λvis-max.

Statistical analysis of data

Evaluation of data was performed to determine significance of findings. Two-

Way ANOVA was performed to evaluate pyranoanthocyanin yield and color changes in respect to substitution, time, and the substitution * time interaction. One-Way ANOVA was also conducted to compare final yields (day 31) to compare pigment choice.

Correlation amongst anthocyanins capable of producing pyranoanthocyanin was conducted to relate pyranoanthocyanin yield to changes in color and spectral

86 characteristics using Pearson’s Correlation (two-tailed). SPSS was used to carry out these tasks (International Business Machines, Armonk, North Castle, NY).

5.4. Results & Discussion

5.4.1. Kinetics of pyranoanthocyanin formation

All isolated Cyanidin derivatives formed pyranoanthocyanins with the addition of pyruvic acid acid, with the exception of cy3soph5glu, further demonstrating the necessity of the availability of the anthocyanin C5 hydroxyl group. However, the percent of anthocyanin converted to pyranoanthocyanin of each isolate was dependent on the chemical structure of the pigment (as shown in Figure 18), and the evolution of the pyranoanthocyanin peak formation from HPLC is displayed in Figure 19. It was hypothesized that pyranoanthocyanin yield would be inversely proportional to the size of the 3-substitution with the expected order: cy3ar (pentose) > cy3gal ≈ cy3glu (hexose) > cy3xylgal ≈ cy3soph ≈ cy3rut ≈ xy3glugal > cy3xylglugal >> cy3soph5glu. However, the following order was observed from the experimental data: cy3rut (32%) > cy3glugal

(31%) > cy3glu (23%) > cy3gal (19%) > cy3xylglugal (11%) > cy3soph (7%) > cy3xylgal (5%) > cy3ar (3%). This order was surprising but could have several important implications in understanding pyranoanthocyanin formation kinetics. Both time and

87 pigment choice were statistically significant parameters in pyranoanthcyanin formation

(p-value < 0.001).

Figure 18. Pyranoanthocyanin yield (top) for anthocyanins (500 µM) subjected to pyruvic acid treatment (x100 molar ratio) and anthocyanin survival (bottom) of control treatments at pH 2.5 acidified water at day 7, 14, and 31. Yield defined as (AUC500-520nm of PACN at tn / AUC500-520nm of ACN at t0)*100. Survival defined as (AUC500-520nm of ACN at tn / AUC500-520nm of ACN at t0)*100. Standard deviation reported as error bars.

88 By comparison to the survival of the anthocyanin in control treatments (Figure

18), it did not appear as if pigment stability dictated the extent of pyranoanthocyanin formation at pH 2.5. Correlation of anthocyanin control survival and pyranoanthocyanin yield of the pyruvic acid treated samples showed no trend between the two variables

(Pearson Correlation = -0.281, p-value = 0.184). The survival of the control pigments ranged from 84 to 52% by day 31. Cy3soph and cy3rut control treatments had shorter half-lives among the 9 pigments, yet cy3rut had one of the highest and cy3soph one of the lowest pyranoanthocyanin yields. At day 7, cy3ara had the third highest pyranoanthocyanin content among the nine isolates (6%); however, the formed pyranoanthocyanin decreased at day 14 (5%) and day 31 (3%). It is likely that pyranoanthocyanin derived from cy3ar was much less stable than pyranoanthocyanins derived from the other isolates. This was an unusual finding considering the stability of the control treatment had 66% anthocyanin survival by day 31 and was not the least stable of the anthocyanins that did yield pyranoanthocyanins.

No pyranoanthocyanins were detected in cy3soph5glu over the course of thirty- one days. This is also evident in Figure 19, showing no new peaks. This provides further evidence that the hydroxyl substitution on C5 being a necessary component in the formation of the pyranoanthocyanin ring. For this reason, anthocyanin sources such as red cabbage or red radish would not be good candidates for pyranoanthocyanin formation due to their high levels of C3,5 glycosidic substitutions.

89 Another interesting observation from this data is the disaccharides cy3rut and

cy3glugal, bearing sugars with 1→6 glycosyl linkages, had the greatest

pyranoanthocyanin yield (31-32% at day 31) amongst all pigments, a 9% increase from

the next greatest yield (cy3glu, 23%). In the case of these select isolates, size might not

Figure 19. Pyranoanthocyanin formation over time as monitored by HPLC. Anthocyanin (ACN) and pyranoanthocyanin (PACN) peaks labeled. #→# showing glycosidic bond of each substitution of C3 for anthocyanin. *Cy3rut at day 14 and 31 ran under alternative HPLC conditions

90 have been the greatest factor in formation; perhaps the free rotation of glycosidic bonds as well as conformation were greater driving forces in pyranoanthocyanin formation.

With disaccharides clustered and the rapid degradation of carboxypyranocyanidin-3-arabinoside dismissed, the order of pyranoanthocyanin formation was the following: 1→6 disaccharides (cy3rut and cy3glugal) > monosaccharides (cy3glu and cy3gal) > tri-substituted (cy3xylglugal) > 1→2 disaccharides (cy3xylgal and cy3soph). The difference between 1→6 and 1→2 disaccharides alone was an over 5-fold difference in final pyranoanthocyanin yield.

Higher yield in 1→6 disaccharides versus monosacharides and 1→2 disaccharides suggested the 1→6 glycosyl linkage, possibly having a greater degree of freedom in rotation, might have enhanced the reaction; Whereas, the 1→2 glycosyl linkage, could have inhibited free movement of the substitution at C3. This is also supported by cy3xylglugal, having both the 1→2 and 1→6 linkage, having greater formation than disaccharides with 1→2 glycosyl linkages.

All disaccharides in this study have two torsion angles, phi (φ) and psi (ψ), around the glycosidic bond. However, 1→6 linked sugars have an additional torsion angle between C5 and C6 of the attached sugar called omega (ω), Figure 17. This additional torsion angle has been reported as having greater rotational freedom in the glycosidic bond (Salisburg et al. 2009; Pereira et al. 2006; Wang & Cui 2005; Perić-

Hassler et al. 2010). Other authors have studied conformational analysis of disaccharide glycosidic bonds for rotational freedom, intramolecular hydrogen bonding, and configurational entropy in water, a relevant matrix to this study. These studies have

91 shown that 1→6 linked disaccharides have significantly greater rotational flexibility in water, more conformational states, lack intramolecular hydrogen bonding, and have greater configurational entropy as compared to 1→2 linked disaccharides (Perić-Hassler et al. 2010; Pereira et al. 2006). This could support the theory that the 1→6 disaccharide substituted anthocyanins possessed greater rotational freedom and might be critical in pyranoanthocyanin formation.

Disaccharide-substituted anthocyanins with 1→2 glycosidic bonds could have had restricted stereochemistries due to the rather limited range in φ and ψ angles and possible intramolecular hydrogen bonding, whereas the the 1→6 glycosidic bond (φ, ψ, and ω) contributed to a greater number of conformational states and enhanced free rotation. This free rotation was thought to be aiding in the correct positioning of pyruvic acid or simply moving out of the way and allowing the heterocyclic SN2 reaction between pyruvic acid and the anthocyanin to occur. Free rotation of these disaccharides may play a role in increasing the reaction kinetics by increasing collision between the reactants, while the intramolecular hydrogen bonding in 1→2 disaccharides could have limited the mobility of the system and access of pyruvic acid to the critical site of the anthocyanin. As 1→6 disaccharide substituted anthocyanins had greater pyranoanthocyanin yield than the reported monosaccharides, it was thought that these disaccharides are more likely aiding in positioning pyruvic acid for reaction versus just moving out of the way. This is further supported by cy3xylglugal, containing both 1→2 and 1→6 glycosyl bonds, having greater formation than 1→2 disaccharide anthocyanins.

92 Development of the new pyranoanthocyanin peaks can be observed in Figure 19.

The area under the curve of the new pyranoanthocyanin peaks was greater over time with two exceptions, cy3soph5glu had no new peak development and newly formed carboxypyranocyanidin-3-arabinsode reduced in AUC over time. Under these HPLC conditions, newly formed pyranoanthocyanin peaks eluted 0.8-1.0 minutes after the original anthocyanin, with the exception of cy3rut. This isolate had the pyranoanthocyanin (carboxypyranocyanidin-3-rutinoside) coelute under original HPLC conditions which was unexpected but confimed by MS/MS data. Due to the observed coelution for this pyranoanthocyanin, a partially isocratic HPLC method was utilized to separate the pyranoanthocyanin and its parent anthocyanin. This resulted in broadening of the peaks as they eluted from the column.

5.4.2. Changes in Spectra during Reaction with Pyruvic Acid

Spectral characteristics of the anthocyanins treated with pyruvic acid were monitored by visible spectrophotometry during storage over 31 days, Figure 20. As would be expected of anthocyanins stored in solution, maximum absorbance decreased for all isolates (Cabrita et al. 2000). The greatest decrease in absorbance was observed for cy3ar, which showed a 71% decrease in absorbance after 31 days, Figure 20. The high rate of degradation likely also contributed to cy3ara having the greatest loss in absorbance. Anthocyanins diglycosylated with 1→6 linked sugars showed smallest decreases in absorbances, 26% and 30% for cy3glugal and cy3rut, respectively. The higher pyranoanthocyanin yield could be responsible for helping better maintain maximum absorbance.

93 Figure 20. Changes in spectral characteristics of anthocyanin isolates (500 µM) treated with pyruvic acid (x100 molar ratio) in pH 2.5 acidified water, stored over a 31 day period at 25 °C. Standard deviation represented as errors bars and (#).

The λvis-max of the solutions of the anthocyanins treated with pyruvic acid were also impacted; there appeared to be a relationship between pyranoanthocyanin yield and the hypsochromic shift of isolates. Those anthocyanins that yielded the highest proportions of pyranoanthocyanins exhibited the largest hypsochromic shifts in λvis-max.

For example cy3glugal and cy3rut yielded 31-32% pyranoanthocyanins, and their λvis-max was shifted -7 nm, Figure 20. The anthocyanin monoglycosides followed in terms pyranoanthocyanin yield as well as hypsochromic shifts in λvis-max. Cy3xylglugal, the anthocyanin trisaccharide, yielded greater proportions of pyranoanthocyanins than the 94 1→2 disaccharide glycosylated anthocyanins and also exhibited greater hypsochromic shifts, Figure 20. Overall, hypsochromic shift and pyranoanthocyanin yield were negatively correlated (Pearson Correlation = -0.575, p-value < 0.001). This relationship could be a useful tool in evaluating the extent of pyranoanthocyanin yield in similar model systems. Unlike the other anthocyanins, cy3ar underwent a bathochromic shift by day 31; however there was also a large standard deviation. These attributes were potentially result of the degradation of this pigment.

Absorbance spectra of isolates and newly formed pyranoanthocyanins were also collected from the PDA detector during HPLC analysis. The formation of the fourth ring on the anthocyanin chromophore resulted in several changes in the typical anthocyanin absorbance spectra. The pH at time of elution was ~1.74 for the peaks. As previously mentioned, hypsochromic shifts in λvis-max are known to occur with anthocyanin to pyranoanthocyanin conversion (Bakker & Timberlake 1997; Oliveira et al. 2006); however, the degree of the shift was found to differ among the anthocyanin isolates and their pyranoanthocyanin-derivatives, Figure 21. Similarly, The λvis-max of four cyanidin- pyruvic adducts bearing glucose, rutinose, sophorose, or sambubiose were reported to range 503 – 506 in aqueous buffer pH 2 (Oliveira et al. 2006). The differences in λvis-max between the anthocyanins and associated pyranoanthocyanins ranged 8-12 nm, with the greatest decreases in λvis-max noted for cy3glu, cy3gal, cy3xylgal, and cy3soph (12 nm). It was hypothesized that conversion to the carboxypyranoanthocyanin would have resulted in nearly the same hypsochromic shifts because the modification to the each

95 Anthocyanin Pyranoanthocyanin

carboxycy3xylglugal λ λ Pigment max L* c* h(°) max L* c* h(°) cy3xylglugal (nm) (nm) cy3ar 515 77.2 52.5 13.6 505 80.1 51.1 37.1

cy3gal 514 77.5 52.3 15.3 502 80.8 51.1 40.6

e

c n

a cy3glu 514 77.2 52.5 14.7 502 80.8 50.5 41.4

b

r o

s cy3xylgal 516 77.2 52.6 11.8 504 79.1 50.2 35.6

b A cy3soph 515 76.8 52.8 14.0 503 79.8 51.6 38.1

cy3rut 515 76.4 53.5 11.1 504 80.3 50.1 37.2

cy3glugal 513 78.0 51.5 15.7 505 80.2 50.4 36.1

cy3xylglugal 515 77.1 52.6 12.7 506 79.1 50.1 31.6

cy3soph5glu 513 79.7 50.5 6.8 N/A N/A N/A N/A

300 400 500 600 700 Wavelength (nm)

Figure 21. Representative UV-Vis absorbance spectra (absorbance standardized at respective λvis-max) of an anthocyanin and pyranoanthocyanin (cy3xylglugal and carboxycy3xylglugal), λvis-max (nm), and color characteristics (CIEL*c*h*) of isolated anthocyanins and their respective pyranoanthocyanins obtained from HPLC-PDA detector. chromophore would have been chemically similar. These findings suggest the structure of

the glycosylating moiety played a role in the λvis-max and color expression of

pyranoanthocyanins. In the case of anthocyanins, the interaction of the chromophore with

light (therefore color expression) can be altered by structural distortions of the aglycone

(stretching, bending, or torsion), which could be affected by chemical substitution

patterns (Malcıoğlu, Calzolari, Gebauer, Varsano, & Baroni, 2011).

The typical UV-Vis absorbance spectrum of anthocyanins was also affected in

other characteristics than just the λvis-max by conversion to pyranoanthocyanins, Figure 21.

Anthocyanins with 3-glycosylations typically exhibit a characteristic absorbance shoulder

between 420-440 nm ( Rodriguez-Saona, & Wrolstad, 1999; Harborne, 1967); However,

this shoulder was eliminated or hidden when the pigment was converted to a

carboxypyranoanthocyanin. Another characteristic difference of pyranoanthocyanin

96 spectra as compared to anthocyanin spectra would be the loss of the shoulder at 280 nm, typical of flavonoids. Additionally a new peak at 352-353 nm was found, consistent with the spectral distribution of pyranoanthocyanin derivative of malvidin-3-glucoside

(Fulcrand et al. 1998). The majority of the changes occurred in the UV absorbance region, and therefore would not have large effects on the colorimetric properties of the pigments.These spectral qualities were observed for all carboxypyranoanthocyanins formed and were similarly observed in literature with vitisin A and B formation (Bakker

& Timberlake 1997).

5.4.3. Changes in Color during Reaction with Pyruvic Acid

Color changes among the 9 anthocyanin isolates were evaluated over the 31 day period using the CIE-L*c*h* color space. Spectral data collected with a plate reader was converted with ColorBySpectra to determine colorimetric values. Changes in Lightness, chroma, and hue angle can be seen in Figure 22. For all isolates, lightness increased by day 31. Cy3soph5glu had a higher initial L (87.5) as compared to anthocyanins lacking a

C5 substitution (68.5 to 72.0). The pKh of Cy-3-glu-5-glu has been reported as 2.23

(Mazza & Brouillard 1987); therefore hydration and lighter color expression of cy3soph5glu would be expected to have occurred under the conditions of this study.

Cy3ar showed the greatest increase in Lightness (ΔL 10.2) and cy3glugal the smallest

(ΔL 1.8). Chroma, or color saturation, showed a similar pattern. Cy3ar, again had the greatest change (Δc -42.5) and cy3rut the smallest (Δc -9.7). During storage, all isolates showed overall reductions in chroma and increases in lightness; however, cy3ar showed the greatest amount of color loss.

97 Hue angles were expected to increase proportionally to increasing pyranoanthocyanin yield. Compared to their respective anthocyanins, pyranoanthocyanins are hypsochromically shifted, which is often thought be a contributor for the color evolution of red wine changing from purple-red to brick-red during aging

(He et al. 2012; Bakker & Timberlake 1997). The following order of pyranoanthocyanin yield was reported above as the following: cy3rut > cy3glugal > cy3glu > cy3gal > cy3xylglugal > cy3soph > cy3xylgal > cy3ar. Interestingly, only the anthocyanins glycosylated with 1→6 disaccharides demonstrated increases in hue angle (1.1° for cy3rut and 0.5° for cy3glugal); while all the other pyranoanthocyanin-forming isolates showed decreases in hue angle over time, Figure 22. Generally, the largest decreases in hue angle were demonstrated by pyruvic acid treated anthocyanins that bore glycosides with 1→2 types of linkages. Cy3ar also showed large decreases in hue as a response to degradation. Unlike the hypothesis that an increase in hue (to appear more red-orange) would occur with pyranoanthocyanin formation, there was not a statistically significant correlation between change in hue angle and pyranoanthocyanin yield.

98 Figure 22. Color parameters for CIE Lightness, Chroma and Hue angle for anthocyanins (500 µM) subjected to pyruvic acid treatment (x100 molar ratio) in pH 2.5 acidified water after 31 days of storage at 25 °C. Standard deviation represented as errors bars and (#).

The changes in color of samples treated with pyruvic acid were also monitored in terms of ΔE, comparing day 0 and day 31 colorimetric parameters. The following total color change values (ΔE) were observed: cy3ar: 16.4, cy3gal: 8.2, cy3glu: 7.2, cy3xylgal:

7.8, cy3soph: 9.2, cy3rut: 4.9, cy3glugal: 3.6, cy3xylglugal: 8.3, and cy3soph5glu: 9.7.

As expected based on the L*, c* and, h* values, cy3ar showed the greatest color change while cy3rut and cy3glugal demonstrated smallest changes in color. The higher pyranoanthocyanin yields observed with cy3rut and cy3glugal may have played important roles in better maintaining colorimetric stability. An increase in L* and

99 decrease in c* was observed in conversion of the anthocyanin to its pyranoanthocyanin derivative, Figure 22.

As another point of comparison, colorimetric data was calculated from the absorbance spectra (standardized to absorbance

1.0 at respective λvis-max) of the peaks of the anthocyanins and their respective pyranoanthocyanins under HPLC conditions, Figure 21.

Small differences in L* and c* between the anthocyanin and its respective pyranoanthocyanin were observed despite being standardized, which indicates that the spectral distribution of the pigments had effects on their overall lightness and saturation. Hue angles were also increased by ~20° for all pigments, which correlated to the observed hypsochromic shifts in λvis-max. Thus, the pyranoanthocyanis were comparatively more orange than the anthocyanin precursors. These changes in hue are consistent with pyranoanthocyanin formation in wine and its color evolution from red- purple to a more orange hue (Fernandes et al. 2017; Quaglieri et al.

2017).

5.5. Conclusion

100 Glycosidic substitution patterns on C3 of cyanidin played important roles on pyrano-formation efficiency. Number of glycosidic substitutions on the anthocyanin did not completely predict the order of efficiency of pyranoanthocyanin formation. Instead cy3rut and cy3glugal, bearing 1→6 disaccharides, converted to carboxypyranoanthocyanins with greater yield. Anthocyanins monoglycosylated with hexoses followed in pyranoanthocyanin yield. Cy3ar showed a high initial rate of carboxypyranoanthocyanin formation, but this was lost by day 31, likely due to the poor stability of both the anthocyanin and its pyrano-derivative. The structure of the glycosylating moiety was demonstrated to play important roles in pyranoanthocyanin formation, particularly in the case of di- and tri-saccharide. The glycosylations comprised of only 1→2 glycosidic bonds showed low carboxypyranoanthocyanin formation while those with 1→6 glycosidic bonds seemed to have enhanced capacity to derivatize. These bonds, as free molecules, have exhibited greater degrees of rotation in solution which may work to increase collision between reactants and attract pyruvic acid to the chromophore. Cy3soph5glu had no pyranoanthocyanin formation, confirming the importance of a hydroxyl group on the C5 site. A decrease in chroma and increase in lightness were observed for all isolates, thought to be the result of degradation; however, greater pyranoanthocyanin yield enhanced color stability (smaller ΔE values).

Glycosylation structural conformation was found to be a critical parameter in pyranoanthocyanin formation and must be considered in order to best control production of these pigments.

101

Chapter 6. Stereochemistry of C3-Glycosidic Attachments Affects the Reactivity of Cyanidin Derivatives 6.1. Abstract

The anthocyanidin glycosylation-reactivity relationship has largely been investigated in terms of sugar type, size, and site. The glycosidic branching patterns have yet to be explored for their impact on anthocyanin stability. Seven cyanidin glycosylates with differing C3 branching patterns were isolated by semi-preparative HPLC, diluted in pH 1-9 buffers and stored for 2 weeks at 25 °C in the dark. Bleaching with bisulfite and ascorbic acid at pH 3 were also tested. Spectral changes were monitored to evaluate kinetics of degradation and the extent of bleaching. Cyanidin-3-galactoside was more suspectible to hydration (14.2% vs 17.8% absorbance retention, pH 3) and bleaching

(KSO2 of 11.0 vs 9.84) than cyanidin-3-glucoside. The 1→2 disaccharides exhibited higher vis-max (up to 16 nm higher) as well as enhanced resistance to hydration and 102 bleaching as compared to their 1→6 counterparts. The 1→6 disaccharides had similar

vis-max (often within 2 nm) and slightly improved resistance to hydration and bleaching to the monosaccharides. The tri-glycosylated anthocyanin behavior was often intermediate to the 1→2 and 1→6 disaccharides. The 1→2 disaccharides generally exhibited lower half-lives in comparison to the monosaccharides; whereas, the 1→6 disaccharides exhibited higher stability. This was contrary to past assumptions that greater number of sugars equates to enhanced stability. Rotational freedom of glycosidic bonds, intramolecular hydrogen bonding, as well as stereochemistry of the sugars were thought to be critical factors in anthocyanin stabilization and reactivity by glycosylation.

6.2. Introduction

The reactivity and stability of anthocyanins are known to be affected by the natural chemical composition of the pigment. Anthocyanins are composed of a basic structure of 2-phenylbenzopyrylium of flavylium salt, with varying degrees of substitution (Jing et al. 2008). Anthocyanidins are inherently unstable and are primarily found glycosylated in nature; the glycosyl addition supports anthocyanin stability by formation of hydrogen-bonding, enhanced steric hindrance, and decreased electron delocalization (Jing et al. 2008; Zhao et al. 2014). Glycosylation at C3 is most common, followed by C3 and C5, and least common are C7 or C4 substitution: As many as three sugars may be connected at a single site but more common of a trisaccharide is a disaccharide present at C3 and an additional sugar at C5 or C7 (Zhao et al. 2014;

Andersen & Jordheim 2006; Markakis & Jurd 1974).

103 It is commonly accepted that increasing the number of glycosyls at C3 results in increased stability of the molecules (Zhao et al. 2014). Several sources have provided evidence of the enhanced stability of 3-trisaccharide anthocyanins in comparison to 3- monosaccharides (Sui et al. 2014; García-Viguera et al. 1998; Eiro & Heinonen 2002;

Bonerz et al. 2007). Evidence regarding stability for anthocyanin 3-disaccharides, however, is generally more conflicting. For example, cyanidin-3-rutinoside

(rhamnosyl(1→6)glucoside) was reported as being more stable to heat treatment as compared to the trisaccharide derivative cyanidin-3-glucosyl(1→2)-rutinoside in sour cherry juice (Szalóki-Dorkó et al. 2015) and also more stable than cyanidin-3-glucoside alone (Sui et al. 2014). In certain cultivars of sour cherry, cyanidin-3-xylosyl(1→2)- rutinoside was reported as being less stable than cyanidin-3-rutinoside and cyanidin-3- sophoroside (glucosyl(1→2)glucoside) (Bonerz et al. 2007). In the same study mentioned by Bonerz and others (2007), cyanidin-3-glucosyl(1→2)-rutinoside always exhibited superior stability to the disaccharides present in sour cherry juice even when cyanidin-3- xylosyl(1→2)-rutinoside did not. For the formation of carboxy-pyranoanthocyanins from anthocyanins, 1→6 disaccharides have previously been reported as having higher yield than monosaccharides and 1→2 disaccharides (Farr et al. 2018). In addition to the number of sugars, the type has also been reported as modifying anthocyanin stability.

Susceptibility to acid hydrolysis of anthocyanin monosaccharides have been reported in the following order arabinoside > galactoside > glucoside (Ichiyanagi et al. 2001). These reports suggest that the type of sugar attachment, their stereochemical organization

104 around the chromophore and thus location of glycosidic linkages within the di- or tri- saccharide all may affect anthocyanin stability.

Beyond being innately reactive, anthocyanins have been found to bleach in response to certain chemical stimuli. This reaction is denoted by anthocyanin color fading

(decrease in Chroma, increase in Lightness) and results in unattractive food products.

Bleaching is thought to occur as a result of electrophilic species such as bisulfite, hydrogen peroxide, and/or ascorbic acid attacking the nucleophilic sites of the aglycone.

There is evidence suggesting these types of reactions are inititated with condensation between the reactive species and anthocyanin primarily at Carbon-4 and and less preferentially at Carbon-2 of the aglycone structure (Berké et al. 1998; Poei-Langston &

Wrolstad 1981). The effects of bleaching from both bisulfite and ascorbic acid are dramatically reduced with pyranoanthocyanins, derivatives of anthocyanins with a fourth ring at Carbon-4 and Carbon-5. Bisulfite is a well known bleaching agent of anthocyanins and is often used as a tool in determining the extent of color of a solution due to polymeric anthocyanins, which are linked by Carbon-4-Carbon-8 bridging (Giusti &

Wrolstad 2001). The 3,5-glycosylated anthocyanins have been reported to be significantly more resistant to bisulfite bleaching as compared to just 3-glycosylated anthocyanins (Garcia-Viguera & Bridle 1999). However, there is limited evidence on the possible influence within 3-substituted glycosylations, which are more commonly encountered in nature. As the stability of anthocyanin 3-glycosides hase been found to vary, the differing number of glycosyl units as well as branching patterns may influence the extent and rate of anthocyanin bleaching.

105 Although it is generally recognized that the stability of anthocyanidins is improved by glycosylation, there is some conflicting evidence regarding the effects of number and types of glycosylations on anthocyanin stability. Therefore, the objective of this study was to evaluate the possible effects of stereochemistry of glycosylations on the reactivity of cyanidin derivatives. It was hypothesized that the number of sugar substitutions would be key for anthocyanin stability, and also that the position of attachments would play an important role. It is expected that increasing the number of glycosylations off C3 of cyanidin will increase stability at pH 1-9 and will also increase resistance to hydration and bleaching. Significant differences in reactivity of anthocyanins bearing 1→2 and 1→6 disaccharides are also hypothesized to occur with

1→6 anthocyanins being less stable and more prone to bleaching.

6.3. Materials and Methods

6.3.1. Materials

106 Seven anthocyanin isolates were obtained from the following plant materials:

Cyanidin-3-glucoside from blackberry (Rubus sp.); Cyanidin-3-galactoside from

chokeberry (Aronia melanocarpa); Cyanidin-3-(xylosyl)galactoside, and Cyanidin-3-

(glucosyl)galactoside, Cyanidin-3-(xylosylglucosyl)galactoside from black carrot

(Daucus carota L.); Cyanidin-3-rutinoside from mulberry (Morus nigra); and Cyanidin-

3-sophoroside from red cabbage (Brassica oleracea L.). The isolated anthocyanins and

their 3-glycosidic substitutions are presented in Figure 23. For the presentation of

Figure 23. Cyanidin aglycone structure, different glycosyl moieties, branching patterns (linkage), and purity of isolated pigments. Purity defined as the % area under curve of the target anthocyanin to any other peaks present in the PDA chromatogram (260-700 nm) findings in this study, the isolated pigments are referred to by the aglycone and individual

sugars and linkages.

6.3.2. Methods

Anthocyanin extraction and semi-purification (SPE)

107 Mulberry and chokeberry anthocyanins were locally harvested and prepared from fresh materials (Columbus, OH, USA). Blackberries were purchased at a local grocery store (Columbus, OH, USA). Anthocyanin extraction from these sources was completed using an acetone-chloroform partition detailed by Rodriguez-Saona et al. (2001) . Black carrot and red cabbage were obtained as commercial powders from Sensient (St. Louis,

Missouri, USA) and were hydrated in acidified water (0.01% HCl v/v). High yields of cyanidin-xylosyl(1→2)glucosyl(1→6)galactoside and cyanidin-xylosyl(1→2)galactoside were achieved through saponification (alkaline hydrolysis) of the solubilized powder.

Red cabbage anthocyanins were also saponified, largely resulting in cyanidin-3- glucosyl(1→2)glucoside-5-glucoside. The black carrot and red cabbage solution was diluted in a 10% KOH (in water) for removal of hydroxycinnamic acids present on the acylated pigments. Cyanidin-3-glucosyl(1→6)galactoside is an anthocyanin that is not naturally present in black carrot and was obtained by partial acid hydrolysis of the saponified black carrot anthocyanins. These pigments were diluted in 2N HCl and subjected to 8 minutes of an acid boil to induce the random cleavage of glycosidic bonds from cyanidin-xylosyl(1→2)glucosyl(1→6)galactoside, resulting in the production of the cyanidin-3-glucosyl(1→6)galactoside. Saponifed red cabbage anthocyanins were subjected to the same partial acid hydrolysis treatment to recover cyanidin-3- glucosyl(1→2)glucoside from cyanidin-3-glucosyl(1→2)glucoside-5-glucoside (Durst &

Wrolstad 2005).

Solid phase extraction of the pigments was completed by loading pigment solutions onto an activated C18 cartridge (Waters Sep-pak, Massachusetts, USA). The

108 adsorbed pigments were then washed with two volumes of acidified water (0.01% HCl v/v) for removal of sugars and other polar components. Less polar components such as the aromatic acids cleaved from the saponification step were washed out with a two column volume wash of ethyl acetate. Bound pigments were recovered from the column with acidified methanol (0.01% HCl v/v). The anthocyanin methanol mixture was then dried using a rotovap (Buchi, New Castle, Delaware, USA) under vacuum at 35 °C

(Rodríguez-Saona et al. 2001). With only the anthocyanin remaining, the pigment was dissolved with acidified water (0.01% HCl v/v) and frozen. Anthocyanins from these sources were then isolated through use of semi-preparative HPLC.

Anthocyanin isolation

Isolation of anthocyanins was achieved by use of semi-preparative HPLC. The parameters of separation and collection are reported by Farr and others (2018). Prior to injection, samples were filtered with a Phenex RC 0.45 µm, 15 mm membrane syringe filter (Phenomenex, California, USA). The semi-preparative HPLC included the following modules: LC-6AD pumps, CBM-20A communication module, SIL-20A HT autosampler, CTO-20A column oven, and SPD-M20A Photodiode Array detector

(Shimadzu, Maryland, USA). A Luna pentafluorophenyl column (250 x 21.2 mm) was use was a with 5 µm particle size and 100Å pore size guard column (Phenomenex,

California, USA). The binary system included a 10 mL/min flow with the following solvents: A, 4.5% formic acid; B, 100% acetonitrile.

Anthocyanin purity and monitoring of pigments

109 Purity was evaluated through the use of an uHPLC (iNexera) system coupled to a tandem MS unit (LCMS 8040)(Shimadzu, Maryland, USA). Method parameters are described by Farr and others (2018). Pigment purity was defined as the % area under curve of the the target anthocyanin to any other peaks present in the PDA chromatogram

(260-700 nm). A Pinnacle DB (Restek Corporation, Pennsylvania, USA) was used in combination with a binary gradient system (Solvent A: 4.5% formic acid and Solvent B:

100% acetonitrile). The peaks of target anthocyanins were further identified by use of the tandem MS. The aglycone weight (cyanidin, 287 m/z) as well as the intact weight (M+H) of all anthocyanins from precursor ion analysis served as an additional means of support to confirm the target isolates. All isolates were found to have the aglycone m/z of 287 m/z and the [M+H]+ were as follows: cyanidin-3-galactoside, 449; cyanidin-3-gucoside,

449; cyanidin-3-xylosyl(1→2)galactoside, 581; cyanidin-3-glucosyl(1→2)glucoside,

611; cyanidin-3-xylosyl(1→2)galactoside, 581; cyanidin-3-rhamnosyl(1→6)glucoside,

595; cyanidin-3-glucosyl(1→6)galactoside, 611; cyanidin-3- xylosyl(1→2)glucosyl(1→6)galactoside, 743.

Evaluation of Reactivity in pH 1-9

For evaluating anthocyanin isolates from pH 1 to 9, isolates were quantified by the pH differential method (Giusti & Wrolstad 2001). The isolates were diluted into pH 1

(0.025 M KCl) and pH 4.5 buffer (0.4 M sodium acetate) and were quantified in 110 Cyanidin-3-glucoside equivalency with a 449.2 MW and molar absorption coefficient (ε) of 26,900 L mol-1 cm-1. Isolates were diluted to 50 µM Cyanidin-3-glucoside equivalents in buffer pH’s 1-9 in a 96 well poly-lysine well plate. Buffers were composed of 0.025 M

KCl for pH 1-2, 0.1 M sodium acetate for pH 3-6, 0.25 M TRIS for pH 7-8, and 0.1 M sodium bicarbonate for pH 9. The pH of the systems was adjusted with concentrated HCl or 10% NaOH prior to final volume adjustment with distilled deionized water. All samples were prepared in triplicate. The pH of all samples was monitored after dilution with a Mettler Toledo Inc. S220 SevenCompact™ pH/Ion meter (Columbus, OH).

Readings were taken on a SpectraMax (Molecular Devices, Sunnyvale, CA) plate reader at the following hour time points: 0, 1, 2, 4, 6, 24, 48, 72, 168, and 336 hours. The readings measured spectra from 380 to 780 nm (1 nm interval) with plates being shaken for 5 seconds before the reading. Samples were sealed and stored in the dark at 25 oC between readings. Kinetics of the stability of isolates for all pH levels were calculated as first order.

Evaluation of Reactivity to Bleaching Agents

Isolates were diluted in pH 3 buffer (0.1 M Citrate) to a maximum absorbance of

1.0 at max to evaluate their extent of bleaching in the presence of bisulfite or ascorbic acid. Samples were prepared in triplicate. Spectra was evaluated using the same instrument and conditions mentioned in the Evaluation of Reactivity in pH 1-9 section.

Anthocyanin bleaching with bisulfite

111 The following levels of bisulfite were added: 0, 50, 100, 150, and 200 ppm. The effect of bisulfite bleaching was quick as compared to ascorbic acid bleaching and a bleaching constant (KSO2) from the first spectra scan was used to differentiate the effects.

The reversal of bisulfite bleaching over time was recorded at 24 and 48 hours following sample preparation. The natural log of maximum absorbance versus ppm bisulfite was plotted, and the absolute value of the slope was considered the bleaching constant. Using the natural log provided a higher R2 versus using the slope of maximum absorbance versus bisulfite level alone.

Anthocyanin bleaching with Ascorbic acid

For ascorbic acid bleaching, the following levels were used: 0, 300, 600, 1500, and 3000 ppm. The same buffer and pigment levels from bisulfite were used for ascorbic acid bleaching evaluation and readings were taken at the following time points: 0, 1, 2, 4,

6 hours, followed by daily readings for 7 days. Kinetics of reactivity were calculated following first-order kinetics.

Statistics

Figures and data means and standard deviations were produced using Microsoft

Office Excel 2010 (Office 14.0, Microsoft. Redmond, WA). The λvis-max, % absorbance retention, bleaching constants and rates of the different pigments were evaluated by 1- way analysis of variance (ANOVA) (2-tailed, α = 0.05) and Tukey’s t-test (α = 0.05) using Minitab 16 (Minitab Inc., State College, PA).

112 6.4. Results

6.4.1. Initial vis-max differences at different pH

The glycosidic subsitution patterns of o-dihydroxylated anthocyanins have previously been demonstrated to uniquely impact their spectal and colorimetric properities, particularly when comparing 3-glycosylation to 3,5-glycosylation patterns

(Sigurdson et al. 2018; Stintzing et al. 2002; Torskangerpoll & Andersen 2005). To further explore the role of the stereochemistry of C3 glycosylations on the spectral characteristics of cyanidin derivatives, the absorbance spectra of the isolates were collected after dilution in pH 1 through 9.

With C3 monoglycosylation, small differences were observed between the vis-max of cyanidin-3-glucoside and cyanidin-3-galactoside. At all pH levels, with the exception of pH 7, the vis-max were not significantly different and within 2 nm of one another. The

vis-max of cyanidin-3-glucoside was 11 nm higher than that of cyanidin-3-galactoside at pH 7. Although not signficantly different (p-value > 0.05), the vis-max of cyanidin-3- glucoside was generally numerically greater, Table 4. This was consistent with a previous report in which the vis-max of Cyanidin-3-glucoside was slightly greater than the galactoside derivative in alkaline pH (Sigurdson et al. 2018).

Di- and tri-glycosylation at C3 of cyanidin generally resulted in expression of greater vis-max than monoglycosides in all pH, Table 4. Interestingly, the vis-max of

Cyanidin-3-Glucosyl(1→2)Glucoside was greatest in all but pH 1, even compared to the triglycosylated derivative. Overall, the anthocyanins bearing 1→2 disaccharides

(Xylosyl(1→2)Galactoside and Glucosyl(1→2)Glucoside) were found to have a greater 113 vis-max compared to other derivatives. While the Cyanidin derivatives bearing 1→6 disaccharides exhibited vis-max similar to their respective monosaccharides. With a few notable exceptions, glucosyl-derived isolates showed the following behavior in vis-max:

Glucosyl(1→2)Glucoside > Rhamnosyl(1→6)Glucoside ≈ Glucoside. The galactoside- derived isolates followed a similar pattern with the order being:

Xylosyl(1→2)Galactoside > Glucosyl(1→6)Galactoside ≈ Galactoside. These findings in

vis-max are corroborated by Stintzing and others (2002): when comparing the vis-max of glucoside, xylosyl(1→2)galactoside, and xylosyl(1→2)glucosyl(1→6)galactoside 3- substitutions of cyanidin they found a 3 nm increase from mono-saccharide to the di- and tri-saccharide and a decrease in hue angle. The modulation of the anthocyanin chromophore has been found to drive the diversity of color expression of the molecule through many theoretical mechanisms. Planar distortion and copigmentation are both of interest as means of altered color expression (Trouillas et al. 2016; Malcıoğlu et al.

2011). Glycosyl moieties, having different stereochemisitries, are also likely to differently impact the spectral properties of anthocyanins. Sugars with 6´´ attachments are thought to have greater free rotation of the secondary glycosyl moiety when compared to1→2 disaccharides (Pereira et al. 2006; Perić-Hassler et al. 2010; Farr et al.

2018) and thus may result in differences in the geometric modification of the chromophore. Further supporting this hypothesis, the tri-substituted Cyanidin-3-

Xylosyl(1→2)Glucosyl(1→6)Galactoside isolate had a max within 1 nm of Cyanidin-3-

Xylosyl(1→2)Galactoside at pH 1-3, but for pH 4-9 it was closer to the average of the

max of the 1→2 and 1→6 Galactoside isolates.

114 Table 4. λmax (in nm) of glycosylated cyanidin derivatives in pH 1-9. Values reported are means (n = 3) and (standard deviation). Different superscript letters indicate significant differences between anthocyanins in the same pH.

λmax at different pH Cyanidin 1 2 3 4 5 6 7 8 9 509c 510b 511c 512c 521bc 517c 553bc 567cd 569c galactoside (0) (1) (0) (2) (4) (1) (2) (0) (2) 509c 510b 511c 513c 522abc 519c 564a 569c 570c glucoside (0) (1) (1) (1) (3) (3) (1) (1) (1) 513a 513a 514a 518a 529a 532a 559ab 578a 578a xylosyl(1à2)galactoside (0) (1) (0) (0) (2) (0) (2) (1) (1) 512ab 513a 513ab 515b 528ab 535a 564a 577a 577a glucosyl(1à2)glucoside (1) (0) (1) (1) (0) (2) (0) (1) (1) 512b 512a 513b 516b 520c 518c 553bc 567cd 568c rhamnosyl(1à6)glucoside (1) (1) (1) (1) (4) (1) (1) (0) (1) 509c 509b 510c 512c 517c 519c 550c 566d 570c glucosyl(1à6)galactoside (0) (1) (0) (0) (2) (0) (2) (1) (1) 513ab 513a 513b 515b 524abc 527b 558ab 572b 574b xylosyl(1à2)glucosyl(1à6)galactoside (1) (0) (1) (1) (0) (1) (6) (1) (1)

6.4.2. Absorbance Retention after Equilibration at Different pH

Anthocyanins are thought to typically express the most vibrant colors at very acidic pH which then decreases at mildly acidic pH and increasing again at alkaline pH.

Glycosylation of cyanidin at both C3 and C5 has been reported to decrease pKh, making the anthocyanin more prone to hydration and color loss (Markakis & Jurd 1974; Sui et al.

2014). In order to assess the role of the glycosylation structures on Carbon-3 on the degree of hydration, the absorbance at pHn was compared to the absorbance at pH1 at the respective λvis-max for each cyanidin isolate, Table 5. These glycosyl moieties were thought to be more prone to forming intramolecular hydrogen bonds with the chromophore were expected to decrease the hydration of the anthocyanin, in turn reducing the rate of degradation.

115 Table 5. % absorbance retention of glycosylated cyanidin derivatives in pH 1-9. % Absorbance retention is defined as Absorbance in pHn / Absorbance in pH1 at respective λmax ×100, after 15-30 min equilibration. Values reported are means (n = 3) and (standard deviation)

% Abs Ret Cyanidin 2 3 4 5 6 7 8 9 92.0ab 71.2d 14.2f 5.0d 15.8d 39.6bc 51.7c 69.4b galactoside (0.6)a (1.4) (0.2) (0.1) (0.1) (0.2) (0.1) (0.5) 95.1 73.3cd 17.8d 9.8b 19.7bc 44.4ab 58.2a 73.0ab glucoside (1.3) (1.9) (0.8) (1.0) (1.8) (4.8) (2.3) (5.5) 95.8a 85.3a 28.1a 13.2a 18.9bc 46.9a 61.4a 78.2a xylosyl(1à2)galactoside (0.6) (0.3) (0.1) (0.1) (0.1) (0.4) (0.2) (0.7) 94.5a 75.8c 19.6c 9.3b 9.2e 42.4abc 57.1ab 73.2ab glucosyl(1à2)glucoside (1.1) (1.1) (0.3) (0.1) (1.0) (0.8) (3.7) (0.5) 88.4b 63.5e 15.4e 7.8c 17.6cd 38.6c 44.2d 55.4c rhamnosyl(1à6)glucoside (3.6) (1.4) (0.3) (0.9) (0.4) (1.5) (0.3) (4.0) 93.9a 72.9cd 19.0c 9.2bc 27.3a 43.1abc 58.1a 77.3a glucosyl(1à6)galactoside (1.3) (1.0) (0.2) (0.3) (0.1) (0.4) (1.7) (1.2) 96.3a 79.6b 23.8b 10.2b 21.1b 40.1bc 53.1bc 68.9b xylosyl(1à2)glucosyl(1à6)galactoside (0.3) (0.4) (0.1) (0.0) (0.3) (0.5) (0.4) (0.9)

Comparing these 3-monoglycosides, Cyanidin-3-glucoside was more resistant to color loss as compared to the Cyanidin-3-galactoside isolate at all pHs. Although both sugars are hexosides, the difference in equatorial and axial position of the –OH on C4 of the sugars appeared to impact the degree of hydration of the anthocyanins more strongly than the impact on their λvis-max. Diglycosylation of C3 on Cyanidin demonstrated much more variable results on the degree of loss of color of the pigments. Generally, Cyanidin derivatives bearing 1→2 disaccharides were most resistant to hydration, especially when comparing among glucoside derivatives or among galactoside derivatives. At pH 2 - 5, the 1→2 disaccharides were more resistant to color loss compared to the 1→6 disaccharides, Table 5. Cyanidin derivatives with 1→6 disaccharides were overall more prone to color loss than 1→2 disaccharides, as hypothesized, but their amount of color loss was sometimes greater than even those with monosaccharides. For example at pH 2

116 and 3, the 1→6 disaccharides (especially Cyanidin-3- Rhamnosyl(1→6)Glucoside) were found to be more prone to hydration compared to the monosaccharides, Table 5. The % absorbance retention at pH 3 suggests that the pKh is lower for 1→6 disaccharides compared to 1→2, resulting in higher predominance of colorless forms. This may also explain why the 1→6 disaccharides had recovered more absorbance than the 1→2 disaccharides at pH 6. The degree of hydration and color loss of Cyanidin bearing the trisaccharide with 1→2 and 1→6 bonds varied between the derivatives having either

1→2 or 1→6 disaccharides, suggesting combative effects of the two types of glycosyl bonds.

6.4.3. pH: stability & half lives

The stability of anthocyanidins is increased by plants in vivo by glycosylation almost immediately after production of the aglycone (Zhao et al. 2014). Glycosylations of different size, type, and polarity interact with the chromophore differently and impact the stability of the anthocyanin uniquely (Zhao et al. 2014). Stability of mono-glycosylated anthocyanins has been reported to follow the order of glucosyl > galactosyl > arabinosyl

(Trošt et al. 2008). The findings of this study generally followed this trend; Cyanidin-3- glucoside was more stable than Cyanidin-3-galactoside at most pH levels, Table 6.

However at pH 3, 4 and 9, Cyanidin-3-galactoside unexpectedly showed greater half- lives. In general, glucose is uniquely, highly, and positively correlated to anthocyanin stability among the common glycosylating units (Zhao et al. 2014). In this study, all anthocyanins evaluated bearing glucose as the first attachment to the aglycone exhibited higher stability than their galactoside counterparts.

117 Generally increasing size of the glycosyl moiety has been reported to further increase stability of anthocyanins, such that di- and tri-saccharide bearing anthocyanins are more stable than monoglycosylated counterparts (Zhao et al. 2014). In this study, this was true in the case of trisaccharide bearing Cyanidin- xylosyl(1→2)glucosyl(1→6)galactoside which generally showed large or the greatest half-lives in the widest pH range, Table 6. Interestingly, di-glycosylation of cyanidin showed opposing effects on stability depending on the 1→2 or 1→6 glycosidic linkages in the sugar moiety. Despite being more prone to hydration, those cyanidin derivatives with 1→6 disaccharides unexpectedly showed greater half-lives than their 1→2 counterparts, Table 6. At many pH, Cyanidin-3-xylosyl(1→2)galactoside showed the poorest stability and smallest half-lives of all the cyanidin derivatives evaluated, even compared to monoglycosylated Cyanidin-3-galactoside. The half-lives of Cyanidin-3-

Glucosyl(1→2)Glucoside and Cyanidin-3-Glucoside are often not significantly different from one another in different pH while the stability of Cyanidin-3- rhamnosyl(1→6)glucoside is usually the greatest, Table 6. The stereochemistry of the glycosylating moiety is demonstrated to have a much larger role on the stability of anthocyanins than simply the size of the substitution.

118 Table 6. Half-lives (hr) of glycosylated cyanidin derivatives in pH 1-9, stored at 25 ˚C in dark. Values reported are means (n = 3) and (standard deviation). Different superscript letters indicate significant differences between anthocyanins in the same pH. R2 ≥ 0.90

Half-life at different pH Anthocyanin 1 2 3 4 5 6 7 8 9 3860bc 4615b 1477b 583ab 1f 11e 13d 11c Cyanidin-3-galactoside (154) (76) (258) (65) N/A (0) (0) (0) (1) 6641a 5438ab 664c 489bc 2c 12d 15f 10d Cyanidin-3-glucoside N/A (533) (605) (40) (34) (0) (1) (0) (0) 2440d 2987c 1349b 270c 5a 4f 14e 13b Cyanidin-3-xylosyl(1à2)galactoside N/A (136) (107) (98) (18) (0) (0) (0) (0) 5870a 4497b 1146bc 369bc 3b 12d 14e 9e Cyanidin-3-glucosyl(1à2)glucoside N/A (233) (130) (129) (33) (0) (0) (0) (0) 6291a 5858a 1672b 745a 1d 15b 21b 10d Cyanidin-3-rhamnosyl(1à6)glucoside N/A (434) (613) (317) (147) (0) (0) (0) (0) 4643b 5291ab 1422b 601ab 1e 14c 17c 12b Cyanidin-3-glucosyl(1à6)galactoside N/A (300) (151) (201) (51) (0) (0) (0) (0) Cyanidin- c ab a ab d a a a 3593 4868 2366 579 N/A 2 19 23 13 xylosyl(1à2)glucosyl(1à6)galactoside (154) (381) (357) (137) (0) (0) (0) (0)

6.4.4. Bisulfite Bleaching

Bisulfite is a common preservative included in many foods and is used extensively in winemaking; therefore, the role of anthocyanin substitution on bisulfite bleaching is of great interest to the food industry (Robinson 2006; Taylor et al. 1986).

Because the isolates have different pKhs, especially relevant when working at pH 3, standardizing by absorbance at vis-max was thought to be a more effective means of benchmarking the overall flavylium cation concentration versus the pH differential method used in the pH stability and half-life section. The number of glycosyls at C3 has previously been reported as increasing the pKh of anthocyanins (Redus et al. 1999).

For all seven anthocyanin isolates, extensive bisulfite bleaching occurred almost instantly (within minutes) and at levels as low as 50 ppm, between 39 and 66% loss in maximum absorbance for the different isolates. Figure 24 shows the spectra of the

119 isolates in response to bisulfite and reveals the extent of bleaching is different among

isolates. The following order was observed for the bisulfite bleaching constant (KSO2):

Galactoside > Glucoside > Glucosyl(1→6)Galactoside > Rhamnosyl(1→6)Glucoside >

Xylosyl(1→2)-Glucosyl(1→6)-Galactoside > Glucosyl(1→2)Glucoside >

Xylosyl(1→2)Galactoside. The rates among pigments were significantly different from one another (p-value < 0.001), and post-hoc tests indicated significant differences

between the specific 1→2 and 1→6 disaccharide bleaching constants (KSO2). Clustered by

size and linkage, KSO2 followed the order: monosaccharides > 1→6 disaccharides > trisaccharide > 1→2 disaccharides. It appeared that glycosidic linkage was important in differentiating the extent of bleaching and that our hypothesis was not rejected.

Figure 24. Bleaching of cyanidin isolates with differing number of glycosylations and patterns in response to bisulfite addition at pH 3. Isolate Name reported alongside % of bleaching in response to 50 ppm (orange arrow) and 200 ppm bisulfite (blue arrow)

120 The order of reactivity (bleaching constant) from this study aligned with previous findings of noticeable differences in carboxypyranocyanidin formation as a result of different 3-glycosylation patterns (Farr et al. 2018). This is not surprising considering bisulfite bleaching is thought to preferentially attack the same site of pyranoanthocyanin formation, Carbon-4. Oliveira and others (2006) evaluated bisulfite bleaching for pyranoanthocyanins with differing glycosylations, but a clear pattern due to glycosylation was not established. Pyranoanthocyanins have a Carbon-4 which is covalently occupied, and the bleaching observed was attributed to reaction with Carbon-2 of the aglycone. It is likely that Carbon-2 accessibility is not modified by different 3-glycosylations in the same manner as Carbon-4 and may be why the differentiation between 1→2 and 1→6 disaccharides is able to be more clearly illustrated with the cyanidin derivatives in this

study. Similar to findings in spectra characteristics and half-life, the KSO2 for Cyanidin-3-

Xylosyl(1→2)Glucosyl(1→6)Galactoside was between the value for the bleaching constant of Glucosyl(1→6)Galactoside and Xylosyl(1→2)Galactoside.

In aqueous solutions as free molecules, disaccharides with 1→2 linkages have been previously reported as having greater intramolecular hydrogen bonding, fewer conformational states, as well as reduced rotational flexibility as compared to 1→6 disaccharides (Pereira et al. 2006; Perić-Hassler et al. 2010). With the findings from this study, it is thought that these chemical properties extend to disaccharides attached to the anthocyanin and alter the stereochemistry, specifically accessibility to Carbon-4, and differentiate the extent of bleaching amongst the 1→2 and 1→6 disaccharides evaluated in this study.

121 6.4.5. Ascorbic Acid Bleaching

Ascorbic acid (AA, vitamin C) is a common fortifying agent in foods and has been previously reported to bleach anthocyanins (Farr. Jacob E. & Giusti 2018; Poei-

Langston & Wrolstad 1981; Garcia-Viguera & Bridle 1999). Understanding the role of 3- glycosylation of anthocyanins in minimizing AA bleaching could be utilized in order to mitigate it. Bleaching induced by ascorbic acid was found over time among between all cyanidin isolates with as little as 300 ppm added. Bleaching was dose dependent and was found to fit first-order kinetics with R2 being 0.94 for isolates containing ascorbic acid.

The half-lives of the pigments with AA were significantly different for all treatment levels (p-value < 0.01). With 300 ppm AA added, half-lives were shortened 60% or more as compared to control treatments, Table 7. The following order of half-lives was observed for anthocyanins with 300 ppm ascorbic acid treatment: Xylosyl(1→2)-

Glucosyl(1→6)-Galactoside > Rhamnosyl(1→6)Glucoside > Galactoside >

Xylosyl(1→2)Galactoside > Glucosyl(1→6)Galactoside > Glucoside >

Glucosyl(1→2)Glucoside. No apparent order was observed as an effect of anthocyanin glycosylation structure by evaluation of half-lives or reaction rates alone, except the tri- substitited cyanidin derivative showed greatest stability against ascorbic acid with a half- life of 77 hours with 300 ppm AA. This was in sharp contrast to bisulfite bleaching and is thought to be a result of the intrinsic instability of each of these isolates over time, a factor not relevant in the near-instantaneous bisulfite-induced bleaching. The stabilities of control samples were over a two-fold difference alone with half-lifes in the following order: Xylosyl(1→2)-Glucosyl(1→6)-Galactoside > Galactoside >

122 Glucosyl(1→6)Galactoside > Rhamnosyl(1→6)Glucoside > Glucoside >

Glucosyl(1→2)Glucoside > Xylosyl(1→2)Galactoside, Table 7. It is apparent by comparison to the stability of controls that these pigments react differently from one another in response to ascorbic acid treatment. The order of stability followed the general trends observed with dilution at pH 3 by concentration; however differences were likely caused by standardizing to equal absorbance rather than concentration. Anthocyanin stability is directly related to concentration (natural log); therefore, increasing the overall quanitity of pigments in solutions will result in modified half-lives.

In order to to consider the impact of the stabilty of each isolate, the data was evaluted for the reaction rate with the AA treatment as a function of the control reaction rate of the pigment. With the k/kcontrol parameter, the following pattern was observed for the glucosyl derived cyanidin pigments and AA: Rhamnosyl(1→6)Glucoside >

Glucoside > > Glucosyl(1→2)Glucoside. For the galactosyl derived pigments in the presence of AA, the following order in reaction rate was observed: Galactoside >

Glucosyl(1→6)Galactoside > Xylosyl(1→2)Glucosyl(1→6)Galactoside > >

Xylosyl(1→2)Galactoside. In both cases, the 1→2 linked disaccharide were found to be significantly more resistant to ascorbic acid induced bleaching compared to the 1→6 disaccharides in terms of k/kcontrol parameter, while the isolates from the pH stability and half-life study were found inversely stable. The restricted stereochemistries of the 1→2 disaccharides are thought to localize the substitution around Carbon-4 and limit the interaction; whereas, the 1→6 disaccharides were thought to have greater rotational

123 freedom and exhibited k/kcontrol values that were more similar to their monosaccharide counterparts.

Cyanidin-3-galactoside appeared to be significantly more reactive (34.9x, 300 ppm AA) compared to its control versus cyanidin-3-glucoside (24.3x, 300 ppm AA) compared to its control (post-hoc p-value < 0.01). Cyanidin-3- xylosyl(1→2)glucosyl(1→2)galactoside had the greatest half-life amongst all treatments at all AA levels, the K/Kcontrol was reported as being intermediate to xylosyl(1→2)galactoside and glucosyl(1→6)galactoside or greater than both.

124 Table 7. Bleaching kinetics of cyanidin-3-glycosylates in response to ascorbic acid at pH 3 over a week (25 C, dark). First-order kinetics applied.

Ascorbic reaction rate 2 Pigment Acid Level -1 k/kcontrol R half life (hours) (ppm) (hours )

0 5.50E-04 1.00 0.981 1259 300 1.34E-02 24.27 0.971 52 Glucoside 600 2.68E-02 48.71 0.988 26 1500 6.29E-02 114.30 0.999 11 3000 9.52E-02 173.03 0.998 7 0 3.46E-04 1.00 0.985 2003 300 1.21E-02 34.92 0.989 57 Galactoside 600 2.31E-02 66.87 0.994 30 1500 5.32E-02 153.75 0.998 13 3000 8.58E-02 248.03 0.998 8 0 7.56E-04 1.00 0.913 916 300 1.27E-02 16.73 0.896 55 Xylosyl(1à2)Galactoside 600 3.32E-02 43.89 0.946 21 1500 9.37E-02 123.88 0.996 7 3000 1.11E-01 146.87 0.975 6 0 6.59E-04 1.00 0.865 1051 300 1.35E-02 20.49 0.959 51 Glucosyl(1à2)Glucoside 600 2.76E-02 41.89 0.972 25 1500 6.96E-02 105.49 0.997 10 3000 9.98E-02 151.32 0.994 7 0 4.38E-04 1.00 0.847 1582 300 1.32E-02 30.15 0.941 52 Glucosyl(1à6)Galactoside 600 3.00E-02 68.53 0.976 23 1500 7.29E-02 166.43 0.999 10 3000 1.05E-01 239.81 0.998 7 0 4.46E-04 1.00 0.951 1553 300 1.19E-02 26.76 0.989 58 Rhamnosyl(1à6)Glucoside 600 2.19E-02 49.17 0.997 32 1500 4.45E-02 99.65 0.999 16 3000 6.80E-02 152.46 0.999 10 0 3.29E-04 1.00 0.950 2105 Xylosyl(1à2)- 300 9.00E-03 27.34 0.969 77 Glucosyl(1à6)-Galactoside 600 1.91E-02 58.16 0.990 36 1500 4.53E-02 137.61 0.998 15 3000 7.06E-02 214.56 0.999 10

125 6.5. Conclusion

Reactivity of mono-, di, and tri-substituted cyanidin-3 isolates were found to be more complex than simply considering the number of sugar units alone. The type of sugar made a difference with Cyanidin-3-galactoside which bleached and hydrated to a greater extent than cyanidin-3-glucoside but had a greater half-life at pH 3, 4, and 9. The position of the glycosidic bond on disaccharides also appeared to have impacted the chemical properties of anthocyanins. The 1→2 disaccharide anthocyanins exhibited higher vis-max, lowered stability in solution during storage but greater resistance to bleaching and hydration than 1→6 disaccharide anthocyanins. This may be the result of intramolecular hydrogen bonding localizing the substitution to the chromophore, specifically around Carbon-4, which contributed to their increased resistance to bleaching and hydration. These conformational modifications may also play a role by inducing greater torsion upon the aglycone structure, which has been suggested to increase the vis- max, while straining the molecule and decreasing its overall stability. The 1→6 disaccharides behaved more similarly to monosaccharides, suggesting less direct interaction with the chromophore; and the trisaccharide often had characteristics intermediate to its 1→2 and 1→6 counterparts with the exception of greater stability. The structure-activity relationship of anthocyanidin glycosylation may be affected by the number of glycosyl units, but the branching patterns of these sugars are also of critical importance to these reactive molecules.

126

Chapter 7. ColorBySpectra: an application to automate conversion of spectral data to color spaces 7.1. Abstract

Color serves key purposes such as evoking emotion and suggesting identity.

Calculation of color parameters involve multi-step processes that require collecting standardized light and object spectra, applying color-matching functions, and a series of steps to derive color parameters. Verbal communication of color is challenging, therefore mathematically-standardized color spaces such as HunterLab, CIEL*c*h* and

CIEL*a*b* are used. ColorBySpectra is software that converts absorbance data into commonly used color spaces. It provides a GUI-oriented solution allowing flexibility to determine color of objects seen under different illuminants using a range of wavelengths and intervals. It reduces time requirements to collect data, relies on absorbance spectra in place of colorimeters, can provide superior color accuracy, and use dynamic sample volume.

127 7.2. Introduction and Background

7.2.1. Background Information on Color

Color is an important parameter for many fields including food science, pharmaceuticals, textiles, medical, marketing, and additional applications. These industries often use color as a quality control metric. Historically, different devices have been designed specifically for measuring color. This has led to the creation of several color spaces. Color vocabulary is not capable of encompassing, quantifying, nor standardizing, the descriptions of all possible colors; therefore, use of standardized color spaces is essential for reproducible communication. Color is a result of three elements: a light source, an object, and an observer; however, there are many additional facets to consider for this three-tuple property (Billmeyer & Saltzman 1981).

Light Source and Illuminant

Objects look different and have distinct color parameters dependent upon the light source. Light sources emit energy over visible ranges, having a unique spectral power distribution (SPD) referred to as illuminants (Minolta 2007). An illuminant can be characterized by an SPD graph, plotting energy across visible range wavelengths accompanied by correlated color temperature (Bedau & Minolta 2015). The International

Commission on Illumination (Commission Internationale I’Eclairage, CIE) has provided two standard illuminants, A and D65, and supplemented these with several secondary illuminants (Schanda n.d.). Standardized illuminants are developed to replicate the most common lighting settings and the L series will be published, which will standardize LED light sources (Commission Internationale de l’Eclairage 2017). With a known SPD of an 128 illuminant, tristimulus values associated to the light source can be derived. All colors humans are capable of perceiving can be reproduced with three primary color

(trichromatic: red, blue, green)(Williamson & Cummins 1983). The tristimulus values from the illuminant reflect the necessary combination of these primaries to emulate the color due to the light source. In , the three illuminants’ SPD graphs and tristimulus values selected for ColorBySpectra are presented. These illuminants were selected because they replicate the three most common lighting conditions that reflect a range of standard settings including noon daylight (D65), incandescent lighting (A), and cool white fluorescence (F2, typical office lighting)(Schanda n.d.). Data for the CIE illuminants in 1 nm intervals used in ColorBySpecta was provided by the Rochester

Institute of Technology and can also be found in CIE’s Technical Report (Rochester

Institute of Technology n.d.)((CIE) 2004).

300 F2 250 A 200

D65

y

g r

e 150

n E 100

50

0 380 480 580 680 780 Wavelength (nm)

Illuminant replicated conditions Xn10 Yn10 Zn10 D65 daylight at midday 94.81 100.00 107.32 A incandescent, tungsten filament 111.14 100.00 35.20 F2 cool white fluorescence - office setting 103.28 100.00 69.03 Figure 25. Common illuminants selected for ColorBySpectra and their tristimulus values and chromaticity coordinates of the light source

129

The Object

An object’s color is not intrinsic(Pascale 2003). Color is the result of the object’s physical and chemical properties. Light strikes an object and it can absorb that energy in a unique way. An object can reflect, transmit, or scatter the light. The resulting reflectance or transmittance over visible ranges is then used in quantifying the color of the object by an observer. To measure the color of an object, the light source selected will strike the object and the result is measured as reflectance or absorbance over visible ranges. Either or those values can be converted to %tranmittance for all visible wavelengths. ColorBySpectra utilizes absorbance values, which are converted to percent transmission (%tranmittance = antilog (2 – absorbance)). Many commercial colorimeters will measure the object in 5 or 10 nm intervals over visible ranges;

ColorBySpectra can perform color calculations down to 1 nm intervals allowing for enhanced color accuracy (Schanda n.d.).

The Observer

After a light source interacts with an object, the resulting signal can be witnessed by an observer. An human observer is thought to rely on three receptors of the eye while a colorimeter would use three filters to replicate all possible colors (Purves D et al. 2001).

The sensitivities of the receptors are modeled by color-matching functions (푥̅, 푦̅, and 푧̅ in

Figure 26). Amongst observers, perceived visual stimulus is variable due to differences in visual abilities (e.g. colorblindness) and capacity (e.g. fatigue, color memory); therefore,

130 a standard observer has been formed to further make color parameters reproducible and

communicable and replicate the average visual capacity of a human observer (X-RIte

1993) . Using the array from the spectral power distribution of the illuminant, the %

transmission of the object, and the color-matching function, tristimulus values of the

object can be derived. Tristimulus values of both the illuminant (denoted with subscript

n10, where n identified the illuminant and 10 refers to the 10 degree observer angle) as

well as the object (subscript 10, denoting the 10 degree observer angle used) can then be

used to determine a spatial location in the CIEL*c*h* and CIEL*a*b* color spaces. A

simplified presentation of the calculation to determine Tristimulus values is shown in

Figure 26.

The tristimulus values for the object being measured are calculated by multiplying the

vector of SPD associated to the illuminant, by the vector resulting from our object’s %

transmittance, and then multiplying that by each of the three color-matching functions in

our standard observer. The resulting integral is the tristimulus value XYZ of the object.

Figure 26. Calculation of Tristimulus values used in calculating Color Spaces

131 Color spaces and their computation

Standardization of color is essential for reproducibility. The use of color spaces has been essential in communicating color data. CIEL*c*h* and CIEL*a*b* are scales based on three-dimensional spaces (solid of color) depicted in

. Hue angle (h*) is a 0 to 360° representation with 0 (and 360 °) being red, 90 °

being yellow, 180 ° is green, and 270 ° is blue. L*, the vertical axis, is lightness, a 0 to

100 scale with 0 being the absence of any light (or black) to 100 being absolute white.

Both CIEL*c*h* and CIEL*a*b* share the same scale of L*. The horizontal axes at the bottom of the solid in

represent the a* and b* axis. a* has both a positive and negative range. Negative a* represents green and scales to positive a* which is red. b* follows a similar pattern with negative b* connected with blue and scaling to positive b* that is yellow. Chroma

132 (c*), at the top of the cylinder, is a scale of the intensity of a color. Zero is the center of the circle, as the distance from the center increase, so does c*. Hue (h*), also at the top of the solid of color, is measured as an angle and differentiates possible saturated colors.

Hunter Lab was historically a more popular color space used. To allow for comparison of new color data to historic datasets, it was included.

+ c* h* 0-360°

- +

b* a*

- + L*

Figure 27. Visual depiction of CIEL*c*h* and CIEL*a*b* color spaces

133 The tristimulus values from the object being measured, depicted in Figure 26, as

well as the tristimulus values associated with the illuminant are then used to calculate the

parameters in the color spaces, transforming abstract output of absorbance spectral data

to an effective color data vector which is interpretable and reproducible. The algorithms

used to do so are shown in Figure 28 and are also provided by CIE (2004). All the

equations for these spaces derive from Tristimulus values and their equations can be

found in Figure 28.

Figure 28. Equations used in calculating color parameters

134 7.2.2. Problem and Target Resolution

To determine an object’s color, one must either purchase specialized equipment or use a spectrophotometer paired with a series of calculations. While it is possible to calculate color through the use of absorbance spectra alone, it is a complex series of calculations that could be easily subjected to human errors as well as tedious to perform one sample or object at a time. A traditional user would be inhibited to do so for larger datasets due to the lack of time and mathematical complexity that is prone to minor errors being compounded. A user would be limited to performing the calculation in software such as Microsoft Excel, which would still be a laborious and sample-by-sample operation. Because of this complexity, many will rely on specialized equipment which performs the necessary calculations in Figure 28. There is a need to automate the mathematical process of converting spectral data to color data, and allow for higher throughput of samples or objects to calculate color parameters. Deriving these parameters in the color spaces would typically require the inquirer to measure all samples, one at a time, and compile the color data generated. With the use of ColorBySpectra, an inquirer could obtain spectrophotometric data, collected in bulk with a device such as a plate reader, use the outputted absorbance data of hundreds of samples collected simultaneously, and convert that to color with ease. Another benefit of pairing the functionality of spectrophotometric data with ColorBySpectra is a user now has the freedom to use any volume of sample that might be required for specialized color measurement equipment. Using the same order of operations and algorithms listed in

Introduction and Background, this software aims to enable researchers to take large

135 datasets of absorbance spectral data and convert to HunterLab, CIEL*c*h*, and

CIEL*a*b* color spaces with incredible ease. It will serve as an effective tool in color computation and enhance the productivity of those who use these important measures.

7.3. Software Architecture, Framework, and Implementation

ColorBySpectra uses Java to be platform agnostic. Using the Swing Java library, a simple graphical user interface has been built that allows a user to import their spectral data (as a .CSV file), select an illuminant, and Translate absorption spectra and save interpretable color vector data. Figure 29 provides an illustrative example of how

ColorBySpectra parses input data and the respective output from it.

Wavelengths Sample #1 … Sample #100 380 0.2989 0.3532 Sample L* a* b* c* h* 381 0.2734 0.3012 Sample #1 382 0.3890 0.2987 … … as many … … … … samples as a … … user decides … … … to add, each … … … … represented … … … by a column … … … … … … … … … 779 .1234 .1233 780 .1230 .1229 Sample #100

input structure of data output structure of data Figure 29. Implemented strategy for parsing input spectral data and producing outputted color parameters

The input of data relies upon an open-source java api called opencsv (Smith &

Conway 2004) The functionality and computation originates with the ColorAnalysis object. It takes the user-provided CSV and initializes the following class variables from it: sample names (String []), wavelengths used (List), absorbance values for all

136 samples (ArrayList>), wavelength interval (Integer), 푥̅, 푦̅, and 푧̅ of the standard observer for the range of user-provided wavelengths (List),

Tristimulus values of the illuminant (Xn10, Yn10, and Zn10, double) and Tristimulus values of the object (X10, Y10, and Z10, double). Several of these variables are reconstructed from the ColorConstants class. The constants used are high precision, going to four or more decimals places.

The top row of the input CSV file is stored as an array of sample names. The first column following is saved in an array of integers, reflecting the user-provided wavelengths. The sample absorbance values are stored in a dynamically sized ArrayList to allow for as many samples as desired by a user, as well as dynamic number of absorbance values used in the final calculation. The dropdown box in the GUI is used to set the illuminant Tristimulus values as well as the spectral power distribution used by the

ColorAnalysis object.

The wavelength array is used to determine the interval (nm step from wavelength to wavelength), and build the standard observer value arrays from constants in

ColorConstants, matching the wavelengths from the input file as well as the interval used.

Each column in the two-dimensional array represents all absorbance or reflectance values for a sample. Iterating across rows of the input, color values are calculated from the absorbance data, the user-selected illuminant, and the appropriate color-matching functions. L*a*b and c*h* are returned as an array of doubles. The data is printed as: sample name + “,” + [all array indices]. The resulting output file is a .CSV that has all the samples and their respective color parameters reflected.

137 In order to guarantee the validity of the color calculations and data, Error capturing checkpoints have been implemented during the parsing and calculation stages.

ColorBySpectra enforces a naming convention for the first column (as present in the template file accompanying the tool) where the first column is named “wavelength

(nm)”. The subsequent wavelength values are checked for even interval spacing between all numbers, and also to guarantee they are in the proper range of 380 to 780 nm, having at minimum visible range of 400 to 700 nm. Guaranteeing an interval of 10 nm or less is essential for color accuracy and users are notified if the data reflects an interval of greater than 10nm. Sanity values of absorbance data is verified to ensure only numeric are present. A popup in the user interface will notify the user of the error encountered and the workflow will reset to receive a new input file.

7.4. Software Functionalities

SpectraToColor serves the purpose of automating the conversion of spectral data

(absorbance) to color spaces. It is designed to allow bulk conversion and for a user to determine HunterLab, CIEL*c*h*, and CIEL*a*b* of as many samples or objects as desired. The program could offer similar functionality to that of a colorimeter with use of absorbance spectra data collected by spectrophotometer, important for those with limited resources, and can further extend the capabilities of a scientist or researcher by allowing rapid calculations. It offers flexibility for color calculations at a range of intervals (less than or equal to 10 nm steps) and specific wavelengths ranges. The use of

ColorBySpectra in combination with a spectrophotometer allows necessary sample volume to become independent of instrument requirements. The selection of D65, A, and 138 F2 provides flexibility in selection of an illuminant to use. Error-checking and -reporting built in allows user to have further confidence in the accuracy of color parameters produced.

7.5. Results

ColorBySpectra has been validated with an array of samples for accuracy in measurement.

7.5.1. Experimental Details

Blue #1, Red #3 and sorghum extract were selected for comparison. A ColorQuest

XE instrument (HunterLab, Reston, VA, USA) with the following settings was used: 2 mm thickness cuvettes, regular transmission, one inch viewing area, 10-degree observer angle, data presented for D65, A, and F2 illuminants. Blue #1 and Red #3 were added to reverse osmosis water and sorghum extract to water acidifed with HCl (0.01%). A

SpectraMax plate reader (Molecular Devices, SunnyVale, Ca, USA) equipped with poly- d-lysine 96 well-microplate (Greiner, Monroe, NC, USA) was also used. A 95 uL aliquot, aligning with the maximum absorbance produced by the ColorQuest XE instrument

(±0.1 Absorbance).

7.5.2. Data Generated

139 ColorBySpectra was found to reliably reproduce results from both the colorimeter color values and also corroborated with data generated from the plate reader spectra, as seen in Figure 30. Color differences between the colorimeter software and the colorimeter spectra and ColorBySpectra, measured by ΔE, were equal to or less than one, anything less than five is considered indistinguishable by the average observer.

CIEL*a*b* as well as the graphical representations of the color data of samples can be found in Figure 30. The use of the plate reader allowed for efficient data collection with the spectra of all samples able to be collected simultaneously. While it is possible to visually compare the color from the plate wells to the 2 mm cuvettes, it should be noted that minimal differences in pathlength (2 mm versus pathlength of 95 uL volume in the

140 well) are likely why differences in CIEL*a*b* are observed. Using the plate reader

resulted in slightly higher variance, likely due to differences in pathlength resulting from

pipetting.

Samples read by Color Measuring instrument

Samples read by UV-Vis plate reader

FD&C 1.0 Blue #1

FD&C 0.8 Red #3

Sorghum extract 0.9

Picture Spectra + Picture of Color data Colorimeter of Plate ColorBySpectra 2 mm from ΔE Spectra + Wells Reproduction cuvette instrument ColorBySpectra

Color parameters obtained using Plate reader color using Color Parameters obtained directly spectral data from Color measuring ColorBySpectra from Instrument for Color Analyses instrument and ColorBySpectra Blue #1 Red #3 Sorghum Blue #1 Red #3 Sorghum Blue #1 Red #3 Sorghum L* 87.3(1.2) 78.9(0.4) 92.8(0.1) 84.3(0.1) 75.2(0.8) 89.5(0.1) 84.2(0.1) 75.1(0.8) 89.6(0.1) a* -31.9(2.3) 60.4(0.3) 9.2(0.2) -29.3(0.2) 58.4(0.2) 9.2(0.1) -29.7(0.3) 58.1(0.3) 8.8(0.2) b* -18.1(1.6) -10.7(0.1) 28.6(0.6) -18.3(0.2) -8.1(0.2) 30.3(0.5) -17.3(0.1) -7.4(0.2) 31.1(0.3)

Figure 30. Visual comparison of color data generated using ColorBySpectra to object images and colorimeter software reproduction

141 7.6. Illustrative Examples

An illustrative example is attached as a video explaining how to use the software.

https://www.youtube.com/embed/1sIoeh8H-x8

7.7. Conclusions

ColorBySpectra was designed to extend the capability of researchers by automating a complex series of calculation and reducing instrumentation to derive color.

The software successfully capitalizes on those goals by allowing users to convert absorbance over visible ranges to HunterLab, CIEL*c*h*, and CIEL*a*b* spaces for many samples or objects with extreme ease. The software allows for selection of three common illuminants D65, A and F2. It was developed using the Java language; this allows for researchers to use the environment they are already familiar with or have available. Error checking and simple GUI design were selected to promote accuracy of color data, user friendliness as well as stability of the software in parsing wavelength and absorbance inputs from a CSV file. The outcome of this software will allow researchers to reduce necessary time and equipment requirements to determine color values, allows dynamic sample volumes, and automates the conversion of large datasets, benefitting researchers in a wide range of fields.

142

Chapter 8. Conclusions and Future Work

Color quality and pigment stability remains a challenge for the use of natural colorants in the food industry. In this work, mitigation strategies for bleaching in response to ascorbic acid were pursued. Green tea catechins initially enhanced color of anthocyanins through copigmentation but played little role in reducing reactivity and color loss in response to ascorbic acid addition. The formation of pyranoanthocyanins and certain stereo-chemical characteristics of anthocyanins played a role in reducing the interaction between pigment and ascorbic acid. Anthocyanins with both C3 and C5 substitution as well as greater number of glycosyl units at C3 alone enhanced stability of the pigment against ascorbic acid. Pyranoanthocyanins showed enhanced color stability against ascorbic acid by formation of colored byproducts of the pigment-ascorbic acid interaction. Anthocyanin extracts containing additional phenols were repeatedly found to possess enhanced stability over their isolated pigment counterparts.

The formation of novel pigments between 5-carboxypyranocyanidin-3-galactoside and ascorbic acid is deserving of future investigation. Additional work to elucidate these structures through one and two-dimensional NMR techniques is of most interest in determining the structure of these novel compounds. A kinetic NMR study was performed to try to directly evaluate the rate and resulting site of interaction between

143 cyanidin-3-galactoside and ascorbic acid with limited success. To successfully elucidate the structure, over 20 mg of anthocyanin in 0.6 mL of acidified deuterated methanol

(CH3OD) is necessary. In addition, an amount of ascorbic acid that is not soluble in the solvent would be necessary to reproduce bleaching. Anthocyanins have high solubility in methanol and lower in water; whereas, ascorbic acid is the inverse. Future attempts at this work could look at mixtures of deuterated methanol and water and minor amounts of deuterated trifluoroacetic acid to stabilize the pigment.

The work involving the stability of different C3-glycosylated anthocyanins provided the most fascinating findings with unexpected differences in the 1→2 and 1→6 glycosyl-branched anthocyanin-3-disaccharides. This work is contrary to generally held observations within the field that have stated increasing number of glycosyl units at C3 results in enhanced pigment stability. This finding alone has spurred several additional projects worth pursuit.

Future work to enhance these findings would focus on the use of circular dichroism and NMR technologies to better understand conformational properties of these pigments. Circular dichroism could be used to investigate chiral properties, hydrogen bonding, and the three dimensional structures of different substituted anthocyanins. In addition, recent progress in the field has allowed for measurement of self-association by

NMR (diffusion ordered spectroscopy (DOSY), specifically). We believe that the differences in the spatial occupation between 1→2 and 1→6 C3 anthocyanin disaccharides causes disruption of planarity (torsion) of the aglycone, with the 1→2 disaccharides being less planar. This could provide a foundation in explaining the

144 remarkable difference in stability, bleaching, and pyranoanthocyanin formation of the

1→2 and 1→6 disaccharide substituted anthocyanins. It could be possible that the disruption of planarity would reduce the extent of pigment self-association and DOSY experiments could be a means of further exemplifying this relationship.

We have shown it is possible to enhance color quality and stability of pigments against ascorbic acid caused color loss. While the mechanism of interaction between pigment and vitamin is not fully understood, the solution of forming derived pigments or selecting anthocyanin sources with certain structural characteristics are ways in which the food industry could improve color quality of products. A connecting theme in all findings seemingly always related to the suspected conformational properties and how derivation or substitution may have influenced the spatial arrangement of the pigment. This understanding could be applied for the use of coloring foods and reducing interaction with ascorbic acid, resulting in better color and nutritional content of foods.

145

References

(CIE), C.I. de l’Eclairage, 2004. The evaluation of whiteness. Colorimetry, 3rd Edition,

552, p.24. Available at: http://div1.cie.co.at/?i_ca_id=551&pubid=23.

Allen, L. et al. eds., 2006. WHO | Guidelines on food fortification with micronutrients 1st

ed., France: World Health Organization. Available at:

http://www.who.int/nutrition/publications/micronutrients/9241594012/en/ [Accessed

November 13, 2015].

Amiot, M.J. et al., 1992. Phenolic Composition and Browning Susceptibility of Various

Apple Cultivars at Maturity. Journal of Food Science, 57(4), pp.958–962.

Andersen, Ø. & Jordheim, M., 2013. Basic Anthocyanin Chemistry and Dietary Sources.

In T. Wallace & M. M. Giusti, eds. Anthocyanin in Health and Disease. Boca Raton,

Florida: CRC Press, pp. 13–90.

Andersen, Ø.M. et al., 2004. Anthocyanin from strawberry (Fragaria ananassa) with the

novel aglycone, 5-carboxypyranopelargonidin. Phytochemistry, 65(4), pp.405–410.

Andersen, Ø.M. & Jordheim, M., 2010. Anthocyanins. In Encyclopedia of Life Sciences.

Chichester, UK: John Wiley & Sons, Ltd. Available at:

http://doi.wiley.com/10.1002/9780470015902.a0001909.pub2 [Accessed March 23,

2017].

Andersen, Ø.M. & Jordheim, M., 2006. The Anthocyanins. In Ø. M. Andersen & K. R. 146 Markham, eds. Flavonoids: Chemistry, Biochemistry, and Applications. Boca Raton,

FL: CRC Press (Taylor & Francis Group), pp. 472–537.

Andres-Lacueva, C. et al., 2005. Anthocyanins in aged -fed rats are found

centrally and may enhance memory. Nutritional Neuroscience, 8(2), pp.111–120.

Available at: https://www.researchgate.net/profile/Cristina_Andres-

Lacueva/publication/7693062_Anthocyanins_in_aged_blueberry-

fed_rats_are_found_centrally_and_may_enhance_memory/links/09e4150ef0557d6c

9f000000/Anthocyanins-in-aged-blueberry-fed-rats-are-found-centrally- [Accessed

October 17, 2017].

Andrew L., W., 2001. Determination of Total Phenolics. In R. E. Wrolstad et al., eds.

Current Protocols in Food Analytical Chemistry. Hoboken, NJ, USA: John Wiley &

Sons, Inc., pp. 463–469.

Arnold, L.E., Lofthouse, N. & Hurt, E., 2012. Artificial food colors and attention-

deficit/hyperactivity symptoms: conclusions to dye for. Neurotherapeutics : the

journal of the American Society for Experimental NeuroTherapeutics, 9(3), pp.599–

609. Available at:

http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=3441937&tool=pmcentr

ez&rendertype=abstract [Accessed September 5, 2015].

Asen, S., Stewart, R.N. & Norris, K.H., 1972. Co-pigmentation of anthocyanins in plant

tissues and its effect on color. Phytochemistry, 11(3), pp.1139–1144.

Bakker, J. & Timberlake, C.F., 1997. Isolation , Identification , and Characterization of

New Color-Stable Anthocyanins Occurring in Some Red Wines. Journal of

147 agricultural and food chemistry, 8561(96), pp.35–43.

Bedau, A. & Minolta, K., 2015. The language of light. www. konicaminolta.

com/instruments/knowledge/light/pdf/language_of_light. pdf, 8(6), pp.11–13.

Belitz, H.D., Grosch, W. & Schieberle, P., 2009. Food Additives. In Food Chemistry.

Berlin, Heidelberg: Springer Berlin Heidelberg, pp. 429–466. Available at:

https://doi.org/10.1007/978-3-540-69934-7_9.

Berké, B. et al., 1998. Bisulfite addition to anthocyanins: revisited structures of

colourless adducts. Tetrahedron Letters, 39(32), pp.5771–5774.

Billmeyer, F.W. & Saltzman, M., 1981. Principles of color technology, New York:

Wiley.

Blanco-Vega, D. et al., 2011. HPLC-DAD-ESI-MS/MS characterization of

pyranoanthocyanins pigments formed in model wine. Journal of Agricultural and

Food Chemistry, 59(17), pp.9523–9531.

Bonerz, D. et al., 2007. Analytical characterization and the impact of ageing on

anthocyanin composition and degradation in juices from five sour cherry cultivars.

European Food Research and Technology, 224(3), pp.355–364.

Boulton, R., 2001. The copigmentation of anthocyanins and its role in the color of red

wine: A critical review. American Journal of Enology and Viticulture, 52(2), pp.67–

87.

Brenes, C.H., Del Pozo-Insfran, D. & Talcott, S.T., 2005. Stability of copigmented

anthocyanins and ascorbic acid in a grape juice model system. Journal of

Agricultural and Food Chemistry, 53(1), pp.49–56.

148 Brouillard, R. et al., 1991. ChemInform Abstract: pH and Solvent Effects on the

Copigmentation Reaction of with Polyphenols, Purine and Pyrimidine

Derivatives. J. Chem. Soc. Perkin Trans. II, 22(45), pp.1235–1241. Available at:

http://doi.wiley.com/10.1002/chin.199145035.

Brouillard, R. et al., 1989. The co-pigmentation reaction of anthocyanins: a microprobe

for the structural study of aqueous solutions. Journal of the American Chemical

Society, 111(7), pp.2604–2610. Available at: http://dx.doi.org/10.1021/ja00189a039

[Accessed November 17, 2015].

Brouillard, R., 1990. The hemiacetal-cis-chalcone equilibrium of malvin, a natural

anthocyanin. Canadian journal of chemistry, 68(5), pp.755–761.

Buettner, G.R. & Jurkiewicz, B.A., 1996. Catalytic metals, ascorbate and free radicals:

combinations to avoid. Radiat Res, 145(5), pp.532–541.

Burrows, J.D., A., 2009. Palette of Our Palates: A Brief History of Food Coloring and Its

Regulation. Comprehensive Reviews in Food Science and Food Safety, 8(4),

pp.394–408. Available at: http://doi.wiley.com/10.1111/j.1541-4337.2009.00089.x

[Accessed October 24, 2016].

Cabrita, L., Fossen, T. & Andersen, Ø.M., 2000. Colour and stability of the six common

anthocyanidin 3-glucosides in aqueous solutions. Food Chemistry, 68(1), pp.101–

107.

Cai, Y., Lilley, T.H. & Haslam, E., 1990. Polyphenol-anthocyanin copigmentation.

Journal of the Chemical Society, Chemical Communications, (5), pp.380–383.

Cai, Y.Z., Sun, M. & Corke, H., 2005. Characterization and application of betalain

149 pigments from plants of the Amaranthaceae. Trends in Food Science and

Technology, 16(9), pp.370–376.

Castañeda-Ovando, A. et al., 2009. Chemical studies of anthocyanins: A review. Food

Chemistry, 113(4), pp.859–871. Available at:

http://dx.doi.org/10.1016/j.foodchem.2008.09.001.

Cavalcanti, R.N., Santos, D.T. & Meireles, M.A.A., 2011. Non-thermal stabilization

mechanisms of anthocyanins in model and food systems-An overview. Food

Research International, 44(2), pp.499–509. Available at:

http://dx.doi.org/10.1016/j.foodres.2010.12.007.

Chebrolu, K.K. et al., 2012. An improved sample preparation method for quantification

of ascorbic acid and dehydroascorbic acid by HPLC. LWT - Food Science and

Technology, 47(2), pp.443–449. Available at:

http://dx.doi.org/10.1016/j.lwt.2012.02.004.

Cheeseman, M. & Ph, D., 2011. Food Advisory Committee Certified Color Additives and

Childhood Hyperactivity.

Chen, L. & Hrazdina, G., 1981. Structural aspects of anthocyanin- complex

formation and its role in plant color. Phytochemistry, 20, pp.297–303. Available at:

http://www.sciencedirect.com/science/article/pii/0031942281851114.

Chen, Z. et al., 1998. Stabilizing Effect of Ascorbic Acid on Green Tea Catechins. ,

8561(97), pp.2512–2516.

Chen, Z.-Y. et al., 1998. Stabilizing Effect of Ascorbic Acid on Green Tea Catechins.

Journal of Agricultural and Food Chemistry, 46(7), pp.2512–2516. Available at:

150 http://pubs.acs.org/doi/abs/10.1021/jf971022g.

Choe, E. & Min, D.B., 2009. Mechanisms of Antioxidants in the Oxidation of Foods.

Comprehensive Reviews in Food Science and Food Safety, 8(4), pp.345–358.

Chung, C. et al., 2016. Stabilization of natural colors and nutraceuticals: Inhibition of

anthocyanin degradation in model beverages using polyphenols. Food Chemistry,

212, pp.596–603.

CIE, 2004. CIE 015:2004 Colorimetry, Available at:

http://div1.cie.co.at/?i_ca_id=551&pubid=23.

Commission Internationale de l’Eclairage, 2017. CIE 2017 Midterm Meeting. In Jeju

Island, Republic of Korea.

Dangles, O., Saito, N. & Brouillard, R., 1993. Anthocyanin intramolecular copigment

effect. Phytochemistry, 34(1), pp.119–124.

Dangles, O., Saito, N. & Brouillard ’, R., 1993. Kinetic and Thermodynamic Control of

Flavylium Hydration in the -Cinnamic Acid Complexation. Origin of

the Extraordinary Flower Color Diversity of Pharbitis nil. J . Am. Chem. SOC,

115(3), pp.125–3.

Danilewicz, J.C. & Wallbridge, P.J., 2010. Further Studies on the Mechanism of

Interaction of Polyphenols, Oxygen, and Sulfite in Wine. American Journal of

Enology and Viticulture, 61(2), p.166 LP-175. Available at:

http://www.ajevonline.org/content/61/2/166.abstract.

Davies, a. J., Mazza, G. & Canada, A., 1993. Copigmentation of Simple and Acylated

Anthocyanins with Colorless Phenolic Compounds. Journal of Agricultural and

151 Food Chemistry, 41(5), pp.716–720.

Dewey, C., 2017. Consumers loved “all-natural” — until Trix cereal lost its neon-bright

glow. The Washington Post. Available at:

https://www.washingtonpost.com/news/wonk/wp/2017/09/23/consumers-loved-all-

natural-until-trix-cereal-lost-its-neon-bright-glow/?utm_term=.931d710b8e58

[Accessed January 5, 2018].

Domonoske, C., 2018. Time To Make The Doughnuts Free Of Artificial Dyes, Dunkin’

Decides : The Two-Way : NPR. the two-way: breaking news from NPR. Available

at: https://www.npr.org/sections/thetwo-way/2018/01/05/576054883/time-to-make-

the-donuts-free-of-artificial-dyes-dunkin-decides [Accessed March 5, 2018].

Du, Q., Zheng, J. & Xu, Y., 2008. Composition of anthocyanins in mulberry and their

antioxidant activity. Journal of Food Composition and Analysis, 21(5), pp.390–395.

Durst, R.W. & Wrolstad, R.E., 2005. Separation and Characterization of Anthocyanins

by HPLC. In Handbook of Food Analytical Chemistry. pp. 33–45.

Eiro, M.J. & Heinonen, M., 2002. Anthocyanin color behavior and stability during

storage: Effect of intermolecular copigmentation. Journal of Agricultural and Food

Chemistry, 50(25), pp.7461–7466.

Eison-Perchonok, M.H. & Downes, T.W., 1982. Kinetics of ascorbic acid autoxidation as

a function of dissolved oxygen concentration and temperature. Food Science, 47,

pp.765–767.

Eldahshan, O.A., Nasser, A. & Singab, B., 2013. Carotenoids. Journal of Pharmacognosy

and Phytochemistry, 8192(1), pp.2668735–5. Available at: www.phytojournal.com.

152 Epley, R.J., Addis, P.B. & Warthesen, J.J., 1992. Nitrite in Meat. Minnesota extension

service, pp.2–3.

Eugster, C.H. & Märki‐Fischer, E., 1991. The Chemistry of Rose Pigments. Angewandte

Chemie International Edition in English, 30(6), pp.654–672.

Fan-Chiang, H.-J. & Wrolstad, R.E., 2005. Anthocyanin Pigment Composition of

Blackberries. Journal of Food Science, 70(3), pp.198–202.

Farr. Jacob E. & Giusti, M.M., 2018. Investigating the interaction of ascorbic acid with

anthocyanins and pyranoanthocyanins. Submitted to Molecules.

Farr, J.E. & Giusti, M.M., 2017. ColorBySpectra. Available at:

https://buckeyevault.com/products/colorbyspectra [Accessed October 18, 2017].

Farr, J.E., Sigurdson, G.T. & Giusti, M.M., 2018. Influence of cyanidin glycosylation

patterns on carboxypyranoanthocyanin formation. Submitted to Food Chemistry.

FDA, 2018a. Code of Federal Regulations: Title 21, Chapter 1, Subpart A, Part 74 -

LISTING OF COLOR ADDITIVES SUBJECT TO CERTIFICATION. Code of

Federal Regulations. Available at: https://www.ecfr.gov/cgi-

bin/retrieveECFR?gp=&SID=3463c48f55ae08efd099682901bb9500&r=PART&n=

pt21.1.74 [Accessed March 5, 2018].

FDA, 2016. Listing of Color Additives Exempt from Certification. 1st April, p.Title 21,

Volume 1. Sec. 73.40, 73.350, 73.575, 73. Available at:

http://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfcfr/cfrsearch.cfm?cfrpart=73&s

howfr=1.

FDA, 2018b. PART 71—COLOR ADDITIVE PETITIONS. Code of Federal

153 Regulations. Available at: https://www.ecfr.gov/cgi-

bin/retrieveECFR?gp=&SID=3463c48f55ae08efd099682901bb9500&r=PART&n=

pt21.1.71 [Accessed March 5, 2018].

Fernandes, A. et al., 2017. A review of the current knowledge of red wine colour. OENO

One, 51(1), pp.1–21.

De Ferrars, R.M. et al., 2014. The pharmacokinetics of anthocyanins and their

metabolites in humans. British Journal of Pharmacology, 171(13), pp.3268–3282.

Fossen, T. & Andersen, Ø.M., 2003. Anthocyanins from red onion, Allium cepa, with

novel aglycone. Phytochemistry, 62(8), pp.1217–1220.

De Freitas, V. & Mateus, N., 2011. Formation of pyranoanthocyanins in red wines: A

new and diverse class of anthocyanin derivatives. Analytical and Bioanalytical

Chemistry, 401(5), pp.1467–1477. Available at:

http://link.springer.com/10.1007/s00216-010-4479-9 [Accessed January 4, 2017].

Fulcrand, H. et al., 1998. A new class of wine pigments generated by reaction between

pyruvic acid and grape anthocyanins. Phytochemistry, 47(7), pp.1401–1407.

García-Viguera, C. et al., 1998. Colour and anthocyanin stability of red raspberry jam.

Journal of the Science of Food and Agriculture, 78(4), pp.565–573.

Garcia-Viguera, C. & Bridle, P., 1999. Influence of structure on color stability of

anthocyanins and flavylium salts. Food Chem., 64(1), p.21.

Giusti, M.M. et al., Electrospray and Tandem Mass Spectroscopy As Tools for

Anthocyanin Characterization. Available at:

http://pubs.acs.org/doi/pdf/10.1021/jf981242%2B [Accessed August 17, 2017].

154 Giusti, M.M. & Wallace, T., 2013. Anthocyanins in Health and Disease 1st ed., CRC

Press.

Giusti, M.M. & Wrolstad, R., 2005. Characterization and measurement of anthocyanins

by UV- Visible spectroscopy. Current Protocols in Food Analytical Chemistry,

(August 2016), pp.19–31. Available at:

http://doi.wiley.com/10.1002/0471142913.faf0102s00.

Giusti, M.M. & Wrolstad, R.E., 2003. Acylated anthocyanins from edible sources and

their applications in food systems. Biochemical Engineering Journal, 14(3), pp.217–

225. Available at: http://www.sciencedirect.com.proxy.lib.ohio-

state.edu/science/article/pii/S1369703X02002218 [Accessed April 24, 2017].

Giusti, M.M. & Wrolstad, R.E., 2001. Characterization and Measurement of

Anthocyanins by UV-visible Spectroscopy. In Current Protocols in Food Analytical

Chemistry. pp. 19–31. Available at:

http://doi.wiley.com/10.1002/0471142913.faf0102s00.

Giusti, M.M. & Wrolstad, R.E., 1996. Characterization of Red Radish Anthocyanins.

Journal of Food Science, 61(2), pp.322–326. Available at:

http://onlinelibrary.wiley.com.proxy.lib.ohio-state.edu/store/10.1111/j.1365-

2621.1996.tb14186.x/asset/j.1365-

2621.1996.tb14186.x.pdf?v=1&t=j9ribdop&s=4440e7894d890c38a96d4b1d98030a

bd12ce30f2 [Accessed November 8, 2017].

Gómez-Alonso, S. et al., 2012. Synthesis, isolation, structure elucidation, and color

properties of 10-Acetyl-pyranoanthocyanins. Journal of Agricultural and Food

155 Chemistry, 60(49), pp.12210–12223.

Goto, T. & Kondo, T., 1991. Structure and Molecular Stacking of Anthocyanins - Flower

Color Variation. Angewandte Chemie-international Edition In English, 30(1),

pp.17–33.

Graham, H.N., 1992. Green tea composition, consumption, and polyphenol chemistry.

Preventive Medicine, 21(3), pp.334–350.

Gu, L. et al., 2002. Fractionation of polymeric procyanidins from lowbush blueberry and

quantification of procyanidins in selected foods with an optimized normal-phase

HPLC-MS fluorescent detection method. Journal of Agricultural and Food

Chemistry, 50(17), pp.4852–4860.

Häkkinen, S.H. et al., 1999. Content of the Flavonols Quercetin, Myricetin, and

Kaempferol in 25 Edible Berries. Journal of Agricultural and Food Chemistry,

47(6), pp.2274–2279. Available at: http://dx.doi.org/10.1021/jf9811065.

Harborne, J.B. & Hall, E., 1964. Plant Polyphenols-XIII. The Systematic Distribution and

Origin of Anthocyanins Containing Branched Trisaccharides. Phytochemistry, 3(3),

pp.453–463. Available at:

http://linkinghub.elsevier.com/retrieve/pii/S0031942200836304.

Harp, B.P. & Barrows, J.N., 2015. US regulation of color additives in foods, Elsevier Ltd.

Available at: http://dx.doi.org/10.1016/B978-1-78242-011-8.00004-0.

He, F. et al., 2012. Anthocyanins and their variation in red wines II. Anthocyanin derived

pigments and their color evolution. Molecules, 17(2), pp.1483–1519.

He, J. et al., 2010. Spectral Features and Stability of Oligomeric Pyranoanthocyanin-

156 flavanol Pigments Isolated from Red Wines. J. Agric. Food Chem, 58, pp.9249–

9258.

Henry, B.S., 1996. Natural food colours. In G. A. F. Hendry & J. D. Houghton, eds.

Natural Food Colorants. Boston, MA: Springer US, pp. 40–79. Available at:

https://doi.org/10.1007/978-1-4615-2155-6_2.

Hons, a F., Asdc, C. & Officer, T., 1994. Perkin’s mauve: the history of the chemistry.

Textile Chemicals, 15(1984), pp.1–5.

Hoshino, T., Matsumoto, U. & Goto, T., 1980. Evidences of the self-association of

anthocyanins I. Circular Dichroism of cyanin anhydrobase. Tetrahedron Letters,

21(18), pp.1751–1754.

Hou, D.-X., 2003. Potential Mechanisms of Cancer Chemoprevention by Anthocyanins.

Current Molecular Medicine, 3(2), pp.149–159. Available at:

http://www.eurekaselect.com/openurl/content.php?genre=article&issn=1566-

5240&volume=3&issue=2&spage=149.

Hsiao, H.-C., 2014. Anthocyanin color enhancement by using catechin as copigments and

stability during storage thesis.

Humphrey, A.M., 2006. Chlorophyll as a Color and Functional Ingredient. Journal of

Food Science, 69(5), pp.C422–C425. Available at:

http://doi.wiley.com/10.1111/j.1365-2621.2004.tb10710.x.

Hunt, K., 2015. General Mills commits to removing artificial flavors and colors from

artificial sources, from cereal | A Taste of General Mills. Available at:

http://www.blog.generalmills.com/2015/06/a-big-commitment-for-big-g-cereal/

157 [Accessed March 23, 2017].

Ichiyanagi, T. et al., 2001. Acid mediated hydrolysis of blueberry anthocyanins.

Chemical & pharmaceutical bulletin, 49(1), pp.114–117.

Jansson, P.J. et al., 2004. Oxidative decomposition of vitamin C in drinking water. Free

Radical Research, 38(8), pp.855–860.

Jing, P. et al., 2008. Structure−Function Relationships of Anthocyanins from Various

Anthocyanin-Rich Extracts on the Inhibition of Colon Cancer Cell Growth. Journal

of Agricultural and Food Chemistry, 56(20), pp.9391–9398. Available at:

http://pubs.acs.org/doi/abs/10.1021/jf8005917 [Accessed April 20, 2017].

Jordheim, M. et al., 2007. Molar absorptivities and reducing capacity of

pyranoanthocyanins and other anthocyanins. Journal of Agricultural and Food

Chemistry, 55(26), pp.10591–10598.

Jurd, L., 1964. Reactions involved in sulfite bleaching of anthocyanins. Journal of Food

Science, 29, pp.16–19.

Jurd, L., 1967. THE CONSTITUTION OF A BLUE CORNFLOWER CRYSTALLINE,

PIGMENT. , 6.

Kay, C.D., Mazza, G.J. & Holub, B.J., 2005. Anthocyanins exist in the circulation

primarily as metabolites in adult men. The Journal of nutrition, 135(11), pp.2582–

2588.

Khan, M.I., 2016. Stabilization of betalains: A review. Food Chemistry, 197, pp.1280–

1285.

Khan, M.M.T. & Martell, A.E., 1967. Metal Ion and Metal Chelate Catalyzed Oxidation

158 of Ascorbic Acid. Journal of the American Chemical Society, 89(16), pp.4176–

4185. Available at: http://pubs.acs.org/doi/pdf/10.1021/ja00992a036 [Accessed

November 9, 2017].

Kobylewski, S. & Jacobson, M.F., 2010. Food Dyes: A Rainbow of Risks. Decision

Sciences, 30(2), pp.337–360.

Krikorian, R. et al., 2010. Blueberry Supplementation Improves Memory in Older Adults.

J Agric Food Chem, 58(7), pp.3996–4000.

Kulling, S.E. & Rawel, H.M., 2008. Chokeberry (Aronia melanocarpa) - A review on the

characteristic components and potential health effects. Planta Medica, 74(13),

pp.1625–1634.

Lambert, S.G. et al., 2011. Copigmentation between malvidin-3-glucoside and some wine

constituents and its importance to colour expression in red wine. Food Chemistry,

125, pp.106–115. Available at: https://ac.els-cdn.com/S0308814610010484/1-s2.0-

S0308814610010484-main.pdf?_tid=8bcde930-cbb9-11e7-b399-

00000aab0f6b&acdnat=1510938523_0c10bc354a6baf6052dcb3f6064e7c73

[Accessed November 17, 2017].

Lehto, S. et al., 2017. Comparison of food colour regulations in the EU and the US: a

review of current provisions. Food Additives & Contaminants: Part A, 34(3),

pp.335–355. Available at: https://doi.org/10.1080/19440049.2016.1274431.

Li, D. et al., 2017. Critical Reviews in Food Science and Nutrition Health benefits of

anthocyanins and molecular mechanisms: Update from recent decade Health

benefits of anthocyanins and molecular mechanisms: Update from recent decade.

159 Available at:

http://www.tandfonline.com/action/journalInformation?journalCode=bfsn20

[Accessed November 6, 2017].

Lu, Y. & Foo, L.Y., 2001. Unusual anthocyanin reaction with acetone leading to

pyranoanthocyanin formation. Tetrahedron Letters, 42(7), pp.1371–1373.

Malcıoğlu, O.B. et al., 2011. Dielectric and Thermal Effects on the Optical Properties of

Natural Dyes: A Case Study on Solvated Cyanin. Journal of the American Chemical

Society, 133(39), pp.15425–15433.

Manach, C., 2004. Polyphenols : food sources and bioavailability . Am J Clin Nutr.

American Journal of Clinical Nutrition, 79(October 2015), pp.727–747.

Mangels, A.R. et al., 2018. Carotenoid content of fruits and vegetables: An evaluation of

analytic data. Journal of the American Dietetic Association, 93(3), pp.284–296.

Available at: http://dx.doi.org/10.1016/0002-8223(93)91553-3.

Markakis, P., 1982. Stability of Anthocyanins in Foods. In P. Markakis, ed. Anthocyanins

as Food Colors. Elsevier Science, p. 280. Available at:

https://books.google.com/books?hl=en&lr=&id=SN8XbvA1w8gC&oi=fnd&pg=PP

1&dq=Anthocyanins+as+food+additives+markakis&ots=bfmkCpG1jX&sig=8lTjO

2EMSeDotG6uAOqMIlrEcvM#v=onepage&q=Anthocyanins as food additives

markakis&f=false [Accessed April 20, 2017].

Markakis, P. & Jurd, L., 1974. Anthocyanins and their stability in foods. C R C Critical

Reviews in Food Technology, 4(4), pp.437–456. Available at:

http://www.tandfonline.com.ezproxy.library.dal.ca/doi/abs/10.1080/1040839740952

160 7165.

Martins, N. et al., 2016. Food colorants: Challenges, opportunities and current desires of

agro-industries to ensure consumer expectations and regulatory practices. Trends in

Food Science and Technology, 52, pp.1–15.

Mazza, G. & Brouillard, R., 1987. Color stability and structural transformations of

cyanidin 3, 5-diglucoside and four 3-deoxyanthocyanins in aqueous solutions.

Journal of Agricultural and Food Chemistry, 35, pp.422–426.

McCann, D. et al., 2007. Food additives and hyperactive behaviour in 3-year-old and 8/9-

year-old children in the community: a randomised, double-blinded, placebo-

controlled trial. Lancet, 370(9598), pp.1560–1567.

Minolta, K., 2007. Precise color communication: color control from perception to

instrumentation,

Moerman, D.E., 2010. Native American Food Plants: An Ethnobotanical Dictionary,

Timber Press. Available at: https://books.google.com/books?id=iYhjlKR7GZEC.

Mokrzycki, W.S. & Tatol, M., 2012. Colour difference ∆ E - A survey. Machine Graphic

and Vision, pp.1–28.

Montilla, E.C. et al., 2011. Anthocyanin composition of black carrot (Daucus carota ssp.

sativus var. atrorubens Alef.) Cultivars antonina, beta sweet, deep purple, and purple

haze. Journal of Agricultural and Food Chemistry, 59(7), pp.3385–3390.

Morata, A. et al., 2006. Effects of pH, temperature and SO2 on the formation of

pyranoanthocyanins during red wine fermentation with two species of

Saccharomyces. International Journal of Food Microbiology, 106(2), pp.123–129.

161 Morrot, G., Brochet, F. & Dubourdieu, D., 2001. The Color of Odors. Brain and

Language, 79, pp.309–320. Available at: http://www.idealibrary.com [Accessed

October 21, 2016].

Mudd, J., 2016. Mars, Incorporated to remove all artificial colors from its human food

portfolio. Available at: http://www.prnewswire.com/news-releases/mars-

incorporated-to-remove-all-artificial-colors-from-its-human-food-portfolio-

300216158.html [Accessed March 23, 2017].

Murray, K., 2017. Enabling innovation: Lessons from Crystal Pepsi. The Conversation.

Available at: http://theconversation.com/enabling-innovation-lessons-from-crystal-

pepsi-84967 [Accessed February 15, 2018].

Nasri, S., 2012. Chronic Cyanidin-3-glucoside Administration Improves Short-term

Spatial Recognition Memory but not Passive Avoidance Learning and Memory in

Streptozotocin-diabetic Rats. Phytotherapy Research, 26(8), pp.1205–1210.

National Public Radio, 2015. ICONIC KRAFT MACARONI & CHEESE TO REMOVE

SYNTHETIC COLORS AND ARTIFICIAL PRESERVATIVES IN THE U.S. IN

2016. https://www.prnewswire.com/news-releases/iconic-kraft-macaroni--cheese-to-

remove-synthetic-colors-and-artificial-preservatives-in-the-us-in-2016-

300068396.html. Available at: https://www.prnewswire.com/news-releases/iconic-

kraft-macaroni--cheese-to-remove-synthetic-colors-and-artificial-preservatives-in-

the-us-in-2016-300068396.html [Accessed March 5, 2017].

Nursten, H.E., 2005. Nonenzymic Browning Mainly Due to Ascorbic Acid. In The

Maillard reaction : chemistry, biochemistry, and implications. Cambridge, UK:

162 Royal Society of Chemistry. Available at:

http://www.books24x7.com/marc.asp?bookid=14424.

Oliveira, J. et al., 2006. Color properties of four cyanidin-pyruvic acid adducts. Journal

of agricultural and food chemistry, 54(18), pp.6894–903. Available at:

http://pubs.acs.org.proxy.lib.ohio-state.edu/doi/abs/10.1021/jf061085b [Accessed

November 17, 2015].

Oliveira, J. et al., 2009. Equilibrium Forms of Vitisin B Pigments in an Aqueous System

Studied by NMR and Visible Spectroscopy. The Journal of Physical Chemistry B,

113(32), pp.11352–11358. Available at:

http://pubs.acs.org/doi/pdf/10.1021/jp904776k [Accessed May 11, 2017].

Oliveira, J., Mateus, N. & De Freitas, V., 2014. Previous and recent advances in

pyranoanthocyanins equilibria in aqueous solution. Dyes and Pigments, 100(1),

pp.190–200. Available at: http://dx.doi.org/10.1016/j.dyepig.2013.09.009.

Oszmianski, J. & Sapis, J.C., 1988. Anthocyanins in Fruits of Aronia Melanocarpa

(Chkeberry). Journal of Food Science, 53(4), pp.1241–1242.

Ozkan, M. et al., 2002. Degradation Kinetics of Anthocyanins from Sour Cherry,

Pomegranate, and Strawberry Juices by Hydrogen Peroxide. Journal of Food

Science, 67(2), pp.525–529. Available at: http://doi.wiley.com/10.1111/j.1365-

2621.2002.tb10631.x [Accessed March 21, 2017].

Pascale, D., 2003. A review of RGB color spaces from xyY to R’G’B’, Available at:

http://scholar.google.com/scholar?hl=en&btnG=Search&q=intitle:A+Review+of+R

GB+color+spaces...+from+xyY+to+R’G’B#0.

163 Pereira, C.S. et al., 2006. Conformational and Dynamical Properties of Disaccharides in

Water: a Molecular Dynamics Study. Biophysical Journal, 90(12), pp.4337–4344.

Available at: http://linkinghub.elsevier.com/retrieve/pii/S0006349506726121.

Perić-Hassler, L. et al., 2010. Conformational properties of glucose-based disaccharides

investigated using molecular dynamics simulations with local elevation umbrella

sampling. Carbohydrate Research, 345(12), pp.1781–1801.

Pesek, C.A. & Warthesen, J.J., 1987. Photodegradation of Carotenoids in a Vegetable

Juice System. Journal of Food Science, 52(3), pp.744–746.

Poei-Langston, M.S. & Wrolstad, R.E., 1981. Color Degradation in an Ascorbic Acid-

Anthocyanin-Flavanol Model System. Journal of Food Science, 46(4), pp.1218–

1236. Available at: http://doi.wiley.com/10.1111/j.1365-2621.1981.tb03026.x

[Accessed November 3, 2015].

Purves D, A., GJ, F.D. & et al, 2001. Vision: The Eye. In Neuroscience. Sunderland,

MA: Sinauer Associates. Available at:

https://www.ncbi.nlm.nih.gov/books/NBK11074/.

Quaglieri, C. et al., 2017. Updated knowledge about pyranoanthocyanins: Impact of

oxygen on their contents, and contribution in the winemaking process to overall

. Trends in Food Science and Technology, 67, pp.139–149.

Redus, M., Baker, D.C. & Dougall, D.K., 1999. Rate and equilibrium constants for the

dehydration and deprotonation reactions of some monoacylated and glycosylated

cyanidin derivatives. Journal of Agricultural and Food Chemistry, 47(8), pp.3449–

3454.

164 Rein, M.J., 2005. Copigmentation reactions and color stability of berry anthocyanins.

University of Helsinki Department.

Rein, M.J. et al., 2005. Identification of novel pyranoanthocyanins in berry juices.

European Food Research and Technology, 220(3–4), pp.239–244.

Rentzsch, M., Schwarz, M. & Winterhalter, P., 2007. Pyranoanthocyanins - an overview

on structures, occurrence, and pathways of formation. Trends in Food Science and

Technology, 18(10), pp.526–534. Available at: http://ac.els-

cdn.com/S0924224407001458/1-s2.0-S0924224407001458-

main.pdf?_tid=96d9a3e2-8815-11e7-85cc-

00000aacb35e&acdnat=1503501376_7dca565efd1b5bbf09b46f3676c0213b

[Accessed August 23, 2017].

Robinson, J., 2006. The Oxford companion to wine. The Oxford companion to wine 3rd

Edn, 1, p.840.

Rochester Institute of Technology, Rochester Institute of Technology Color Science.

Resources: Useful Color data. Available at: http://www.rit-

mcsl.org/UsefulData/D65_and_A.htm [Accessed May 18, 2017].

Rodríguez-Pérez, C. et al., 2015. Assessment of the stability of proanthocyanidins and

other phenolic compounds in cranberry syrup after gamma-irradiation treatment and

during storage. Food Chemistry, 174, pp.392–399.

Rodríguez-Saona, L.E. et al., 2001. Extraction, Isolation, and Purification of

Anthocyanins. In R. E. Wrolstad et al., eds. Current Protocols in Food Analytical

Chemistry. Hoboken, NJ, USA, NY: John Wiley & Sons, p. F1.1-F1.11. Available

165 at: http://doi.wiley.com/10.1002/0471142913.faf0101s00.

Rowe, A.P., Miller, L.M. & Meisch, L.A., 2007. Weaving and dyeing in highland

ecuador, Available at: http://www.scopus.com/inward/record.url?eid=2-s2.0-

84898356986&partnerID=tZOtx3y1.

Salisburg, A.M. et al., 2009. Ramachandran-type plots for glycosidic linkages: Examples

from molecular dynamic simulations using the glycam06 force field. Journal of

Computational Chemistry, 30(6), pp.910–921.

Santos-Buelga, C. & Scalbert, A., 2000. Proanthocyanidins and tannin-like compounds -

Nature, occurrence, dietary intake and effects on nutrition and health. Journal of the

Science of Food and Agriculture, 80(7), pp.1094–1117.

Sawant, L., Prabhakar, B. & Pandita, N., 2010. Quantitative HPLC Analysis of Ascorbic

Acid and Gallic Acid in Phyllanthus Emblica. Journal of Analytical & Bioanalytical

Techniques, 1(3), pp.3–6. Available at: https://www.omicsonline.org/2155-

9872/2155-9872-1-111.digital/2155-9872-1-111.html.

Scalzo, R. Lo et al., 2008. Anthocyanin composition of cauliflower (Brassica oleracea L.

var. botrytis) and cabbage (B. oleracea L. var. capitata) and its stability in relation to

thermal treatments. Food Chemistry, 107(1), pp.136–144.

Schanda, J., CIE Colorimetry. In Colorimetry. Hoboken, NJ, USA: John Wiley & Sons,

Inc., pp. 25–78. Available at: http://doi.wiley.com/10.1002/9780470175637.ch3

[Accessed May 18, 2017].

Schwartz, S.J. et al., 2017. Colorants. In Fennema’s Food Chemistry. CRC Press.

Schwartz, S.J. & Lorenzo, T. V., 1990. Chlorophylls in Foods. Critical Reviews in Food

166 Science and Nutrition, 29(1), pp.1–17.

Sharma, V., McKone, H.T. & Markow, P.G., 2010. A Global Perspective on the History,

Use, and Identification of Synthetic Food Dyes. Journal of Chemical Education,

88(1), pp.24–28. Available at: http://pubs.acs.org/doi/pdf/10.1021%2Fed100545v

[Accessed August 26, 2015].

Shi, M.-Z. & Xie, D.-Y., 2014. Biosynthesis and Metabolic Engineering of Anthocyanins

in Arabidopsis thaliana. Recent Patents on Biotechnology, 8(1), pp.47–60. Available

at: http://www.eurekaselect.com/openurl/content.php?genre=article&issn=1872-

2083&volume=8&issue=1&spage=47.

Shriner, R.L. & Moffett, R.B., 1941. Benzopyrylium Salts. III. Syntheses from

Substituted Coumarins and Chromones. Journal of the American Chemical Society,

63(6), pp.1694–1698.

Sigurdson, G.T. et al., 2017. Effects of hydroxycinnamic acids on blue color expression

of cyanidin derivatives and their metal chelates. Food Chemistry, 234, pp.131–138.

Available at: http://dx.doi.org/10.1016/j.foodchem.2017.04.127.

Sigurdson, G.T. et al., 2016. Evaluating the role of metal ions in the bathochromic and

hyperchromic responses of cyanidin derivatives in acidic and alkaline pH. Food

Chemistry, 208, pp.26–34.

Sigurdson, G.T. et al., 2018. Impact of Location, Type, and Number of Glycosidic

Substitutions on the Color Expression of o-dihdroxylated Anthocyanidins. Food

Chemistry, Under Revi.

Sigurdson, G.T., Tang, P. & Giusti, M.M., 2017. Natural Colorants: Food Colorants from

167 Natural Sources. Annual Review of Food Science and Technology, 8(1), pp.261–280.

Available at: http://www.annualreviews.org/doi/10.1146/annurev-food-030216-

025923 [Accessed March 14, 2017].

Smith, G. & Conway, S., 2004. opencsv. Available at:

http://sourceforge.net/projects/opencsv/.

Somers, T.C., 1971. The polymeric nature of wine pigments. Phytochemistry, 10(9),

pp.2175–2186.

Sondheimer, E. & Kertesz, Z.I., 1953. Participation of ascorbic acid in the destruction of

anthocyanin in strawberry juice and model systems. Biol.Chem.,, 18(922), pp.475–

479. Available at: http://doi.wiley.com/10.1111/j.1365-2621.1953.tb17740.x

[Accessed January 9, 2017].

Sondheimer, E. & Kertesz, Z.I., 1952. The kinetics of the oxidation of strawberry

anthocyanin by hydrogen peroxide. Journal of Food Science, 17(1–6), pp.288–298.

Available at: http://doi.wiley.com/10.1111/j.1365-2621.1952.tb16766.x [Accessed

March 21, 2017].

Starr, M.S. & Francis, F.J., 1968. Oxygen and ascorbic acid effect on the relative stability

of four anthocyanin pigment in cranberry juice. Food Technology, 22, pp.1293–

1295.

Stebbins, N.B. et al., 2016. Ascorbic acid-catalyzed degradation of cyanidin-3-O-β-

glucoside: Proposed mechanism and identification of a novel hydroxylated product.

Journal of Berry Research, 6(2), pp.175–187.

Stillman, J.A., 1993. Color influences flavor identification in fruit-flavored beverages.

168 Journal of Food Science, 58(4), pp.810–812.

Stintzing, F.C. et al., 2002. Color and antioxidant properties of cyanidin-based

anthocyanin pigments. Journal of Agricultural and Food Chemistry, 50(21),

pp.6172–6181.

Sui, X., Dong, X. & Zhou, W., 2014. Combined effect of pH and high temperature on the

stability and antioxidant capacity of two anthocyanins in aqueous solution. Food

Chemistry, 163, pp.163–170. Available at:

http://dx.doi.org/10.1016/j.foodchem.2014.04.075.

Szalóki-Dorkó, L. et al., 2015. Degradation of anthocyanin content in sour cherry juice

during heat treatment. Food Technology and Biotechnology, 53(3), pp.354–360.

Talcott, S.T. et al., 2003. Phytochemical stability and color retention of copigmented and

processed muscadine grape juice. Journal of Agricultural and Food Chemistry,

51(4), pp.957–963.

Taylor, S.L., Higley, N.A. & Bush, R.K., 1986. Sulfites in foods: Uses, analytical

methods, residues, fate, exposure assessment, metabolism, toxicity, and

hypersensitivity. Advances in Food Research, 30(C), pp.1–76.

Tian, Q. et al., 2005. Screening for anthocyanins using high-performance liquid

chromatography coupled to electrospray ionization tandem mass spectrometry with

precursor-ion analysis, product-ion analysis, common-neutral-loss analysis, and

selected reaction monitoring. Journal of Chromatography A, 1091(1–2), pp.72–82.

Available at: http://ac.els-cdn.com/S0021967305014615/1-s2.0-

S0021967305014615-main.pdf?_tid=fa83127e-835c-11e7-994c-

169 00000aacb35f&acdnat=1502982283_a779c1c4873a37bf4c667ceb6a5d6f02

[Accessed August 17, 2017].

Torskangerpoll, K. & Andersen, Ø.M., 2005. Colour stability of anthocyanins in aqueous

solutions at various pH values. Food Chemistry, 89(3), pp.427–440.

Trošt, K. et al., 2008. Anthocyanin degradation of blueberry-aronia nectar in glass

compared with carton during storage. Journal of Food Science, 73(8), pp.405–411.

Trouillas, P. et al., 2016. Stabilizing and Modulating Color by Copigmentation: Insights

from Theory and Experiment. Chemical Reviews, 116(9), pp.4937–4982. Available

at: http://pubs.acs.org/doi/abs/10.1021/acs.chemrev.5b00507 [Accessed October 28,

2016].

Tsao, C.S., Packer, L. & Fuchs, J., 2009. An overview of ascorbic acid chemistry and

biochemistry. J. Fac. Pharm., 38(3), pp.233–255.

Varvara, M. et al., 2016. The use of the ascorbic acid as food additive and technical-legal

issues. Italian Journal of Food Safety, 5(1). Available at:

http://www.pagepressjournals.org/index.php/ijfs/article/view/ijfs.2016.4313.

Vuong, Q. V et al., 2011. Optimizing conditions for the extraction of catechins from

green tea using hot water. Journal of separation science, 34(21), pp.3099–106.

Available at: http://www.ncbi.nlm.nih.gov/pubmed/21905216 [Accessed January 12,

2015].

Wang, D. et al., 2012. Gut microbiota metabolism of anthocyanin promotes reverse

cholesterol transport in mice via repressing miRNA-10b. Circulation Research,

111(8), pp.967–981.

170 Wang, Q. & Cui, S.W., 2005. Food Carbohydrates. Chemistry, Physical Properties, and

Applications, CRC Press.

Williamson, S.J. & Cummins, H.Z., 1983. Light and color in nature and art, New York:

Wiley.

Willstätter, R. & Everest, A.E., 1913. Untersuchungen über die Anthocyane. I. Über den

Farbstoff der Kornblume. Justus Liebigs Annalen der Chemie, 401(2), pp.189–232.

X-RIte, I., 1993. A Guide to Understanding Color Communication, Grandville,

Michigan: X-Rite. Available at:

https://books.google.com/books?id=bFJkGwAACAAJ.

Yanishlieva, N. V. & Marinova, E.M., 2001. Stabilisation of edible oils with natural

antioxidants. European Journal of Lipid Science and Technology, 103(11), pp.752–

767.

Yokota, K. et al., 2013. Analysis of A-type and B-type highly polymeric

proanthocyanidins and their biological activities as nutraceuticals. Journal of

Chemistry, 2013, pp.1–8.

Yuan, J.-P. & Chen, F., 1998. Degradation of Ascorbic Acid in Aqueous Solution.

Journal of Agricultural and Food Chemistry, 46(12), pp.5078–5082. Available at:

http://dx.doi.org/10.1021/jf9805404 [Accessed September 10, 2015].

Zhang, C. et al., 2009. Influence of copigmentation on stability of anthocyanins from

purple peel in both liquid state and solid state. Journal of agricultural and

food chemistry, 57(20), pp.9503–8. Available at:

http://www.ncbi.nlm.nih.gov/pubmed/19791791.

171 Zhao, C.L. et al., 2017. Stability-increasing effects of anthocyanin glycosyl acylation.

Food Chemistry, 214, pp.119–128.

Zhao, C.L.L. et al., 2014. Structure–activity relationships of anthocyanidin glycosylation.

Mol Divers, 18(3), pp.687–700. Available at:

https://link.springer.com/content/pdf/10.1007%2Fs11030-014-9520-z.pdf [Accessed

October 18, 2017].

172