Mass Spectrometry Based Investigation of Chlorogenic Acid Reactivity and Profile in Model Systems and Coffee Processing

Sagar Deshpande

A Thesis submitted in partial fulfillment of the requirements for the degree of

Doctor of Philosophy in Chemistry

Approved Dissertation Committee

Prof. Dr. Nikolai Kuhnert (Chair) Prof. Dr. Gerd-Volker Röschenthaler (Reviewer) Prof. Dr. Michael N. Clifford (External Reviewer)

Date of Defense: 24th January 2014

School of Engineering and Science Abstract

Beneficial health and biological effects of coffee as well as its sensory properties are largely associated with chlorogenic acids (CGAs) since; coffee is the richest dietary source of CGAs and their derivatives. From green coffee beans to the beverage, chemical components of the green coffee undergo enormous transformations, which have been studied in great details in the past. Roasted coffee melanoidines are extensively contributed by the products formed by the most relevant secondary metabolite- chlorogenic acids. For every 1% of the dry matter of the total CGA content in the green coffee beans, 8-10% of the original CGAs are transformed or decomposed into respective derivatives of cinnamic acid and quinic acid. The non-volatile fraction of the roasted coffee remains relatively unravelled in the aspects of its chemistry and structural information. Coffee roasting, along with the other processes brings about considerable changes in the chlorogenic acid profile of green coffee through number of chemical processes. In roasting, chlorogenic acids evidently undergo various processes such as, acyl group migration, transesterification, thermal trans-cis isomerization, dehydration and epimerization. To understand the chemistry behind roasted coffee melanoidines, it is of utmost importance to study the changes occurring in CGAs and their derivatives through food processing.

The isomeric transformations of the chlorogenic acids resulting due to the migration of hydroxycinnamoyl group from any of the four hydroxyl groups of quinic acid to another have been thoroughly investigated in this work by LC-MSn. In this thesis, the acyl migration phenomenon under the treatment of tetramethylammonium hydroxide (TMAH) hydrolysis, model roasting experiments and by brewing at pH 5 (water reflux, 5 h) of the seven commercially available mono- and di-caffeoylqunic acids was studied in detail. Intermolecular acyl migration (transesterification) was also studied by tetramethylammonium hydroxide (TMAH) hydrolysis and model roasting experiments in between 5-CQA and p- coumaric acid as well as in 5-CQA and ferulic acid.

In this thesis, four diastereoisomers of quinic acid have been synthesized selectively, namely, epi-quinic acid, muco-quinic acid, cis-quinic acid and scyllo-quinic acid by applying appropriate hydroxyl group protection and deprotection strategies in order to study their behavior in LC-MSn along with commercially available (-)-quinic acid. We report for the first time that these diastereoisomers are distinguishable on the basis of their fragmentation behavior as well as their chromatographic elution order. In this study, we also observed that

i muco-quinic acid, scyllo-quinic acid and epi-quinic acid are present in hydrolyzed Guatemala roasted coffee sample as possible products of roasting. The synthetic work accomplished in present work will provide for the generation of the reference standards to identify remaining epimers of CGAs in roasted coffee.

Considering the fact that relatively large amount of the degradation products of CGAs such as, quinic acid, caffeic acid along with the most prominent member of the CGAs profile in roasted coffee, 5-caffeoylquinic acid itself are present in the roasted coffee along with free small, non-volatile organic acids, we examined in details the further esterification phenomenon among themselves. To investigate transesterification in roasted coffee in details we designed a thorough analytical plan involving four experiments. A selection of small organic acids were heated in the presence of 5-CQA to check if simulated roasting conditions facilitate the formation of the transesterification, caffeic acid and quinic acid with the mixture of all the organic acids separately to check, which of the organic acid show greater affinity towards the formation of the condensed esters. With the experimental results in hand, we then identified transesterification products in different roasted coffee samples by LC-MSn, LC- TOF-MS and FT-ICR-MS.

Data generated by different analytical techniques such as, NMR-, CD-, IR spectroscopy and LC-MS was used to differentiate the Arabica and Robusta green coffee extracts by principal component analysis (PCA) to determine, which spectroscopic technique allows the best discrimination of coffee varieties. A total of 38 green bean extracts were characterized using NMR-, CD-, and IR spectroscopy along with LC-MS and the data was further analyzed by PCA using different PCA processing parameters by unsupervised non-targeted approach. Distinction between different groups of samples, in particular, Arabica versus Robusta green coffee beans successfully achieved using IR- spectroscopy and LC-MS. Surprisingly, both CD- and NMR spectroscopy fail to achieve in this case, an adequate level of distinction. This is the first study that directly compares the value of various spectroscopic techniques if multivariant statistical techniques are employed to them.

Acknowledgements

I, hereby acknowledge that most of the times, this specific segment of the thesis write-up is supposed to be formal. But, I want to take up the liberty of using the language in which, I am comfortable in to thank people who actually thought they would make a possible ‘scientist’ out of something like ‘me’ risking a doubt which may arise in my examiners mind of being too non-scientific in my thesis. I am taking this risk because, to my personal belief, I think the apology and the gratefulness should come by heart when you are down two beers.

Firstly, I would like to thank Mr. Yadav who awaken the interest in me to peruse chemistry as a career as opposed to being a play-writer or a Himalayan monk in my adolescent age. Big thanks with a little judgment goes to the committee, which accepted my application for masters in Nanomolecular science in Jacobs University after having a degree in organic chemistry with moderate grades. I must thank Mr. Sunil Joshi for giving me an opportunity to work in a reputed research institute such as, National Chemical Laboratory (NCL), Pune, which made my application to Jacobs look considerable. I would like to thank Prof. Dr. Nugent who pumped a great deal of professionalism into me starting from how I should reply to the formal emails. Prof. Nugent pulled me out of some tough times and taught me to be serious about the opportunities I have in present and may have in my future. I am grateful to Ms. Shalaka Shah, Mr. Ketan Kulkarni and my lecturer first and a very close friend later, Mr. Suparna Tambe for making my masters days in India and in Germany so easy and joyful. My lab-mates, Rakesh, Tina, Aga, Hande, Marius, Maria, Rohan, Boris, Mohammed, Abhinandan (names not in order of importance!) and their partners have been the family away from my home. All of whom, suffered me, my inappropriate jokes and occasional emotional outbursts with so much of tolerance, that I am unable express my gratitude towards them even if they might not understand the necessity in doing so.

I cannot even begin to thank Prof. Dr. Nikolai Kuhnert for trusting me to perform this work as a PhD candidate. My intelligence level matches to a Chimp in front of him; still it’s his skills that he managed to pull a significant contribution out of me to the coffee chemistry. I want to mention that if I manage to incorporate 10% of his knowledge and 1% of his modesty and humbleness in myself when I am of his age, I would consider myself to be very successful in my life. I want to thank Prof. Dr. Clifford for accepting to be an external examiner for my dissertation although; it is of considerable inconvenience for him to travel from U.K. and Prof. Dr. Gerd-Volker Röschenthaler for taking time to revise my thesis. I am thankful to Ms.

iii

Anja Mueller for her technical support for five years. I appreciate my collaborating professor, Prof. Dr. Materny and his group, in particular Dr. Rasha El-Abasssy for very fruitful collaborative work. I would like to thank Dr. Bassem Bassil from the group of Prof. U. Kortz for the measurement and solution of the single crystal X-ray structures.

I have been in Germany for more than five years now and most of the times my wife Neha and me, had to live apart from each other. She has been and will be a definition of unconditional love for me. She stretched me to the extremes of the feeling of happiness and (rarely) suicidal throughout the entire time before and after marriage. She trusted me with the fatherhood of our child, Raghav. Although, she would doubt it, there is no one close to my heart than her. We are a proud evidence for a working long distance happy relationship.

There are actually very few words in which I could be expressing my gratefulness towards my father, mother, sister and uncle completely. I am very aware of the fact that it is due to their financial and emotional sacrifices I am what I am today.

I am thankful to Jacobs community for teaching me how to accept and respect diverse nationalities and religions. It is the healthy international environment of Jacobs University, which gave me conversational confidence and a feeling of not being special or unique than other communities all over the world. I am thankful to my Indian friends in India and in Jacobs as well, for support and long lasting memories throughout the entire stay in Germany. My former flat mate turned very good friend Naveen, thanks to you too.

Last but not the least at all, I am very thankful to Jacobs University and Kraft Food for financial support.

Contents

Abstract………………………………………………………………………………………..i

Acknowledgements…………………………………………………………………………..iii

List of Figures………………………………………………………………………………...ix

List of Tables………………………………………………………………………………..xiii

Abbreviations………………………………………………………………………………..xv

CHAPTER 1: Introduction ...... 1

1.1 Introduction ...... 1

1.2 Coffee: Commercial aspects and biological relevance ...... 1

1.3 Classes of compounds present in coffee ...... 4

1.3.1 Carbohydrates...... 4

1.3.2 Lipids...... 5

1.3.3 Amino acids and protein...... 6

1.3.4 Acids and phenolic compounds in coffee ...... 7

1.4 Chlorogenic acids: Definition, history, occurrence and biological relevance ...... 9

1.5 Fate of CGAs in food processing and Aim of the project ...... 17

1.5.1 Acyl migration...... 18

1.5.2 Epimerization...... 19

1.5.3 Transesterification ...... 20

1.5.4 Suitability of an analytical technique to differentiate large set of green coffee extracts from various origins by PCA...... 20

References...... 22

CHAPTER 2: LC-MSn identification of CGAs in green and roasted coffee ………….31 2.1 Introduction ...... 31

2.2 LC-MSn identification of CGAs in green coffee ...... 32

2.3 LC-MSn identification of CGAs in roasted coffee ...... 45

References……………………………………………………………………………..50 v

CHAPTER 3: Acyl migration in mono- and di-caffeoylquinic acids under basic and aqueous acidic conditions and dry roasting conditions ...... 52

3.1 Introduction ...... 52

3.2 Materials and methods ...... 53

3.3 Results and discussion ...... 55

3.3.1 Intramolecular acyl migration: hydrolysis by TMAH of 2-5 and 8-10...... 55

3.3.2 Intermolecular acyl migration (Transesterification): hydrolysis by TMAH (Cross- over experiment) ...... 63

3.3.3 Intramolecular acyl migration: model roasting of 2-5 and 8-10 ...... 72

3.3.4 Transesterification: model roasting (Cross-over experiment) ...... 77

3.3.5 Intramolecular acyl migration: Brewing of CGAs ...... 80

3.4 Conclusions……………………………………………………………………………….83

References……………………………………………………………………………..86

CHAPTER 4: Synthesis, structure and tandem MS investigation of diastereomers of quinic acid ...... 89

4.1 Introduction ...... 89

4.2 Experimental ...... 91

4.2.1 Synthesis of the mixture of the epimers of (-)-quinic acid ...... 91

4.2.2 Synthesis of the epi-quinic acid (2)...... 92

4.2.3 Synthesis of the muco-quinic acid (3) ...... 95

4.2.4 Synthesis of the cis-quinic acid (4) ...... 95

4.2.5 Synthesis of the scyllo-quinic acid (5) ...... 99

4.2.6 Hydrolysis of the CGAs in roasted coffee ...... 101

4.2.7 Synthesis of the methyl esters of epi-, muco-, cis-, scyllo-quinic acids and (-)-quinic acid ...... 101

4.2.8 X-ray crystallography ...... 102

4.3 Results and discussions ...... 102

4.4 Discussion of the X-ray structures ...... 120

4.5 Conclusions ...... 122

References...... 123

CHAPTER 5: Transesterification of chlorogenic acids with small organic acids present in the coffee bean ...... 125

5.1 Introduction ...... 125

5.2 Materials and methods ...... 127

5.2.1 Chemicals and materials ...... 127

5.2.2 Model roasting ...... 128

5.2.3 Aqueous extract of roasted coffee ...... 129

5.2.4 Roasted coffee samples for ESI-FT-ICR-MS analysis ...... 129

5.3 Results and discussion ...... 135

5.3.1 Transesterification of 5-CQA(2) in model roasting and in roasted coffee samples ...... 136

5.3.2 Transesterification of quinic acid (1) in model roasting and in roasted coffee samples ...... 151

5.3.3 Transesterification of caffeic acid (3) in model roasting and in roasted coffee samples ...... 158

5.4 Conclusions...... 160

References...... 161

CHAPTER 6: Which spectroscopic technique allows best differentiation of coffee varieties: Comparing principal component analysis using data derived from CD-, NMR- , IR- spectroscopy and LC-MS in the analysis of the chlorogenic acid fraction in green coffee beans ...... 163

6.1 Introduction ...... 163

6.2 Materials and methods ...... 166

6.2.1 Statistical analysis ...... 166

6.3 Experimental ...... 167

6.4 Results and discussion ...... 168

vii

6.4.2 Circular Dichroism spectroscopy ...... 170

6.4.3 Infrared spectroscopy ...... 177

6.4.4 1H NMR spectroscopy ...... 179

6.5 Conclusions...... 181

References...... 182

Conclusions...... 184

List of Publications...... 186

Curriculum Vitae...... 188

viii

List of figures

Figure 1.1 Structures of caffeine and adenosine 3

Figure 1.2 Basic chemical structures of aliphatic carboxylic acids which are present in coffee in free form and in mixed esters of CGAs 8 Figure 1.3 Basic structures of quinic acid, derivatives of cinnamic acid and typical chlorogenic acids 10

Figure 1.4 Basic structures of feruloyl- and p-coumaroylquinic acid and mono- and dicaffeoylquinic acids 14

Figure 1.5 Chlorogenic acids and their derivatives 15

Figure 1.6 Representative scheme showing possible chemical transformations in CGA 18

Figure 1.7 Stereoisomers of quinic acid 19

Figure 2.1 TIC of green Robusta coffee extract in negative ion mode 32

Figure 2.2 Structures of the fragments generated by quinic and cinnamic acid derivatives 36

Figure 2.3 MS2 and MS3 spectra of 3-acyl chlorogenic acids in negative ion mode 37

Figure 2.4 MS2 and MS3 spectra of 4-acyl chlorogenic acids in negative ion mode 38

Figure 2.5 MS2 and MS3 spectra of 5-acyl chlorogenic acids in negative ion mode 39

Figure 2.6 MS2, MS3, and MS4 spectra of 3,4-diCQA, 3,5-diCQA, and 4,5-diCQA in negative ion mode (m/z 515) 41

Figure 2.7 MS2, MS3, and MS4 spectra of 3,4-diFQA 16 and 3D-4FQA 34 in negative ion mode (m/z 543 and m/z 557, respectively) 42

Figure 3.1 Structure of mono and di caffeoylquinic, p-coumaroylquinic and feruloylquinic acids 57

Figure 3.2 UV Chromatograms (318-322 nm) at 2, 5, 10 and 30 minutes of base hydrolysis of 5-CQA (3) 58

Figure 3.3 Amount of the transformation products after base hydrolysis for different time intervals of 5-CQA (3), 4-CQA (4) and 3-CQA (2) 59

Figure 3.4 Mechanism of the acyl migration through an ortho-ester intermediate formation 60

Figure 3.5 Amount of the transformation products after base hydrolysis for different time intervals of di-acylated reference standards 63

Figure 3.6 UV Chromatograms (318-322 nm) at 2, 5, 10, 30 and 60 minutes of base hydrolysis of 3, 5-diCQA (9) 66

Figure 3.7 Compounds identified during acyl migration studies 67

Figure 3.8 Comparison between the peak areas of compounds formed during TMAH treatment of 5-CQA (3) with p-coumaric acid (pCoA) 69

Figure 3.9 EIC and fragmentation patterns for 1-cis-caffeoylquinic acid (49) at m/z 353 and caffeoyl-feruloylquinic acid (51) at m/z 529 in transesterification induced by TMAH 70

Figure 3.10 Comparison between the peak areas of compounds formed during TMAH treatment of 5-CQA (3) with ferulic acid (FA) 72

Figure 3.11 EIC and fragmentation patterns for m/z 671 observed during model roasting 76

Figure 3.12 EIC and fragmentation patterns for m/z 497 observed during model roasting 77

Figure 3.13 MS3 and MS4 of 4-pCoQA (14) and 5-FQA (23) respectively observed during cross-over experiment by model roasting 79

Figure 3.14 Structures identified after brewing of the reference standards 84

Figure 4.1 Stereoisomers of quinic acid 88

Figure 4.2 Reaction scheme for obtaining scyllo-quinic acid (5) and epi-quinic acid (2) 91

Figure 4.3 X-ray structure of methyl 3,5-Di-O-(tert-butyldimethylsilyl)quinate (7)

Conformer A 92

Figure 4.4 X-ray structure of methyl 3,5-Di-O-(tert-butyldimethylsilyl)quinate (7)

Conformer B 93

Figure 4.5 Reaction scheme for obtaining cis-quinic acid (4) and Methyl-cis-quinate (17) 95

Figure 4.6 X-ray structure 3,4-O-Cyclohexylidene-1,5-quinide (13) 96

Figure 4.7 X-ray structure of cis-quinic acid (4) 98

Figure 4.8 X-ray structures of 3-O-cinnamoyl-1,4-scyllo-quinide (9) and 3-O-cinnamoyl-1,5- quinide (10) 99

Figure 4.9 X-ray structure of methyl cis-quinate (17) 101

Figure 4.10 EIC at m/z 191 in negative mode and MSn of quinic acid (1), epi-quinic acid (2), muco-quinic acid (3), cis-quinic acid (4) and scyllo-quinic acid (5) 104

Figure 4.11 MSn of the acidic fraction of non-selectively isomerized quinic acid 109

Figure 4.12 Diastereomers identified in the EIC of the hydrolyzed roasted coffee in negative ion mode 110

Figure 4.13 MSn of the diastereomers of the quinic acid and reference compounds obtained from the direct infusion experiments 113

Figure 4.14 Proposed mechanisms for the fragmentation of 1, 2, 3 and 4 118

Figure 5.1 Structures of all the reactants involved in the model roasting experiments 127

Figure 5.2 Tentative structures of transesterification products 130

Figure 5.3 UV chromatogram at 254 nm of the model roasting experiment sample generated by heating 5-CQA (2) with succinic acid (6) 144

Figure 5.4 Fragmentation scheme for compound 19 (glutaric acid+ caffeoylquinic acid) 145

Figure 5.5 Fragmentation patterns for m/z 481 (21) and m/z 463 (22) 148

Figure 5.6 Total ion chromatogram in the negative mode of the model roasting experiment sample generated by heating QA (1) with glutaric acid (7) 151

Figure 5.7 EIC at m/z 341 159

Figure 6.1 Representative chromatogram of green coffee extract of sample No. 33 (Tanzania Robusta), a) TIC in negative ion mode; b) UV-VIS chromatogram monitored at 320 nm 171 Figure 6.2 CD spectra of 3-CQA and 3,5-diCQA 172

Figure 6.3 The PCA score and loading plots of the obtained CD spectral data 177

Figure 6.4 ATR-IR spectrum of Panama Boguete Arabica extract 179

Figure 6.5 The PCA score and loading plots of the obtained IR spectral data 179 xi

1 Figure 6.6 H-NMR spectra of Tanzania Robusta in DMSO-d6 181 Figure 6.7 The PCA score plot of the obtained NMR spectral data 181

xii

List of tables

Table 1.1 Typical quantities of CGAs found in vegetables and fruits 11

Table 1.2 Biological activities associated with CGAs 13

Table 2.1 Chlorogenic acids identified in green coffee beans 32

Table 2.2 MS2 and MS3 data of monoacyl CGAs in negative ion mode 43

Table 2.3 MS2, MS3, and MS4 data of diacyl CGAs in negative ion mode 44

Table 2.4 Chlorogenic acids identified in roasted coffee 46

Table 2.5 Negative ion mode MS2, MS3 and MS4 fragmentation data for the cinnamoylshikimate esters and chlorogenic acid lactones 48

Table 3.1 Compounds identified after base treatment of CGA with p-coumaric acid and CGA with ferulic acid for various time intervals 64

Table 3.2 Compounds identified after heating (model roasting) reference standards 73

Table 3.3 Compounds identified after heating (Model roasting) of 5-CQA (3) and p-coumaric acid 78

Table 3.4 Compounds identified after heating (Model roasting) 5-CQA (3) and ferulic acid 80

Table 3.5 Compounds identified after hydrolysis (Brewing) of reference standards 81

Table 4.1 MS2 data of quinic acid diastereomers in negative ion mode at 75% collision energy 103

Table 4.2 Crystal data and structure refinement for compounds 17, 4, 9, 10, 15 and 7 120

Table 5.1 Transesterification products of 4-11 with 5-CQA (2) identified with LC-TOF-MS in the samples generated by heating each acid separately with 5-CQA 137

Table 5.2 Transesterification products of 4-11 with 5-CQA (2) identified with LC-MSn in the samples generated by heating each acid separately with 5-CQA 141

Table 5.3 Compounds transesterified with 5-CQA identified in FT-ICR-MS data of roasted coffee samples 146

Table 5.4 Transesterification products of 4-11 with 5-CQA (2) identified with LC-TOF-MS in the roasted coffee samples 143

xiii

Table 5.5 Transesterification products of 4-11 with 5-CQA (2) identified with targeted LC- MSn in the samples generated by heating all of the acids collectively with 5-CQA 149

Table 5.6 Transesterification products of 4-11 with 5-CQA (2) identified with LC-TOF-MS in the samples generated by heating all of the acids collectively with 5-CQA 149

Table 5.7 Transesterification products of 4-11 with quinic acid (1) identified with LC-TOF- MS in the samples generated by heating each acid separately with quinic acid (QA) 153

Table 5.8 Compounds transesterified with quinic acid (QA) identified in FT-ICR-MS data of roasted coffee samples 156

Table 5.9 Transesterification products of 4-11 with quinic acid identified with LC-TOF-MS in the samples generated by heating all of the acids collectively with quinic acid 157

Table 5.10 Compounds transesterified with caffeic acid (CA) identified in FT-ICR-MS data of roasted coffee samples 158

Table 5.11 Transesterification products of 4-11 with CA (3) identified with LC-TOF-MS in the roasted coffee samples 155

Table 6.1 Origins, nature and grouping of green bean coffee samples anayzed and included in PCA analysis 173

Table 6.2 Numbering, nomenclature and high resolution MS data of selected secondary metabolites identified in green bean coffee samples 174

xiv

Abbreviations

Ac Acetyl APCI Atmospheric Pressure Chemical Ionisation Bn Benzyl CD Circular Dichorism CFQA Caffeoyl-feruloylquinic Acid CGA Chlorogenic Acid CGAs Chlorogenic Acids CQ Caffeoylquinate CQA Caffeoylquinic Acid CQL Caffeoylquinic Acid Lactone/ Caffeoylquinide CSA Caffeoylshikimic Acid / Caffeoylshikimate CSiQA Caffeoyl-Sinapoylquinic Acid DAD Diode Array Detector DCC N,N’-Dicyclohexylcarbodiimide DCE Dichloroethane DCM Dichloromethane DEAD Diethyl Azodicarboxylate DIAD Diisopropyl Azodicarboxylate DiCQA Dicaffeoylquinic Acid DMAP N,N’-Dimethylaminopyridine DMP 2,2-Dimethoxypropane DMSO Dimethylsulfoxide DMF N,N’-Dimethylformamide ESI Electrospray Ionisation FQA Feruloylquinic Acid FQL Feruloylquinic acid lactone/ Feruloylquinide FSA Feruloylshikimic Acid/ Feruloyl Shikimate FSiQA Feruloyl-Sinapoylquinic Acid

FT-ICR Fourier transform ion cyclotron resonance HPLC High Performance Liquid Chromatography HRMS High Resolution Mass Spectometry ISCID In Source Collision Induced Dissociation LC Liquid Chromatography LC-MSn Liquid Chromatography Tandem Mass Spectrometry MeOH Methanol MRM Multi Reaction Monitoring MS Mass Spectrometry PCA Principal Component Analysis p-CoQA p-Coumaroylquinic Acid RP Reverse Phase SIM Selected Ion Monitoring SiQA Sinapoylquinic Acid THF Tetrahydrofuran TLC Thin Layer Chromatography TMB 2,2,3,3-Tetramethoxybutane TBDMS-Cl tert-butyldimethylsilyl chloride TOF Time of Flight TriCQA Tricaffeoylquinic Acid Troc 2,2,2,-Trichloroethylformyl Troc-Cl 2,2,2-Trichloroethylformyl Chloride

xvi

CHAPTER 1: INTRODUCTION

1.1 Introduction The profound health and biological effects of coffee have received a lot of scientific attention in last decade. Beneficial health effects of coffee are largely associated with chlorogenic acids (CGAs) and their derivatives present in coffee, which have been extensively investigated in animals, in vitro models and in humans also through numerous epidemiological studies. 1-7 CGA content of coffee also contributes considerably towards the uniqueness of coffee’s sensory and organoleptic properties. 8

In this thesis, the work carried out and accomplished results are presented in six chapters: first chapter covers the introduction of coffee underlying its commercial aspects, classes of compounds present in coffee, introduction to CGAs; definition, occurrence in coffee and biological relevance, food processing and CGAs, and the statement of the aim of the project. The second chapter covers profiling, structural elucidation and identification of CGAs by tandem mass spectrometry (MS) in green and roasted coffee. The following three chapters describe the outcomes from three different projects undertaken to understand in details, the fate of CGAs in food processing. The last chapter deals with the challenges in verification of the possible adulterations in coffee varieties and focuses on, which spectrometric technique allows the best differentiation of coffee varieties by comparing the principle component analysis data generated from CD-, NMR-, IR- spectroscopy and LC-MS in the analysis of the chlorogenic acid fraction in green coffee beans.

1.2 Coffee: Commercial aspects and biological relevance Coffee is one of the most valued agricultural commodities in terms of the economic aspects of the exports from the developing coffee producing countries, accounting to ca. 8 million metric tonnes per year. About 70 to 80% of the total human population consumes coffee as a beverage regularly. More than 70 countries in the world cultivate coffee, which grows in the form of ‘cherries’ on a coffee plant. Approximately, 2.3 billion cups of coffee are consumed worldwide per day. 9 After water and black tea, coffee is the third most consumed beverage on this planet with a market value in excess of 5 Billion US $ of the raw material alone. In countries like USA and Germany coffee is second most consumed beverage. Approximately, 450 million cups of coffee are consumed every day in United States only. Green coffee beans

2 are produced in two varieties, Coffea Arabica (known as Arabica coffee) and Coffea canephora (known as Robusta coffee) holding 70% and 30%, respectively of the total coffee market in the world.7 The conventional coffee whether instant, filter or freshly ground is made from roasting the green beans of coffee obtained from cherry fruits of either Arabica or Robusta varieties of the coffee plant.

Similar to any other plant food material, coffee contains a complex mixture of over 1000 chemical components. In recent times, it has been acknowledged on a wider platform that consumption of coffee under specific conditions provide consumers with certain physiological benefits sourced from certain chemical components of coffee beyond basic nutritional functions. Assessment of the impact of the coffee consumption henceforth on the human health must consider potential beneficial as well as adverse health effects arising from coffee and its constituents. Specially, if we consider the fact that although coffee consumption goes back to over 1000 years in history, until recently most of the studies on health effects of coffee were based on potential adverse and toxic effects. More than 100 diseases have been linked to the coffee consumption in humans in literature.10 Among other diseases, hypertension, cardiovascular disease, cancer, fatalities to the fetus and child birth such as, early abortions and low birth weight and osteoporosis were often linked directly to coffee consumption. Roasted coffee contains series of compounds that have been shown to be carcinogens in animal models including acrylamide and furan derivatives. Regardless of the huge research attention put into the establishment of the direct link between these diseases and coffee consumption, the evidence to support the links has been inconsistent and limited. For example, a large epidemiological study involving 43,000 people was conducted in Norway which confirmed that there is no association between overall risk of cancer and coffee consumption 11 this phenomenon has been described as the coffee paradox. Not to mention that Norway is the country where per capita coffee consumption is among the largest in the world.

The beneficial health effects of coffee are attributed to the polyphenolic content present in the coffee. It varies based on the concentration of the polyphenols present in the green bean and roasted bean because, different roasting conditions gives rise to the different derivatives of the phenolic compounds in different concentrations in roasted coffee, which can affect the antioxidative properties of coffee variety directly. Commonly, protective effects of coffee consumption are generally accompanied with the antioxidant activity resulting in protection of the cells from the oxidative damage in body. However, it is now clear that these antioxidative

3 properties are caused by the phenolic compounds and their metabolites. Apart from CGAs, caffeoyl-tryptophan and caffeine present in green coffee beans have also been shown to contribute to the antioxidative activity of coffee 12, 13 along with roasted coffee melanoidines and phenylindans. The mechanism of the antioxidative activity of coffee is complex and believed to be a collective contribution towards different processes such as, radical scavenging, transition metal chelation and active oxygen trapping. The importance of antioxidant activity for the observed health effects of coffee and other phenol rich dietary materials is under intense discussion.

O N N N OH O N N H2N N OH N N O HO Caffeine Adenosine

Figure 1.1 Structures of caffeine and adenosine

Neuroactive behavioral modification induced by coffee consumption is also counted amongst positive health effects of coffee. Caffeine plays a key role as a coffee component in these effects. Caffeine blocks adenosine receptors in the central nervous system and at high doses exceeding 500 mg it interferes with gamma amino butyric acid (GABA) transmission in the brain (Figure 1.1). A typical serving of coffee contains 80-200 mg of caffeine. LD50 has been established for humans to be around 150-200 mg/ kg (192 mg/ kg in rats) with dose above 500 mg leading to initial intoxication symptoms. Caffeine has been reported to be responsible for increased alertness, performance improvement and fatigue reduction in the literature.14 Behavioral modification through coffee- induced neuroactivity has also been documented to possess properties towards prevention of suicidal tendencies and caffeine has directly been related as active anti-depression factor. Epidemiological studies report three to five fold decreases in the risk of suicide in both men and women compared to the placebo group, directly related to the coffee intake.15 Griffiths et al. established a direct relationship between the experimental administrations of caffeine to the increase in subjective feeling of wellness, self-confidence and motivation and decrease in the social anxiety. 16 Caffeine has also been reported as a responsible factor for improved mood 17 and decreased irritability 18 in psychiatric studies.

As stated earlier, coffee consumption has been attempted to be linked as one of the causes of various negative health effects by the scientific community in the past. But, recent epidemiological evidences and results from different in vivo and in vitro studies on the effect of coffee constituents on human health suggests that coffee consumption may prove protective against some type of cancers such as, colon cancer. 19 It is postulated that the antioxidant activity induced by the constituents of coffee stimulates the chemo-detoxification processes in body based on the experimental data, resulting in the chemoprotective properties of coffee. As reported by Klatsky et al. and Corrao et al., risk of acquiring cirrhosis through excessive alcohol consumption can be reduced by coffee consumption.20-22 It was also observed that consumption of coffee adversely affect the advancement of hepatitis B and C infections on development of cirrhosis thus indicating positive effect of coffee in case of non- alcoholic cirrhosis. Stelzer et al. published an epidemiological study in which he found that coffee consumption in patients undergoing radiotherapy as a part of cervical cancer treatment significantly decreases the chances of severe late radiation injuries. 23

Caffeine and certain chlorogenic acids have been reported to have glycemic effects in humans. 102, 105,106 Johnston and Clifford reported that coffee consumption modulated plasma glucose levels, gastrointestinal hormone and insulin secretion in humans. Their study confirmed the role of caffeine in glucose uptake, gastrointestinal hormone and insulin secretion also, 5-CQA was observed to impart an antagonistic effect on glucose transport.24

1.3 Classes of compounds present in coffee Carbohydrates, proteins and amino acids, lipids and organic acids and phenolics are the most prominent classes of compounds in the non-volatile fraction of coffee, which have been investigated in details.

1.3.1 Carbohydrates Carbohydrates content in coffee is very important, as it contributes approximately half of the dry weight of the green coffee bean. 25 The roasting process brings out extensive changes in the chemical constituents of the green coffee in which, both low and high molecular weight carbohydrates present in green coffee bean play a major role. Monosaccharides are found in negligible amounts in coffee with sucrose being present as major component of the low molecular weight sugar content. Arabica coffee variety contains about twice amount of sucrose than Robusta variety. Clifford summarized the reported amount in literature to be in the range of 2- 5% for Robusta and 5- 8.5% for Arabica coffee green beans 26 also, Robustas were found to contain more reducing sugars than Arabicas. There was no evidence found for

5 the existence of other simple oligomeric sugars such as, raffinose or stachyose except for sucrose and mannose in very minute quantity (0.1%). 27 However, upon roasting the sucrose content is degraded almost completely and roasted coffee contains only negligible amounts of sucrose totaling to 0.24- 0.33%. 28 The hydrolysis products of sucrose, which are reducing sugars, were expected to be identified in the roasted coffee samples in quantitative amount but, evidently glucose and fructose undergo thermal degradation even more rapidly than sucrose only to leave their trace amounts in light roasted coffee. As reported by Noyes and Chu, 21 samples from roasted Arabica or Robusta coffees of Brazilian origin showed the presence of only 0.1% sugars in total with 0.8% of sucrose itself. 29 In the roasting conditions involving higher temperature since, water is absent for the glycosidic bond cleavage, cellulose is expected to form through pyrolysis of oligomers or polymers of monosaccharides. 30 However, anhydro-sugars were also detected in the range of 0.1%.

On the other hand, high molecular weight sugars or polysaccharides were found to be relatively stable in the roasting process since, they are the principle cell wall component of coffee bean, which through its thickness, provide the characteristic hardness to the bean. Polysaccharide content in green coffee bean is dominated by arabinogalactan, mannan and/or galacto-mannan and cellulose. 31 Cellulose was found to be the most stable polysaccharide during the roasting procedure while, arabinogalactan being the most labile. Mannan remains more stable to roasting than galacto-mannan polysaccharides. 31 Rich crema or foam generated in espresso coffee signifies the quality of the brew. Nunes et al. established a direct relationship between the stability of the foam and concentration of the polysaccharide content present in the coffee variety. Brazilian Arabicas and Ugandan Robustas found out to contain highest polysaccharide content hence, producing richest ‘crema’ of espresso brew. 32

Polysaccharides present in coffee are not hydrolyzed by mammalian enzymes therefore, are considered as dietary fibers. Accordingly, Rao et al. reported potential properties of coffee fiber as anti-colon cancer agent. 33

1.3.2 Lipids Most amount of the coffee-oil or the lipids are found to be present in the endosperm of the coffee bean. 34 The lipid content varies in green Arabica and Robusta coffee. About 15% of the dry weight of green Arabica coffee bean is comprised of lipids while, green Robusta coffee bean possesses 10% of the lipids on a dry weight basis. 35 Out of the total green coffee-oil, about 75% is comprised of triglycerides, which mean most of the lipid content of green coffee is unsaponifiable. Free and esterified diterpene alcohols contribute about 19% to

6 the rest 25% of the coffee oil whereas, esterifed sterols contribute 5%. Very small quantities of other substances such as, tocopherols are also present in the composition of the green coffee oil. 36 Upon roasting, fatty acid content does not undergo major changes as far as the amounts are concerned. Trans fatty acid levels are found to be increasing after roasting green Arabica and Robusta coffee beans. Free fatty acids (FFAs) were found to exist at around 1 g/ 100 g of total lipids whereas, green Robustas FFA content was found at comparatively higher amount; about 2 g/ 100 g of lipids. Roasting does not bring about noticeable changes in the amount of distribution of FFA content with only exception of linoleic acid, which decreases slightly during roasting at higher temperatures. 37

In this work, we selected linoleic acid and palmitic acid as representatives of FFA content of lipids in green coffee for the purpose of studying the possibility of transesterification between FFA and quinic acid (1) during model roasting experiments since, both of them are observed in significant amounts in their free form in roasted coffee. Detailed results from this experiment are presented in Chapter 4.

1.3.3 Amino acids and protein Both amino acids and proteins contribute towards the color, flavor and aroma of the brewed coffee. Quantity and variety of the amino acids and proteins present in coffee play a vital role to the aromatic qualities of coffee variety. 38 Free amino acid (FAA) content of the green coffee bean is largely transformed by the roasting process resulting in trace amounts in roasted coffee.

The post-harvesting processes such as, drying, fermentation and storage affect the content of free amino acids (FAAs) considerably. For example, if adhering pulp was not removed from the coffee bean before drying, amount of glutamic acid increases after drying the freshly harvested coffee beans to 500 mg/ kg dry basis. Whereas, in most samples, aspartic acid was found to be decreasing by 110-780 mg/ kg dry basis. 39 Decaffeination by steam treatment decreases free amino acids content significantly. It was observed that decrease in the levels of FAA by steaming is grater in Arabica coffee than Robusta coffee. The amount of protein bound amino acids was also found to be decreasing with the increase in the duration of steam treatment. The industrially steamed coffee bean contains 10% less of protein bound amino acids and 50% less of FAAs compared to their original amounts. On the other hand, protein bound amino acids were found to be more stable in the roasting environment than free amino acids, this observation is supported by the recent studies confirming the role of proteins in the formation of aroma and metal-chelating compounds in coffee brew. 38

Derivatives of amino acids with caffeic acid and other hydroxycinnamic acids such as, ferulic acid and p-coumaric acid have been detected in green coffee bean. 40-42 However, in roasted coffee, no information about these derivatives found reported in the literature.

1.3.4 Acids and phenolic compounds in coffee Acids present in coffee are the major contributors to the perceived taste characteristic to coffee. Acidity in coffee is considered as one of the important parameters for the quality of the coffee variety. 43 11 % of the total weight of the green coffee bean is contributed by the acid content, which decreases upon roasting to 6 %. 44 This acid content is contributed by various volatile and non-volatile acids. As Clifford reported earlier, in brewed coffee citric acid (9), phosphoric acid (3), phytic acid (2), quinic acid (1), chlorogenic acids and malic acid (10) are the most important acids contributing to the perceived acidity. 43 Other free organic acids present in roasted coffee such as oxalic acid (4), malonic acid (5), glutaric acid (7), adipic acid (8), tartaric acid (11) and succinic acid (6) 45-54 need to be considered as factors affecting the acidic taste as well. These acids do not contribute to the titrable acidity of the coffee as established by Engelhard and Maier 54, 55 but they might exist as anions in a coffee brew or in a coffee extract providing protons. Volatile acids, which can be isolated from coffee sample by distillation processes and detected by GC directly are basically low molecular weight aliphatic compounds bound to a carboxylic functionality. Rancid smelling volatile acids such as 2- and/or 3-methylbutyric acid (13, 12) contribute to the aromatic properties of coffee. 56 Apart from formic and acetic acid, propanoic, butanoic, isomers of methyl propanoic and methyl butanoic acids are also found in Arabica and Robusta roasted coffee. Small organic acid content was found to be increasing in proportion with the higher degree of roasting; however, higher saturated fatty acids did not show significant increase with higher degree roasts. Straight chain fatty acids from C5 to C10 were also observed in roasted coffee as hydrolysis products of high molecular weight fatty acids. 57 Approximately 10% of the total composition of the processed seeds of green Coffea canephora (robusta coffee) and around 6-10% of green Arabica coffee variety comes from chlorogenic acids on dry basis, out of the total content, 5-O-caffeoylquinic acid (32) contributes about half. 7 3-O-caffeoylquinic acid (33) and 4-O-caffeoylquinic acid (34) contribute significantly after 5-O-caffeoylquinic acid to the total composition of the coffee chemistry accompanied by feruloylquinic acids (35), dicaffeoylquinic acids such as, 3,4-O-, 4,5-O- and 3,5-O-dicaffeoylquinic acids (40-42). Mono- and di-acyl chlorogenic acids

8 involving p-coumaric acid (24) and 3,4-dimethoxycinnamic acid (26) have also been observed to contribute in minor proportions to the chlorogenic acid profile of green coffee. 7

OH O O HO HO P OH O O P OH O OH OH O O HO P OH HO OH O P O O OH HO O HO OH HO O P HO OH HO P O O O O 3 4 1 OH P OH 2 HO Oxalic acid (-)-qunic acid Phosphoric acid phytic acid O O O O O O HO HO OH OH OH HO OH HO O O 5 6 7 8 Malonic acid Succinic acid Glutaric acid Adipic acid O OH O OH OH O O O O OH HO HO OH HO HO OH OH O O OH 12 10 11 9 3-methylbutyric acid Citric acid Malic acid Tartaric acid O O O OH O O OH OH HO HO HO OH O O 13 14 15 16

2-methylbutyric acid citraconic acid itaconic acid mesaconic acid

O O O OH O OH HO O HO O OH O HO O HO OH O O CH3 17 18 19 20

fumaric acid maleic acid methoxyoxalic acid 3-Hydroxy-3-methylglutaric acid Figure 1.2 Basic chemical structures of aliphatic carboxylic acids which are present in coffee in free form and in mixed esters of CGAs

50% of the chlorogenic acids are lost through decomposition during roasting thus, producing half of the decomposition products such as, quinic acid and hydroxycinnamic acids (Figure 1.3) through hydrolysis. 7, 9, 58 Quinic acid (1) ranges from 3 to 6 g/kg in green robusta and

Arabica coffee beans from various origins in the free form as reported earlier. 53, 59 After steaming the green coffee beans as a part of the decaffeination process, original quinic acid content rises by up to 15% as shown by Hucke and Maier. 60 Similarly, roasting also helps to elevate the quinic acid content. 44, 60, 61 Although chlorogenic acid lactones and quinic acid lactones are among the degradation products from chlorogenic acid, free quinic acid in roasted coffee is still found out in roasted coffee to maintain its existence between the ranges of 6.63 to 9.47 g/kg. 62 Among the other non-volatile organic acids, citric acid (9) and malic acid (10) are present in green coffee in the ranges of 5 to 15 g/kg in arabica and 3 to 10 g/kg in robusta respectively. 59, 63 12 to 18 % of these acids are degraded during roasting process. Citric acid mainly yields citraconic acid (14), glutaric acid (7), itaconic acid (15), mesaconic acid (16) and succinic acid (6) as decomposition products during roasting whereas, malic acid (10) generate fumaric acid (17) and maleic acid (18). Among these degradation products of citric and malic acids, succinic acid, glutaric acid and fumaric acid are included in this work to study their significance towards the formation of transesterification products produced in coffee roasting (Chapter 4). 62

1.4 Chlorogenic acids: Definition, history, occurrence and biological relevance Chlorogenic acids (CGAs) are generally defined as a family of the esters between quinic acids and certain trans-cinnamic acids, most commonly caffeic acid (21), p-coumaric acid (24) and ferulic acid (22) 64, 65 (Figure 1.3). This report will use the nomenclature defined by the IUPAC system for (-)-quinic acid as, 1L-1(OH),3,4/5-tetrahydroxycyclohexane carboxylic acid. 66

The most common chlorogenic acid (CGA) is 5-O-caffeoylquinic acid (32) (5-caffeoylquinic acid or 5-CQA), which formerly referred to as 3-CQA (pre-IUPAC). In this work, the current nomenclature is used as per IUPAC guidelines and the numbering in some of the references is changed as per requirement of consistency. Shorthand, which is used for abbreviation of the CGAs in this work is as follows: A-XQA or AX,BY-QA where A and B signify the position of acyl substituent and X and Y define the chemical nature of the substituent e.g. C = caffeoyl, pCo = p-coumaroyl, F = feruloyl, Si = sinapoyl, D = dimethoxycinnamoyl. Quinic acid will be referred as, QA, shikimic acid as, SA and quinic acid lactone as, QL (L= lactone). Subsequently, 3-caffeoylquinide will be written as, 3-CQL (44) and 5-caffeoylshikimic acid as, 5-CSA. Similarly, 4Si,5-CQA will stand for 4-sinapoyl-5-caffoylquinic acid (43).

The term, ‘chlorogenic acid’ arises from the chemical reaction of CGAs to generate green pigment when reacted with ferric chloride falsely indicating the presence of chlorine, which

10 was first introduced by Payen. 67 Fischer and Dangschat firstly identified CGAs to be the esters of caffeic and quinic acid. 68

O HO HO HO HO OH OH OH HO O O OH O O O O 21 22 23 24

Caffeic acid Ferulic acid Sinapic acid p-Coumaric acid

O OH O O HO

OH OH OH OH HO O O HO O O O O 25 26 27 28

m-Coumaric acid Dimethoxycinnamic acid Trimethoxycinnamic acid Trihydroxycinnamic acid

OH O OR 5 OH 5 HOOC 1 4 OH HOOC 1 4 OH OR HO 3 3 O OR OR O O OH OH 5 HOOC 1 4 OR 29 1 3 OR OR Isoferulic acid (-) Quinic acid Chlorogenic acid OR= Cinnamoyl or Alkoyl 32 R= H, 5-Caffeoylquinic acid Common name: Chlorogenic acid

Figure 1.3 Basic structures of quinic acid, derivatives of cinnamic acid and typical chlorogenic acids

CGAs may be subdivided by the identity, number, and positions of the individual acyl residues. The following subgroups can be identified:

1. Mono esters of hydroxycinnamic acids (caffeic acid (21), ferulic acid (22), sinapic acid (23), etc.). 69-71

2. Diesters, triesters, and a single tetraester of a single hydroxycinnamate moiety (e.g. diferuloyl, tricaffeoyl, or tetracaffeoyl quinic acid). 69, 71-74 This class of compounds can be referred to as homo-di or homo-tri esters. 3. Mixed diesters, triesters of caffeic acid (21) and ferulic acid (22) or any other hydroxycinnamate moieties (referred to as hetero-diesters or hetero-triesters). 71, 73, 74 4. Mixed esters involving various permutations of a hydroxycinnamate and other aromatic or aliphatic ester substituent (e.g. oxalic (4), methoxyoxalic (19), fumaric (17), succinic (6), malic (10), glutaric acid (7) characteristic for many plants of the Asteraceae family and 3-hydroxy-3-methylglutaric acid found in Gardeniae Fructus, Rubiaceae family) (Figure 1.2). 75 5. Other derivatives including cis-hydroxycinnamate esters or esters of diastereoisomers of quinic acid (1). 73 The dietary burden of CGAs with their occurrence in plants as a secondary metabolite was reviewed comprehensively by Clifford. 64, 65 Additionally, an excellent and comprehensive review summarizing the chemistry of CGAs has also been published by Clifford. 76 Table 1.1 summarizes the typical quantities of CGAs found in vegetables and fruits.

Table 1.1 Typical quantities of CGAs found in vegetables and fruits

Food source Source Amount (mgkg-1) Ref

Coffee Roast coffee 20-675 mg(200ml)-1 77

Tea Black tea 10-50 gkg-1 78, 79

Maté Maté 107-133 mg(200ml)-1 80

Pome fruits Apple 62-385 mgkg-1 81, 82

Pear 60-280 mgkg-1 83, 84

Stone fruits Cherries, apricot 150-600 mgkg-1 85

Berry fruits Blueberries 0.5-2 gkg-1 86

Blackcurrants 140 mgkg-1 87

Blackberries 70 mgkg-1 88

Raspberries 20-30 mgkg-1 88

Strawberries 20-30 mgkg-1 88

Redcurrant 20-30 mgkg-1 88

Gooseberries 20-30 mgkg-1 88

Citrus fruits Oranges 170-250 mgkg-1 89, 90

Grapefruit 27-62 mgkg-1 89, 91

Lemon 55-67 mgkg-1 89, 91

Grapes and wines Grape juice 10-430 mgl-1 81

American wine 9-116 mgl-1 92

Other fruits Pine apple 3 mgl-1 93

Kiwi 11 mgl-1 94

Brassica vegetables Kale 6-120 mgkg-1 89, 95

Cabbage 104 mgkg-1 89, 95

Brussels sprouts 37 mgkg-1 89, 95

Broccoli 60 mgl-1 96

Cauliflower 20 mgkg-1 96

Radish 240-500 mgkg-1 97

Chenopodiaceae Spinach 200 mgkg-1 98, 99

Asteraceae Lettuce 50-120 mgkg-1 98

Endive 200-500 mgkg-1 98

Chicory 20 mgkg-1 98

Solanaceae Potato 500-1200 mgkg-1 100

Aubergines 600 mgkg-1 100

Tomatoes 10-80 mgkg-1 101

Apiaceae Carrot 20-120 mgkg-1 89

Cereals Barley bran 50 mgkg-1 94

Rice 12 gkg-1 102

Table 1.2 Biological activities associated with CGAs

Source Chlorogenic Acid/ Biological Ref Derivative Activity Food materials (Diets), 32 Antioxidant 103-105 Synthetic 45 Increase hepatic 106 glucose utilization Synthetic 47 Antidiabetic 102

Synthetic 46-70 Antidiabetic 102

- 32 Antidiabetic 107 Coffee 32 Glycemic effects, 108, 109 Anticancer Baccharis genistelloides 38, 40, 41, 42 Anti-HIV 110 Achyrocline satureioides 71 Anti-HIV 110 - 32, 38, 40, 41, 42, Anti-HIV 111 71 Aster scaber 73 Anti-HIV 112 Evolvulus alsinoides 37 Antistress 113 Ipomoea pes-caprae 73, 74 Collagenase 114 inhibitor Lonicera japonica 40, 41, 42 Anti-HIV, 115 inhibition of DNA polymerase-α Lonicera japonica 74, 76 Anti-HIV, 115 inhibition of DNA polymerase-α Lonicera bournei 32, 75, 77, 78, 79, Hepatocyte 116 80, 81, 82, 83, 84, protective activity 85, 86 Hedera helix 32, 41, 42, 87 Antispasmodic 116 Roasted coffee 88, 89, 90 Antidiabetic, 117, 118 Inhibition of adenosine transporter

Approximately, a normal human being ingests 1 to 2.5 g of chlorogenic acids per day. The figures for average daily intake vary considerably depending on the diet types and food habits of the consumer as well as the insufficiency of the accurate data on these compounds. Exact quantities of CGAs present in different food materials are rarely known accurately. Inadequate knowledge on statistical variances between the content of CGAs in plant materials of different geographical origins and species adds up to the difficulties in describing CGA profile in particular food material. In many cases, the analytical data is obtained through the techniques based on derivatization followed by colorimetry, thus quantifying only one

14 particular class of CGAs e.g., caffeoyl or feruloyl esters, and therefore resulting in a possible underestimation of actual CGA content present in the sample. 119 Old data in many cases proves to be obsolete to be considered as relevant currently, because of the change in agricultural practices.

OH O OH OH 5 HOOC HOOC 1 4 OH 5 OH OH HOOC OH 3 HOOC 1 4 O 3 OH O OH O OH O O O O OH OH OH HO

HO O 4-CQA OH OH OH 35 36 33 FQA 3-CQA OH p-CoQA O OH OH 5 OH 5 HOOC 1 4 HOOC 1 4 OH O 3 O 3 O 5 OH OH O O O O O O HOOC 1 4 O OH 5 O 3 HO HOOC 1 4 OH 3 O O OH OH O O OH HO

HO HO OH OH OH OH 38 OH 40 37 OH OH 39 1,5-diCQA 3,4-diCQA 1,3-diCQA (Cynarin) OH 1,4-diCQA

OH O 5 O OH OH OH 1 4 OH 3 OH O O OH OH O O O 5 O HOOC 1 4 O O O O 3 O O 5 5 OH OH OH HOOC 1 4 OH HOOC 1 4 O HO 3 3 O OH OH O O OH OH OH 44 OH 43

3-CQL 42 4Si,5-CQA

HO 4,5-diCQA 41 OH 3,5-diCQA

Figure 1.4 Basic structures of feruloyl- and p-coumaroylquinic acid and mono- and dicaffeoylquinic acids

OH OH OH OH OH OH S OH OH N OH

O OH O O O O O O O O O NH HOOC OH O HOOC OH O O O O OH OH HOOC OH H3COOC O OH HOOC OH HOOC OH HOOC OH OH OH HOOC OH OH OH OH OH OH OH OH OH OH H Cl 49 50 51 52 H 46 47 48 45 OH OH S-3483 OH OH OH OH OH OH OH

O O O O O O O O O O O O HOOC OH O OH O O OH O HOOC OH HOOC OH HOOC OH OH HOOC OH O OH HOOC OH HOOC OH O OH O OH O OH HOOC OH OH OH OH OH

53 54 55 56 57 58 59 60 OH OH OH OH OH OH

O O O O O O O O O O O O O O O HOOC OH HOOC OH OH O HOOC OH HOOC OH HOOC OH HOOC OH HOOC O OH O OH HOOC OH O OH O OH O OH O OH O OH O OH

S 61 62 O Cl Cl Cl 65 66 67 68 63 64 Cl

OH OH OH OH

N OH N OH OH OH O S O O O O O HOOC OH O O O O O O O HOOC OH O 5 5 5 OH O OH HOOC 1 4 OH HOOC 1 4 O HOOC 1 4 O O OH HOOC OH 3 O 3 3 O OH O OH O O OH O O O O OH O O O

O O

Cl HO OH OH HO Cl 70 72 73 74 69 71 OH OH OH OH

OH OH OH 5 OH O H COOC 1 4 OH OH OH 3 OH 5 O OH 3 ROOC 1 4 O 5 OH OH O O 3 ROOC 1 4 O OH OH OH O 3 O O O OH O O OH O O OH 5 5 OH ROOC 1 4 ROOC 1 4 5 O OH OH O R=Methyl 84 3 3 ROOC 1 4 OH OH O O OH OH OH 3 R=Ethyl 85 OH OH HO R=Methyl 81 OH R=Methyl 79 83 R=Methyl 75 R=Ethyl 82 HO R=Ethyl 80 R=Ethyl 76 OH R=Methyl 77 R=Ethyl 78 HO OH O O O OH OH O O O 5 1 1 1 5 5 HOOC 1 4 OH HO HO 5 HO OH 3 3 4 3 4 3 4 OH O O O O O O O O O O O O O O O

5 O HOOC 1 4 OH 3 OH OH OH HO O OH HO O HO 87 OH OH OH 86 88 OH OH 89 90

Figure 1.5 Chlorogenic acids and their derivatives

Coffee is the richest source of chlorogenic acids and most of the beneficial health effects from coffee consumption including anti-diabetic, antioxidant, anticancer, and protective effects on Parkinson’s disease and Alzheimer’s disease 104, 120-123 have been associated with the CGA content present in coffee. CGAs are ubiquitous plant material and have been reported previously to possess important positive biological effects such as, antioxidant capacity, radical scavenging activity, antimutagenic/anticarcinogenic effect, anti-diabetic, anti-HIV, anti-bacterial, anti-HBV, inflammation inhibiting, endothelial protective properties and so on. 102, 103, 110, 111 Table 1.2 gives a brief summary of the biological effects associated with specific CGAs present in the respective food source.

1.5 Fate of CGAs in food processing and Aim of the project The volatile compounds formed in coffee roasting have been extensively researched in the past. More than 800 different volatile compounds have been identified in roasted coffee, which includes heterocycles such as, pyrazines, furanethiols, disulfides and aldehydes. 9 However, attempts to unravel the chemistry and structures of the non-volatile fraction of the roasted coffee have been reported on very few occasions. The non-volatile fraction; so called, ‘melanoidines’ is a very complex mixture composed from several fractions of diverse molecular weight, which can be separated using dialysis as shown by Schols and co-workers. 124, 125 Lack of analytical techniques and strategies to undertake complex task to obtain useful structural information of the majority of the compounds in melanoidines has been an impossible venture until now. Main constituents of the green coffee are chlorogenic acids as major secondary metabolites in green coffee, proteins and carbohydrates. Considering the fact that the CGA fraction accounts for an estimated value of 10 w% of the green coffee bean, in this work we have undertaken the task of unravelling the fate of CGA fraction during processes like roasting and brewing. During food processing, CGAs undergo chemical processes such as, acyl migration, Transesterification, thermal trans-cis isomerization, dehydration and epimerization (Figure 1.6).

O O OH HO

O O O O O O OH HO OH OH OH O Lactonisation HO HO HO O Dehydration OMe OMe OH OMe a O b OH O OH Acyl migration HO OH g O c O O O OH H HO OH OH O OH O f OH OH O OH HO HO O OMe d OMe e HO O Oxidation OMe O HO OH HO OH O O O OH O OH OH OH Hydration and trans-cis isomerisation OH HO Epimerisation OMe OMe

Figure 1.6 Representative scheme showing possible chemical transformations in CGA

1.5.1 Acyl migration Roasted coffee melanoidines are extensively contributed by the products formed by the most relevant secondary metabolite- chlorogenic acids. For every 1% of the dry matter of the total CGA content in the green coffee beans, 8-10% of the original CGAs are transformed or decomposed into respective cinnamic acid derivatives and quinic acid. 126 Hydroxycinnamoyl group in chlorogenic acids undergoes positional exchanges during coffee roasting. The isomeric transformations of the chlorogenic acids resulting due to the migration of a hydroxycinnamoyl group from any of the four hydroxyl groups of quinic acid to another have been thoroughly investigated in this work by LC-MSn. Previously, Clifford described the acyl group migration in selected chlorogenic acids in aqueous basic solutions and Dawidowicz reported on acyl migration in 5-caffeoylquinic acid (32) in aqueous acidic solutions. However, mechanistic study comparing inter- and intra-molecular acyl migration involving a number of chlorogenic acids in basic conditions and in dry roasting conditions as well as in

19 brewing process is not yet reported. In this contribution, we studied the acyl migration phenomenon under the treatment of tetramethylammonium hydroxide (TMAH) hydrolysis, model roasting experiments and by brewing at pH 5 (water reflux, 5 h) of the seven commercially available mono- and di-caffeoylquinic acids. Intermolecular acyl migration (transesterification) was also studied by tetramethylammonium hydroxide (TMAH) hydrolysis and model roasting experiments in between 5-CQA (32) and p-coumaric acid (24) as well as in 5-CQA and ferulic acid (22).

1.5.2 Epimerization CGA derivatives undergo isomerization to form CGA derivatives based on one or more of the six possible diastereoisomers of quinic acid. To verify this hypothesis the synthesis of CGA derivatives (mono and diacyl derivatives) based on the diastereoisomers of (-)-quinic acid (muco-, epi-, scyllo-, cis-, neo-, iso-quinic acid) should be accomplished.

O OH O OH O OH O OH OH OH OH OH HO OH HO HO OH HO OH OH OH OH OH OH (-)-quinic acid (1) (-)-epi-quinic acid (91) muco-quinic acid (92) cis-quinic acid (93) C inverted C4 inverted 3 C5 inverted

O OH O OH OH OH OH HO OH HO OH OH OH HO OH OH OH OH HO OH O O OH OH (+)-quinic acid (+)-epi-quinic acid scyllo-quinic acid (94) neo-quinic acid (95) C and C inverted C and C inverted 4 5 3 4

Figure 1.7 Stereoisomers of quinic acid

In this contribution, we have selectively synthesized four isomers namely, epi-quinic acid (91), muco-quinic acid (92), cis-quinic acid (93) and scyllo-quinic acid (94) in order to study their behavior in LC-MSn along with commercially available (-)-quinic acid (1) (Figure 1.7). We report for the first time that these isomers are distinguishable on the basis of their fragmentation behavior as well as their chromatographic elution order. In this study, we also observed that muco-quinic acid (92), scyllo-quinic acid (94) and epi-quinic acid (91) are present in hydrolyzed Guatemala roasted coffee sample as possible products of roasting. Non selective isomerization of (-)-quinic acid using acetic acid/conc. H2SO4 was performed from

20 which, we could identify epi-quinic acid , scyllo-quinic acid and (-)-quinic acid using newly assigned fragmentation schemes and retention times characteristic to the specific compound.

1.5.3 Transesterification The perceived taste of the coffee is largely contributed by the acids present in the coffee. In fact, acidity in coffee is considered as one of the important parameters for the quality of the coffee variety. 11 % of the total weight of the green coffee bean is contributed by the acid content, which decreases upon roasting to 6 %. 63 This acid content is contributed by various volatile and non-volatile acids. Considering the fact that relatively large amount of the degradation products of CGAs such as, quinic acid (1), caffeic acid (21) along with the most prominent member of the CGAs profile in roasted coffee, 5-caffeoyl quinic (32) acid itself are present in the roasted coffee along with free small, non-volatile organic acids. In this work, we report the further condensation of the CGAs and their decomposition products with the non-volatile fraction of the total acid content. We have taken a set of small organic acids and heated each of them individually with 5-caffeoyl quinic acid to check if simulated roasting conditions facilitate the formation of the transesterification products. Same experimental conditions were used incorporating caffeic acid and quinic acid as well. Also, we heated 5- caffeoyl quinic acid, caffeic acid and quinic acid with the mixture of all the organic acids separately to check, which of the organic acid show greater affinity towards the formation of the condensed esters. The set of eight organic acids contained oxalic acid (4), malonic acid (5), succinic acid (6), glutaric acid (7), adipic acid (8), citric acid (9), malic acid (10) and dextrotartaric acid (11) (Figure 1.2). In each of the experiment, the roasting conditions were simulated by keeping the temperature at 200 oC for the duration of 12 minutes. All the samples acquired from these experiments were analyzed by high resolution ESI-TOF-MS. Four green coffee samples were also roasted in the conditions described earlier and then analyzed by ESI-FT-ICR-MS to identify the transesterification products in roasted coffee samples.

1.5.4 Suitability of an analytical technique to differentiate large set of green coffee extracts from various origins by PCA In order to investigate parameters like geographic origin, varieties, adulterations, processing conditions, sensory properties, beneficial health effects, shelf-life or any other desirable or undesirable property of a food, a detailed knowledge of its composition and chemistry is required and therefore becomes foremost a problem of analytical chemistry. Once the chemical constituents of food have been elucidated comparison of the chemical

21 profile of different samples allows differentiation between samples and identification of variations that are of interest to both producer and consumer. Most popular statistical method aimed at data reduction in current times is Principle component analysis (PCA). The question however arises when choosing the most suitable analytical technique which can provide optimum differentiation between given set of food samples. In this work, we tried to answer this question in terms of differentiation of green coffee extracts. Within this contribution we have analysed aqueous methanolic extracts of a total of 38 green bean coffee samples, which vary in terms of coffee variety and processing conditions. We have characterized these extracts using NMR-, IR- and CD spectroscopy along with LC-MS. All spectroscopic data have been analysed by principal component analysis (PCA) using different PCA processing parameters using an unsupervised non-targeted approach. We could show, that distinction between different groups of samples, in particular, Arabica versus Robusta green coffee beans can be successfully carried out using IR- spectroscopy and LC- MS. Surprisingly both CD- and NMR spectroscopy fail to achieve in this case, an adequate level of distinction. This is to our knowledge the first study that directly compares the value of various spectroscopic techniques if multivariant statistical techniques are employed to them.

References

1. Czok, G. Coffee and health. Z. Ernaehrungswiss. 1977, 16, 248-255.

2. Higdon, J.V.; Frei, B. Coffee and health: a review of recent human research. Crit. Rev. Food Sci. Nutr. 2006, 46, 101-123.

3. Dorea, J.G.; da Costa, T.H.M. Is coffee a functional food?. Br. J. Nutr. 2005, 93, 773-782.

4. Hamer, M. Coffee and health: explaining conflicting results in hypertension. J. Hum. Hypertens. 2006, 20, 909-912.

5. van Dam, R.M. Coffee consumption and risk of type 2 diabetes, cardiovascular diseases, and cancer. Appl Physiol Nutr Metab 2008, 33, 1269-1283.

6. Tavani, A.; La, V.C. Coffee and cancer: a review of epidemiological studies, 1990-1999. Eur J Cancer Prev 2000, 9, 241-256.

7. Crozier, A.; Jaganath, I.B.; Clifford, M.N. Dietary phenolics: chemistry, bioavailability and effects on health. Nat. Prod. Rep. 2009, 26, 1001-1043.

8. Frank, O.; Blumberg, S.; Kunert, C.; Zehentbauer, G.; Hofmann, T. Structure Determination and Sensory Analysis of Bitter-Tasting 4-Vinylcatechol Oligomers and Their Identification in Roasted Coffee by Means of LC-MS/MS. J. Agric. Food Chem. 2007, 55, 1945-1954.

9. Jaiswal, R.; Matei, M.F.; Golon, A.; Witt, M.; Kuhnert, N. Understanding the fate of chlorogenic acids in coffee roasting using mass spectrometry based targeted and non-targeted analytical strategies. Food Funct. 2012, 3, 976-984.

10. Leviton, A.; Pagano, M.; Allred, E.N.; el Lozy, M. Why those who drink the most coffee appear to be at increased risk of disease. A modest proposal. Ecol. Food Nutr. 1994, 31, 285- 293.

11. Stensvold, M.I.; Jacobsen, B.K. Coffee and cancer: a prospective study of 43,000 Norwegian men and women. Cancer Causes & Control 1994, 5, 401-408.

12. Ohnishi, M.; Morishita, H.; Toda, S.; Yase, Y.; Kido, R. Inhibition in vitro linoleic acid peroxidation and haemolysis by caffeoyltryptophan. Phytochemistry 1998, 47, 1215-1218.

13. Devasagayam, T.; Kamat, J.; Mohan, H.; Kesavan, P. Caffeine as an antioxidant: inhibition of lipid peroxidation induced by reactive oxygen species. Biochimica et Biophysica Acta (BBA)-Biomembranes 1996, 1282, 63-70.

14. Smith, A. Effects of caffeine on human behavior. Food and chemical toxicology 2002, 40, 1243-1255.

15. Kawachi, I.; Willett, W.C.; Colditz, G.A.; Stampfer, M.J.; Speizer, F.E. A prospective study of coffee drinking and suicide in women. Arch. Intern. Med. 1996, 156, 521.

16. Griffiths, R.R.; Evans, S.M.; Heishman, S.J.; Preston, K.L.; Sannerud, C.; Wolf, B.; Woodson, P. Low-dose caffeine discrimination in humans. J. Pharmacol. Exp. Ther. 1990, 252, 970-978.

17. Furlong, F. Possible psychiatric significance of excessive coffee consumption. The Canadian Psychiatric Association Journal/La Revue de l'Association des psychiatres du Canada 1975

18. Stephenson, P.E. Physiologic and psychotropic effects of caffeine on man. A review. J. Am. Diet. Assoc. 1977, 71, 240-247.

19. Giovannucci, E. Meta-analysis of coffee consumption and risk of colorectal cancer. Am. J. Epidemiol. 1998, 147, 1043-1052.

20. Klatsky, A.L.; Armstrong, M.A.; Friedman, G.D. Coffee, tea, and mortality. Ann. Epidemiol. 1993, 3, 375-381.

21. Armstrong, M.A. Alcohol, Smoking, Cofee, and Cirrhosis. Am. J. Epidemiol. 1992, 136, 1248-1257.

22. Corrao, G.; Lepore, A.; Torchio, P.; Valenti, M.; Galatola, G.; D'Amicis, A.; Arico, S.; Di Orio, F. The effect of drinking coffee and smoking cigarettes on the risk of cirrhosis associated with alcohol consumption. Eur. J. Epidemiol. 1994, 10, 657-664.e

23. Stelzer, K.J.; Koh, W.; Kurtz, H.; Greer, B.E.; Griffin, T.W. Caffeine consumption is associated with decreased severe late toxicity after radiation to the pelvis. International Journal of Radiation Oncology* Biology* Physics 1994, 30, 411-417.

24. Johnston, K.L.; Clifford, M.N.; Morgan, L.M. Coffee acutely modifies gastrointestinal hormone secretion and glucose tolerance in humans: Glycemic effects of chlorogenic acid and caffeine. Am. J. Clin. Nutr. 2003, 78, 728-733.

25. Trugo, L.C.; Macrae, R. The use of the mass detector for sugar analysis of coffee products. Colloq. Sci. Int. Cafe, [C. R. ] 1985, 11th, 245-251.

26. Clifford, M. Chemical and physical aspects of green coffee and coffee products, In Coffee, Anonymous ; Springer: 1985; pp. 305-374.

27. Silwar, R.; Luellmann, C. The determination of mono- and disaccharides in green Arabica and Robusta coffees using high-performance liquid chromatography. Cafe, Cacao, The 1988, 32, 319-322.

28. Hughes, W.; Thorpe, T. Determination of Organic Acids and Sucrose in Roasted Coffee by Capiillary Gas Chromatography. J. Food Sci. 1987, 52, 1078-1083.

29. Noyes, R.; Chu, C. Material balance on free sugars in the production of instant coffee. 1993

30. Maga, J.A. Thermal decomposition of carbohydrates. An overview. ACS Symp. Ser. 1989, 409, 32-39.

31. Bradbury, A.G.W.; Halliday, D.J. Chemical structures of green coffee bean polysaccharides. J. Agric. Food Chem. 1990, 38, 389-392.

32. Nunes, F.M.; Coimbra, M.A.; Duarte, A.C.; Delgadillo, I. Foamability, Foam Stability, and Chemical Composition of Espresso Coffee As Affected by the Degree of Roast. J. Agric. Food Chem. 1997, 45, 3238-3243.

33. Rao, C.V.; Chou, D.; Simi, B.; Ku, H.; Reddy, B.S. Prevention of colonic aberrant crypt foci and modulation of large bowel microbial activity by dietary coffee fiber, inulin and pectin. Carcinogenesis 1998, 19, 1815-1819.

34. Wilson, A.; Petracco, M.; Illy, E. Some preliminary investigations of oil biosynthesis in the coffee fruit and its subsequent re-distribution within green and roasted beans. 1997

35. Folstar, P. The composition of wax and oil in green coffee beans. Versl. Landbouwkd. Onderz. 1976, 854, 65.

36. Maier, H.G. Kaffee. Parey: 1981; Vol. 18

37. Speer, K.; Sehat, N.; Montag, A. Fatty acids in coffee. 1993

38. Macrae, R. Nitrogenous components, In Coffee, Anonymous ; Springer: 1985; pp. 115- 152.

39. Steinhart, H.; Luger, A. Amino acid pattern of steam treated coffee. Colloq. Sci. Int. Cafe, [C. R. ] 1995, 16th, 278-285.

40. Morishita, H.; Takai, Y.; Yamada, H.; Fukuda, F.; Sawada, M.; Iwahashi, H.; Kido, R. Caffeoyltryptophan from green robusta coffee beans. Phytochemistry 1987, 26, 1195-1196.

41. Murata, M.; Okada, H.; Homma, S. Hydroxycinnamic acid derivatives and p-coumaroyl- (L)-tryprophan, a novel hydroxycinnamic acid derivative, from coffee beans. Bioscience Biotechnology and Biochemistry 1995, 59, 1887-1895.

42. Balyaya, K.; Clifford, M. Individual chlorogenic acids and caffeine contents in commercial grades of wet and dry processed Indian green robusta coffee. JOURNAL OF FOOD SCIENCE AND TECHNOLOGY-MYSORE- 1995, 32, 104-108.

43. Clifford, M. What factors determine the intensity of coffee’s sensory attributes. Tea Coffee Trade J 1987, 159, 35-39.

44. Maier, H. The acids of coffee. Proceedings 12th Asic College 1987, 229-237.

45. Mabrouk, A.F.; Deatherage, F.E. Organic acids in brewed coffee. Food Technol. (Chicago, IL, U. S. ) 1956, 10, 194-197.

46. Lentner, C.; Deatherage, F.E. Organic acids in coffee in relation to the degree of roast. Food Res. 1959, 24, 483-492.

47. Nakabayashi, T. Chemical studies on the quality of coffee. VI. Changes in organic acids and pH of roasted coffee. Nippon Shokuhin Kogyo Gakkaishi 1978, 25, 142-146.

48. Engelhardt, U.H.; Maier, H.G. Acids in coffee. XI. The proportion of individual acids in the total titratable acid. Z Lebensm Unters Forsch 1985, 181, 20-23.

49. Woodman, J.S.; Giddey, A.; Egli, R.H. Carboxylic acids of brewed coffee. Int. Colloq. Chem. Coffee, 3rd 1968, 137-143.

50. Hughes, W.J.; Thorpe, T.M. Determination of organic acids and sucrose in roasted coffee by capillary gas chromatography. J. Food Sci. 1987, 52, 1078-1083.

51. Engelhardt, U.H. Nichtflüchtige Säuren im Kaffee. Technische Universität Carolo- Wilhelmina zu Braunschweig: 1984

52. Clifford, M. Chemical and physical aspects of green coffee and coffee products, In Coffee, Anonymous ; Springer: 1985; pp. 305-374.

53. Scholze, A. Quantitative Bestimmung von Säuren im Kaffee durch Kapillar- Isotachophorese. Technische Universität Carolo-Wilhelmina zu Braunschweig: 1983;

54. Engelhardt, U.H.; Maier, H.G. Säuren des Kaffees. Zeitschrift für Lebensmittel- Untersuchung und Forschung 1985, 181, 206-209.

55. Balzer, H.H. Chemistry I: Non-Volatile Compounds, In Anonymous ; Blackwell Science Ltd: 2008; pp. 18-32.

56. Holscher, W.; Vitzthum, O.; Steinhart, H. Identification and sensorial evaluation of aroma-impact-compounds in roasted Colombian coffee. Café, cacao, thé 1990, 34, 205-212.

57. Wöhrmann, R.; Hojabr-Kalali, B.; Maier, H. Volatile minor acids in coffee. I. Contents of green and roasted coffee. Dtsch. Lebensm. -Rundsch. 1997, 93, 191-194.

58. Maeaettae, K.R.; Kamal-Eldin, A.; Toerroenen, A.R. High-Performance Liquid Chromatography (HPLC) Analysis of Phenolic Compounds in Berries with Diode Array and Electrospray Ionization Mass Spectrometric (MS) Detection: Ribes Species. J. Agric. Food Chem. 2003, 51, 6736-6744.

59. Kampmann, B.; Maier, H.G. Säuren des Kaffees. Zeitschrift für Lebensmittel- Untersuchung und Forschung 1982, 175, 333-336.

60. Hucke, J.; Maier, H.G. Quinic acid lactone in coffee. Z Lebensm Unters Forsch 1985, 180, 479-484.Journal Article

61. Galli, V.; Barbas, C. Capillary electrophoresis for the analysis of short-chain organic acids in coffee. Journal of Chromatography A 2004, 1032, 299-304.Journal Article

62. Ginz, M.; Balzer, H.H.; Bradbury, A.G.; Maier, H.G. Formation of aliphatic acids by carbohydrate degradation during roasting of coffee. European Food Research and Technology 2000, 211, 404-410.

63. Balzer, H.H. Chemistry I: Non-volatile compounds. Acids in coffee. Coffee 2001, 18-32.

64. Clifford, M.N. Chlorogenic acids and other cinnamates - nature, occurrence and dietary burden. J. Sci. Food Agric. 1999, 79, 362-372.

65. Clifford, M.N. Chlorogenic acids and other cinnamates - nature, occurrence, dietary burden, absorption and metabolism. J. Sci. Food Agric. 2000, 80, 1033-1043.

66. Anonymous IUPAC Commission on the Nomenclature of Organic Chemistry (CNOC) and IUPAC-IUB Commission on Biochemical Nomenclature (CBN). Nomenclature of cyclitols. Recommendations, 1973. Biochem J 1976, 153, 23-31.

67. Payen, S. Untersuchung des Kaffees. Annalen 1846, 60, 286-294.

68. Fischer, H.O.L.; Dangschat, G. Konstitution der Chlorogensäure (3. Mitteil. über Chinasäure Derivate). Berichte der deutschen chemischen Gesellschaft (A and B Series) 1932, 65, 1037-1040.

69. Jaiswal, R.; Patras, M.A.; Eravuchira, P.J.; Kuhnert, N. Profile and Characterization of the Chlorogenic Acids in Green Robusta Coffee Beans by LC-MSn: Identification of Seven New Classes of Compounds. J. Agric. Food Chem. 2010, 58, 8722-8737.

70. Clifford, M.N.; Johnston, K.L.; Knight, S.; Kuhnert, N. Hierarchical Scheme for LC-MSn Identification of Chlorogenic Acids. J. Agric. Food Chem. 2003, 51, 2900-2911.

71. Clifford, M.N.; Marks, S.; Knight, S.; Kuhnert, N. Characterization by LC-MSn of four new classes of p-coumaric acid-containing diacyl chlorogenic acids in green coffee beans. J. Agric. Food Chem. 2006, 54, 4095-4101.

72. Jaiswal, R.; Kuhnert, N. Hierarchical scheme for liquid chromatography/multi-stage spectrometric identification of 3,4,5-triacyl chlorogenic acids in green Robusta coffee beans. Rapid Communications in Mass Spectrometry 2010, 24, 2283-2294.

73. Jaiswal, R.; Sovdat, T.; Vivan, F.; Kuhnert, N. Profiling and Characterization by LC-MSn of the Chlorogenic Acids and Hydroxycinnamoylshikimate Esters in Mate (Ilex paraguariensis). J. Agric. Food Chem. 2010, 58, 5471-5484.

74. Clifford, M.N.; Knight, S.; Surucu, B.; Kuhnert, N. Characterization by LC-MSn of four new classes of chlorogenic acids in green coffee beans: Dimethoxycinnamoylquinic acids, diferuloylquinic acids, caffeoyl-dimethoxycinnamoylquinic acids, and feruloyl- dimethoxycinnamoylquinic acids. J. Agric. Food Chem. 2006, 54, 1957-1969.

75. Clifford, M.N.; Wu, W.; Kirkpatrick, J.; Jaiswal, R.; Kuhnert, N. Profiling and characterisation by liquid chromatography/multi-stage mass spectrometry of the chlorogenic acids in Gardeniae Fructus. Rapid Communications in Mass Spectrometry 2010, 24, 3109- 3120.

76. Clifford, M. Chlorogenic acids, In Coffee, Anonymous ; Springer: 1985; pp. 153-202.

77. Clifford, M.; Walker, R. Chlorogenic acids—confounders of coffee-serum cholesterol relationships. Food Chem. 1987, 24, 77-80.

78. Cartwright, R.; Roberts, E. Theogallin, a polyphenol occurring in tea. J. Sci. Food Agric. 1954, 5, 593-597.

79. Hara, Y.; Luo, S.; Wickremasinghe, R.; Yamanishi, T. Special issue on tea. Food Rev. Int. 1995, 11, 371-542.

80. Clifford, M.N.; Ramirez-Martinez, J.R. Chlorogenic acids and purine alkaloids contents of Maté (Ilex paraguariensis) leaf and beverage. Food Chem. 1990, 35, 13-21.

81. Spanos, G.A.; Wrolstad, R.E. Phenolics of apple, pear, and white grape juices and their changes with processing and storage. A review. J. Agric. Food Chem. 1992, 40, 1478-1487.

82. Mosel, H.; Herrmann, K. The phenolics of fruits. Zeitschrift für Lebensmittel- Untersuchung und Forschung 1974, 154, 6-11.

83. Igile, G.O.; Oleszek, W.; Jurzysta, M.; Burda, S.; Fafunso, M.; Fasanmade, A.A. Flavonoids from Vernonia amygdalina and their antioxidant activities. J. Agric. Food Chem. 1994, 42, 2445-2448.

84. Wald, B.; Wray, V.; Galensa, R.; Herrmann, K. Malonated flavonol glycosides and 3, 5- dicaffeoylquinic acid from pears. Phytochemistry 1989, 28, 663-664.

85. Risch, B.; Herrmann, K. Contents of hydroxycinnamic acid derivatives and catechins in pome and stone fruit. Z. Lebensm. Unters. 1988, 186, 225-230.

86. Schuster, B.; Herrmann, K. Hydroxybenzoic and hydroxycinnamic acid derivatives in soft fruits. Phytochemistry 1985, 24, 2761-2764.

87. Gao, L.; Mazza, G. Quantitation and distribution of simple and acylated anthocyanins and other phenolics in blueberries. J. Food Sci. 1994, 59, 1057-1059.

88. Koeppen, B.H.; Herrmann, K. Flavonoid glycosides and hydroxycinnamic acid esters of blackcurrants (Ribes nigrum). Zeitschrift für Lebensmittel-Untersuchung und Forschung 1977, 164, 263-268.

89. Risch, B.; Herrmann, K. Hydroxyzimtsäure-Verbindungen in Citrus-Früchten. Zeitschrift für Lebensmittel-Untersuchung und Forschung 1988, 187, 530-534.

90. Winter, M.; Brandl, W.; Herrmann, K. Determination of hydroxycinnamic acid derivatives in vegetables. Z. Lebensm. -Unters. Forsch. 1987, 184, 11-16.

91. Risch, B.; Herrmann, K.; Wray, V.; Grotjahn, L. 2'-(E)-O-p-coumaroylgalactaric acid and 2'-(E)-O-feruloylgalactaric acid in citrus. Phytochemistry 1987, 26, 509-510.

92. Okamura, S.; Watanabe, M. Determination of phenolic cinnamates in white wine and their effect on wine quality. Agric. Biol. Chem. 1981, 45, 2063-2070.

93. Fernandez de Simon, B.; Perez-Ilzarbe, J.; Hernandez, T.; Gomez-Cordoves, C.; Estrella, I. Importance of phenolic compounds for the characterization of fruit juices. J. Agric. Food Chem. 1992, 40, 1531-1535.

94. Hernandez, T.; Auśn, N.; Bartolomé, B.; Bengoechea, L.; Estrella, I.; Gómez-Cordovés, C. Variations in the phenolic composition of fruit juices with different treatments. Zeitschrift für Lebensmitteluntersuchung und-Forschung A 1997, 204, 151-155.

95. Brandl, W.; Herrmann, K. Hydroxycinnamic acid esters in cabbage-like vegetables and garden cress. Z. Lebensm. -Unters. Forsch. 1983, 176, 444-447.

96. Plumb, G.W.; Price, K.R.; Rhodes, M.J.C.; Williamson, G. Antioxidant properties of the major polyphenolic compounds in broccoli. Free Radical Res. 1997, 27, 429-435.

97. Brandl, W.; Herrmann, K.; Grotjahn, L. Hydroxycinnamoyl esters of malic acid in small radish (Raphanus sativus L. var. sativus). Z. Naturforsch. , C: Biosci. 1984, 39C, 515-520.

98. Tadera, K.; Mitsuda, H. Isolation and chemical structure of a new fluorescent compound in spinach leaves. Agr. Biol. Chem. 1971, 35, 1431-1435.

99. Winter, M.; Herrmann, K. Esters and glucosides of hydroxycinnamic acids in vegetables. J. Agric. Food Chem. 1986, 34, 616-620.

100. Malmberg, A.; Theander, O. Analysis of chlorogenic acid, coumarins and feruloylputrescine in different parts of potato tubers infected with Phoma. Swed. J. Agric. Res. 1984, 14, 63-70.

101. Brandl, W.; Herrmann, K. Occurrence of chlorogenic acids in potatoes. Z. Lebensm. - Unters. Forsch. 1984, 178, 192-194.

102. Shibuya, N. Phenolic acids and their carbohydrate esters in rice endosperm cell walls. Phytochemistry 1984, 23, 2233-2237.

103. Hemmerle, H.; Burger, H.; Below, P.; Schubert, G.; Rippel, R.; Schindler, P.W.; Paulus, E.; Herling, A.W. Chlorogenic Acid and Synthetic Chlorogenic Acid Derivatives: Novel Inhibitors of Hepatic Glucose-6-phosphate Translocase. J. Med. Chem. 1997, 40, 137-145.

104. del Castillo, M.D.; Ames, J.M.; Gordon, M.H. Effect of Roasting on the Antioxidant Activity of Coffee Brews. J. Agric. Food Chem. 2002, 50, 3698-3703.

105. Luzia, M.R.; Da Paixao, C.C.; Marcilio, R.; Trugo, L.C.; Quinteiro, L.M.C.; De Maria, C.A.B. Effect of 5-caffeoylquinic acid on soybean oil oxidative stability. Int. J. Food Sci. Technol. 1997, 32, 15-19.

106. Herling, A.W.; Burger, H.; Schwab, D.; Hemmerle, H.; Below, P.; Schubert, G. Pharmacodynamic profile of a novel inhibitor of the hepatic glucose-6-phosphatase system. Am. J. Physiol. 1998, 274, G1087-G1093.

107. Arion, W.J.; Canfield, W.K.; Ramos, F.C.; Schindler, P.W.; Burger, H.; Hemmerle, H.; Schubert, G.; Below, P.; Herling, A.W. Chlorogenic acid and hydroxynitrobenzaldehyde: new inhibitors of hepatic glucose 6-phosphatase. Arch. Biochem. Biophys. 1997, 339, 315-322.

108. Stich, H.F.; Rosin, M.P.; Bryson, L. Inhibition of mutagenicity of a model nitrosation reaction by naturally occurring phenolics, coffee and tea. Mutat. Res. 1982, 95, 119- 128.Journal Article

109. Johnston, K.L.; Clifford, M.N.; Morgan, L.M. Coffee acutely modifies gastrointestinal hormone secretion and glucose tolerance in humans: Glycemic effects of chlorogenic acid and caffeine. Am. J. Clin. Nutr. 2003, 78, 728-733.

110. Robinson, W.E., Jr.; Cordeiro, M.; Abdel-Malek, S.; Jia, Q.; Cho, S.A.; Reinecke, M.G.; Mitchell, W.M. Dicaffeoylquinic acid inhibitors of human immunodeficiency virus integrase: inhibition of the core catalytic domain of human immunodeficiency virus integrase. Mol. Pharmacol. 1996, 50, 846-855.

111. Mcdougall, B.; King, P.J.; Wu, B.W.; Hostomsky, Z.; Reinecke, M.G.; Robinson, W.E.,Jr Dicaffeoylquinic and dicaffeoyltartaric acids are selective inhibitors of human immunodeficiency virus type 1 integrase. Antimicrob. Agents Chemother. 1998, 42, 140-146.

112. Kwon, H.C.; Jung, C.M.; Shin, C.G.; Lee, J.K.; Choi, S.U.; Kim, S.Y.; Lee, K.R. A new caffeoyl quinic acid from Aster scaber and its inhibitory activity against human immunodeficiency virus-1 (HIV-1) integrase. Chem. Pharm. Bull. 2000, 48, 1796-1798.

113. Gupta, P.; Akanksha; Siripurapu, K.B.; Ahmad, A.; Palit, G.; Arora, A.; Maurya, R. Anti-stress constituents of Evolvulus alsinoides: an ayurvedic crude drug. Chem. Pharm. Bull. 2007, 55, 771-775.

114. Teramachi, F.; Koyano, T.; Kowithayakorn, T.; Hayashi, M.; Komiyama, K.; Ishibashi, M. Collagenase inhibitory quinic acid esters from Ipomoea pes-caprae. J. Nat. Prod. 2005, 68, 794-796.

115. Chang, C.; Lin, M.; Lee, S.; Liu, K.C.S.C.; Hsu, F.; Lin, J. Differential inhibition of reverse transcriptase and cellular DNA polymerase-Î± activities by lignans isolated from Chinese herbs, Phyllanthus myrtifolius Moon, and tannins from Lonicera japonica Thunb and Castanopsis hystrix. Antiviral Res. 1995, 27, 367-374.

116. Xiang, T.; Xiong, Q.; Ketut, A.I.; Tezuka, Y.; Nagaoka, T.; Wu, L.; Kadota, S. Studies on the hepatocyte protective activity and the structure-activity relationships of quinic acid and caffeic acid derivatives from the flower buds of Lonicera bournei. Planta Med. 2001, 67, 322- 325.

117. de Paulis, T.; Schmidt, D.E.; Bruchey, A.K.; Kirby, M.T.; McDonald, M.P.; Commers, P.; Lovinger, D.M.; Martin, P.R. Dicinnamoylquinides in roasted coffee inhibit the human adenosine transporter. Eur. J. Pharmacol. 2002, 442, 215-223.

118. Shearer, J.; Farah, A.; de Paulis, T.; Bracy, D.P.; Pencek, R.R.; Graham, T.E.; Wasserman, D.H. Quinides of roasted coffee enhance insulin action in conscious rats. J. Nutr. 2003, 133, 3529-3532.

119. Jaiswal, R. Synthesis and Analysis of the Dietary Relevant Isomers of Chlorogenic Acids, Their Derivatives and Hydroxycinnamates. 2012. PhD Thesis.

120. Panagiotakos, D.B.; Lionis, C.; Zeimbekis, A.; Makri, K.; Bountziouka, V.; Economou, M.; Vlachou, I.; Micheli, M.; Tsakountakis, N.; Metallinos, G.; Polychronopoulos, E. Long- term, moderate coffee consumption is associated with lower prevalence of diabetes mellitus among elderly non-tea drinkers from the Mediterranean Islands (MEDIS Study). Rev Diabet Stud 2007, 4, 105-111.

121. Daglia, M.; Racchi, M.; Papetti, A.; Lanni, C.; Govoni, S.; Gazzani, G. In Vitro and ex Vivo Antihydroxyl Radical Activity of Green and Roasted Coffee. J. Agric. Food Chem. 2004, 52, 1700-1704.

122. Tan, E.-.; Tan, C.; Fook-Chong, S.M.C.; Lum, S.Y.; Chai, A.; Chung, H.; Shen, H.; Zhao, Y.; Teoh, M.L.; Yih, Y.; Pavanni, R.; Chandran, V.R.; Wong, M.C. Dose-dependent protective effect of coffee, tea, and smoking in Parkinson's disease: a study in ethnic Chinese. J Neurol Sci 2003, 216, 163-167.

123. Lindsay, J.; Laurin, D.; Verreault, R.; Hebert, R.; Helliwell, B.; Hill, G.B.; McDowell, I. Risk factors for Alzheimer's disease: a prospective analysis from the Canadian Study of Health and Aging. Am J Epidemiol 2002, 156, 445-453.

124. Bekedam, E.K.; Roos, E.; Schols, H.A.; Van Boekel, M.A.J.S.; Smit, G. Low molecular weight melanoidins in coffee brew. J. Agric. Food Chem. 2008, 56, 4060-4067.

125. Bekedam, E.K.; Loots, M.J.; Schols, H.A.; Van Boekel, M.A.J.S.; Smit, G. Roasting effects on formation mechanisms of coffee brew melanoidins. J. Agric. Food Chem. 2008, 56, 7138-7145.

126. Crozier, A.; Jaganath, I.B.; Clifford, M.N. Dietary phenolics: chemistry, bioavailability and effects on health. Nat. Prod. Rep. 2009, 26, 1001-1043.

CHAPTER 2: LC-MSn identification of CGAs in green and roasted coffee

2.1 Introduction Mono-caffeoylquinic acids are generally identified in their tandem mass spectra in negative ion mode, a pseudomolecular ion [M-H] at m/z 353, which as precursor ions in tandem MS yields fragment ions at m/z 191 or m/z 173, which is characteristic to the quinic acid and dehydrated quinic acid moiety respectively. If these two ions are observed in MS2, it signifies the presence of mono-acyl CGAs and if observed in MS3, they are originated by di-acyl CGAs. Similarly, tri-acyl CGAs generate m/z 191 or m/z 173 in MS4. Retention times and resulting elution order on a reverse phase column help in the preliminary assessment of the regio-isomers of CGAs. For example, mono-acyl CGAs normally elute in the following order: 1-CQA (76) > 5-CQA (3)> 3-CQA (1)> 4-CQA (2). Also, the m/z value of the parent ion will in most cases reveal the chemical nature of the cinnamic acid moiety e.g. m/z 353- caffeoylquinic acid, m/z 367- feruloylquinic acid, m/z 515-dicaffeoylquinic acid etc.

Furthermore, targeted MSn experiments in the negative ion mode can be performed on the intact acyl moiety if observed. Quinic acid moiety stabilizes the negative charge hence, observed in negative ion mode very often. On the other hand, the side chain of the intact ion of the acyl moiety particularly stabilizes the positive charge in order to be observed in positive ion mode. In this manner, targeted MSn experiments in positive ion mode also help to elucidate the structure of the acyl substituent.

Additionally, confirmation of the molecular formulas of the CGAs is achieved by high resolution mass measurement typically through LC-TOF-MS or FT-ICR-MS techniques. For publication standards, mass error lower than five ppm is considered to be acceptable.

For the assignment of the regio-chemistry of the CGAs through tandem MS spectra, Clifford and Kuhnert’s hierarchical schemes can be exclusively employed. Direct comparison of the experimentally obtained fragment spectra with those already published allows the unambiguous assignment.

In this work, we have achieved to synthesize four diastereomers of (-)-quinic acid in order to distinguish quinic acid stereoisomers by tandem MS. This will enable us in future to discriminate between the hydroxycinnamic esters, which are bound to the diastereomers of quinic acid and do show their existence in roasted coffee. The detailed results are discussed in Chapter 4.

2.2 LC-MSn identification of CGAs in green coffee Presently, around 48 different chlorogenic acids in green coffee have been identified, which are characterized to their regio-isomeric level based on their fragmentation behavior in tandem MS and retention times in LC (Table 2.1). 1-5 Unlike in Artichoke, no mono- or di- acylated CGA shows the presence of acyl group on C1 of quinic acid in the case of green coffee beans. A representative total ion chromatogram in negative mode of the methanolic extract of the green coffee bean is shown in the Figure 2.1. With the peaks assigned to the CGAs identified.

[%] 100 3 TIC -All MS 6

15 13 2 14

26 22 18 1 5 25

4 36 9 41 0 10 20 30 40 50 Time [min]

Figure 2.1 TIC of green Robusta coffee extract in negative ion mode 6 (peak numbers corresponds to CGAs listed in Table 2.1)

Table 2.1 Chlorogenic acids identified in green coffee beans

No. Compound Abbreviation R3 R4 R5

1 3-O-caffeoylquinic acid 3-CQA C H H

2 4-O-caffeoylquinic acid 4-CQA H C H

3 5-O-caffeoylquinic acid 5-CQA H H C

4 3-O-feruloylquinic acid 3-FQA F H

5 4-O-feruloylquinic acid 4-FQA H F H

6 5-O-feruloylquinic acid 4-FQA H H F

7 3-O-p-coumaroylquinic acid 3-pCoQA pCo H H

8 4-O-p-coumaroylquinic acid 4-pCoQA H pCo H

9 5-O-p-coumaroylquinic acid 5-pCoQA H H pCo

10 3-O-dimethoxycinnamoylquinic acid 3-DQA D H H

11 4-O-dimethoxycinnamoylquinic acid 4-DQA H D H

12 5-O-dimethoxycinnamoylquinic acid 5-DQA H H D

13 3,4-di-O-caffeoylquinic acid 3,4-diCQA C C H

14 3,5-di-O-caffeoylquinic acid 3,5-diCQA C H C

15 4,5-di-O-caffeoylquinic acid 4,5-diCQA H C C

16 3,4-di-O-feruloylquinic acid 3,4-diFQA F F H

17 3,5-di-O-feruloylquinic acid 3,5-diFQA F H F

18 4,5-di-O-feruloylquinic acid 4,5-diFQA H F F

19 3,4-di-O-p-coumaroylquinic acid 3,4-dipCoQA pCo pCo H

20 3,5-di-O-p-coumaroylquinic acid 3,5-dipCoQA pCo H pCo

21 4,5-di-O-p-coumaroylquinic acid 4,5-dipCoQA H pCo pCo

22 3-O-feruloyl-4-O-caffeoylquinic acid 3F-4CQA F C H

23 3-O-caffeoyl-4-O-feruloylquinic acid 3C-4FQA C F H

24 3-O-feruloyl-5-O-caffeoylquinic acid 3F-5CQA F H C

25 3-O-caffeoyl-5-O-feruloylquinic acid 3C-5FQA C H F

26 4-O-feruloyl-5-O-caffeoylquinic acid 4F-5CQA H F C

27 4-O-caffeoyl-5-O-feruloylquinic acid 4C-5FQA H C F

28 3-O-dimethoxycinnamoyl-4-O-caffeoylquinic acid 3D-4CQA D C H

29 3-O-dimethoxycinnamoyl-5-O-caffeoylquinic acid 3D-5CQA D H C

30 4-O-dimethoxycinnamoyl-5-O-caffeoylquinic acid 4D-5CQA H D C

31 3-O-caffeoyl-4-O-dimethoxycinnamoylquinic acid 3C-4DQA C D H

32 3-O-caffeoyl-5-O-dimethoxycinnamoylquinic acid 3C-5DQA C H D

33 4-O-caffeoyl-5-O-dimethoxycinnamoylquinic acid 4C-5DQA H C D

34 3-O-dimethoxycinnamoyl-4-O-feruloylquinic acid 3D-4FQA D F H

35 3-O-dimethoxycinnamoyl-5-O-feruloylquinic acid 3D-5FQA D F H

36 4-O-dimethoxycinnamoyl-5-O-feruloylquinic acid 4D-5FQA H D F

37 3-O-p-coumaroyl-4-O-caffeoylquinic acid 3pCo-4CQA pCo C H

38 3-O-caffeoyl-4-O-p-coumaroylquinic acid 3C-4pCoQA C pCo H

39 3-O-p-coumaroyl-5-O-caffeoylquinic acid 3pCo-5CQA pCo H C

40 3-O-caffeoyl-5-O-p-coumaroylquinic acid 3C-5pCoQA C H pCo

41 4-O-caffeoyl-5-O-p-coumaroylquinic acid 4C-5pCoQA H C pCo

42 4-O-p-coumaroyl-5-O-caffeoylquinic acid 4pCo-5CQA H pCo C

43 3-O-p-coumaroyl-4-O-feruloylquinic acid 3pCo-4FQA pCo F H

44 3-O-p-coumaroyl-5-O-feruloylquinic acid 3pCo-5FQA pCo H F

45 4-O-p-coumaroyl-5-O-feruloylquinic acid 4pCo-5FQA H pCo F

46 4-O-dimethoxycinnamoyl-5-O-p-coumaroylquinic acid 4D-5pCoQA H D pCo

47 3-O-p-coumaroyl-4-O-dimethoxycinnamoylquinic acid 3pCo-4DQA pCo D H

48 3-O-p-coumaroyl-5-O-dimethoxycinnamoylquinic acid 3pCo-5DQA pCo H D

C = caffeoyl; F = feruloyl; D = dimethoxycinnamoyl; H = hydrogen; pCo = p-coumaroyl.

Mono-acyl CGAs being more polar, elute earlier than the di-acylated CGAs in the reverse phase packings. 1-53, 4, 7, 8 Generally, water, acetic acid, methanol, and acetonitrile have been used as mobile phases for HPLC with reverse phase stationary phases like C18, C8, phenylhexyl, and diphenyl etc. 1-4, 7-12 Clifford et al. developed the hierarchical scheme for the identification of the CGAs considering relative hydrophobicity, fragmentation patterns and retention times. 1-4, 8 In this work, we have applied the same strategies to identify mono- and di-acylated CGAs especially, when we studied the acyl migration in CGAs during roasting, brewing and by base treatment (Chapter 3). Figure 2.2 Shows the postulated fragment structures in negative ion mode generated by different chlorogenic acids.

The following guidelines could be useful to identify mono-acylated CGAs:

1. 3-acylated mono-acyl CGAs such as, 3-FQA (4), 3-pCoQA (7), 3-DQA (10) and 3- sinapoylquinic acid (49) (3-SiQA) base peak ions in MS2 and MS3 are generated from 1, 6 cinnamic acid moiety (A4, A1, A2 and A3 respectively). Whereas, remaining mono- acyl CGAs such as, 3-CQA generates base peak ions in MS2 and MS3, which are derivatives of quinic acid moiety. Figure 2.3 shows representative MSn spectra of 3- acyl CGAs. 2. 4-acylated CGAs can be readily identified by the dehydrated base peak in MS2 at m/z 3 1-5 173, followed by MS base peak at m/z 93 (Q6) and Q7 at m/z 111. Figure 2.4 shows representative MSn spectra of 4-acyl CGAs.

3. 3-CQA (1) and 5-CQA (3) generate same base peak ion in MS2 at m/z 191. However, 2 intense secondary ion A1 at m/z 179 in the MS of 3-CQA allows the distinction between the two (Figure 2.5).

O O O O HOOC OH HOOC HOOC HOOC OH OH OH OH OH O OH O O C HC Q1 Q2 Q3 Q4 CH

R1 R 2 OH O O O O

OH OH OH O

Q5 Q6 Q7 Q8

OH OH OH R R O R2 1 R2 1 R2

HC HC HC CH CH CH C O O A B C

Fragment R1 R2 Cinnamic acid Accurate mass

Q1 191.06

Q2 173.04

Q3 172.04

Q4 OH H Caffeic 335.08

OCH3 H Ferulic 349.08

H H p-Coumaric 319.08

OCH3 OCH3 Sinapic 379.10

Q5 85.03

Q6 93.03

Q7 111.04

Q8 127.04

A1 OH H Caffeic 179.04

A2 OCH3 H Ferulic 193.04

A3 H H p-Coumaric 163.04

A4 OCH3 OCH3 Sinapic 223.06

B1 OH H Caffeic 135.04

B2 OCH3 H Ferulic 149.04

B3 H H p-Coumaric 119.04

C1 Ferulic 134.04

C2 Sinapic 164.04

Figure 2.2 Structures of the fragments generated by quinic and cinnamic acid derivatives 15

[%] 2 3-pCoQA (7) MS

162.6 100

118.8 0 190.7 MS3 118.8 100

0 50 100 150 200 250 300 350 m/z

Figure 2.3 (Continued)

[%] 2 3-FQA (4) MS 192.8 100

134.0 0 MS3 133.8 100

148.9 0 50 100 150 200 250 300 350 m/z

[%] 3-CQA (1) MS2 190.8 100

178.9 135.0 0 MS3 100 126.9 172.8 85.3

0 50 100 150 200 250 300 350 m/z

[%] MS2 3-SiQA (49) 222.9 100

164.0 0 MS3 100 163.9

148.9 178.9 0 100 150 200 250 300 350 400 m/z

Figure 2.3 MS2 and MS3 spectra of 3-acyl chlorogenic acids in negative ion mode

[%] MS2 4-CQA (2) 172.9 100 178.9

0 191.0 0 MS3 93.1 100

110.9 154.8 0 100 150 200 250 300 350 m/z

[%] 2 4-pCoQA (8) MS 172.7 100

0 MS3 93.0 100 110.8 154.7 71.2 136.6 0 50 100 150 200 250 300 350 m/z

Figure 2.4 MS2 and MS3 spectra of 4-acyl chlorogenic acids in negative ion mode (Continued)

[%] 2 4-FQA (5) MS 172.8 100

192.8 0 MS3 93.1 100

111.0 71.4 154.8 0 100 150 200 250 300 350 m/z

[%] 4-SiQA (50) MS2 172.9 100

222.9 0 MS3 100 93.3

111.1 71.6

154.9 0 100 150 200 250 300 350 400 450 m/z

Figure 2.4 MS2 and MS3 spectra of 4-acyl chlorogenic acids in negative ion mode

[%] 5-CQA (3) MS2 190.8 100

0 MS3 126.9 100 85.2

111.1 172.9

0 50 100 150 200 250 300 350 m/z

Figure 2.5 (Continued)

[%] 2 5-SiQA (51) MS 100 190.9

222.9 0 MS3 100 127.0

172. 85.493.3

109.1 71. 0 100 150 200 250 300 350 400 m/z

Figure 2.5 MS2 and MS3 spectra of 5-acyl chlorogenic acids in negative ion mode

As mentioned earlier, mono-acylated CGAs elute before di-acyl CGAs. Although in green coffee, 1-acylated CGAs are not identified, they appear first in the chromatogram in both mono- and di-acylated CGAs. The elution order for di-acylated CGAs is as follows: 1,3 (80)>1,4 (81)>1,5 (82)>3,4 (13)>3,5 (14)>4,5 (15). 6, 13, 14 This elution order was first established by Clifford and then Jaiswal pointed out that 3,5- elutes first followed by 3,4- and 4,5-di-acylated CGAs. 15

As in the case of mono-acylated CGAs we can lay out some guidelines for identification of di-acyl CGAs along with their regio-chemistry in negative ion mode. They are as follows:

1. Similar to the mono-acylated CGAs, di-acyl CGAs generate equivalent parent ion in MS1 as, [di-acyl CGA -H+]-. In MS2, all di-acyl CGAs either produce [di-acyl CGA – + - + - 1- cinnamoyl -H ] (Figure 2.6) or [diacyl CGA – cinnamoyl – H2O -H ] (Figure 2.7). 3, 6, 13 2. Vicinal di-caffeoylquinic acids such as, 3,4-diCQA (13) and 4,5-diCQA (15) remain consistent with the fragmentation behavior of the 4-acylated mono-acyl CGAs 3 4 producing Q2 as a base peak in MS at m/z 173 followed by strong MS ions at m/z 93

(Q6) and Q7 at m/z 111. (Figure 2.6 ) 3. The difference in two vicinal di-caffeoylquinic acids arises from the intensity of the 2 fragment Q4, which is high in the case of the MS of 3,4-diCQA(13) and barely 2 3 detected in the MS of 4,5-diCQA(15). Similarly, the 3,4-isomer produces Q1 in MS 4 and Q7 in MS with approximately double the intensities if compared to the 4,5- isomer. 4

On the other hand, tandem MS of the 3,5-diCQA (14) remains consistent with 3-CQA (1) and 3 5-CQA (3), by producing base peak Q1 at m/z 191, supported by strong MS ions at m/z 86

(Q5), m/z 127 (Q8), and m/z 172 (Q3). Fragmentation spectra are shown in the Figure 2.6.

[%] MS2 3,4-diCQA (13) 353.0 100 172.9 254.9 0 MS3 172. 100 9 178.9 190.9 134.9 0 MS4 100 93.1 154.8 0 100 150 200 250 300 350 400 450 500 m/z

[%] 2 3,5-diCQA (14) MS 353.1 100

190.9 0 MS3 190.8 100 178.9 0 135.0 MS4 100 85.3111.0 170.8 0 100 150 200 250 300 350 400 450 500 m/z

Figure 2.6 MS2, MS3, and MS4 spectra of 3,4-diCQA, 3,5-diCQA, and 4,5-diCQA in negative ion mode (m/z 515) (Continued)

[%] 2 4,5-diCQA (15) MS 353.1 100

172.9 202.9 255.0 299.0 0 MS3 172.9 100 178.9 135.0 191.0 0 MS4 100 93.1 154.8 0 100 150 200 250 300 350 400 450 500 m/z

Figure 2.6 MS2, MS3, and MS4 spectra of 3,4-diCQA, 3,5-diCQA, and 4,5-diCQA in negative ion mode (m/z 515)

[%] 3,4-diFQA (16) MS2 349.1 100

172.8 472.7 0 MS3 192.8 100 172.8 269.0 0 MS4 100 133.9 148.9 0 100 150 200 250 300 350 400 450 500 550 m/z

[%] MS2 3D-4FQA (34) 348.9 100

206.7 486.7 395.0 261.0 574.6 0 MS3 100 172.8 268.8 304.9 133.8

0 100 20 300 400 500 600 m/z 0 Figure 2.7 MS2, MS3, and MS4 spectra of 3,4-diFQA 16 and 3D-4FQA 34 in negative ion mode (m/z 543 and m/z 557, respectively)

Table 2.2 MS2 and MS3 data of monoacyl CGAs in negative ion mode

No. CGA MS1 MS2 MS3

Parent ion Base peak Secondary peak Base peak Secondary peak

m/z m/z int m/z int m/z int m/z m/z int m/z int

1 3-CQA 353.1 190.9 178.5 50 134.9 7 85.3 127.0 71 172.9 67

2 4-CQA 353.1 172.9 178.9 60 190.8 20 135.0 9 93.2 111.0 48

3 5-CQA 353.2 190.0 178.5 5 135.0 15 85.2 126.9 66 172.9 27

4 3-FQA 367.2 192.9 191.5 2 173.2 2 133.9 148.9 23

5 4-FQA 367.2 172.9 192.9 16 93.1 111.5 44

6 5-FQA 367.2 190.9 172.9 2 85.2 126.9 70

7 3-pCoQA 337.1 162.9 190.0 5 118.9

8 4-pCoQA 337.1 172.7 93.0 111.0 61

9 5-pCoQA 337.2 190.9 162.9 5 85.2

Table 2.3 MS2, MS3, and MS4 data of diacyl CGAs in negative ion mode (n.d. = not detected)

No. CGA MS1 MS2 MS3

Parent ion Base peak Secondary peak Base peak Secondary peak

m/z m/z int m/z int m/z int m/z m/z int m/z int m/z int

13 3,4-diCQA 515.2 353.1 335.1 4 172.9 20 172.9 178.9 68 191.0 32 135.1 9

15 4,5-diCQA 515.2 353.1 335.1 2 172.9 6 172.9 178.9 76 190.9 9 135.0 19

18 4,5-diFQA 543.2 367.1 349.1 35 172.9 178.9 60 190.8 20 135.0 9

24 3F-5CQA 529.2 367.1 353.1 60 349.0 32 335.0 32 192.7 172.6 36 133.8 36

26 4F-5CQA 529.2 367.1 335.0 4 172.7 21 172.9 192.9 71 133.8 8

27 4C-5FQA 529.1 353.1 367.1 25 172.9 178.9 49 190.8 35 134.7 10

40 3C-5pCoQA 499.0 353.1 337.0 15 190.7

41 4C-5pCoQA 499.3 353.0 172.8 15 172.9 178.7 66 190.6 29 134.8 12

42 4pCo-5CQA 499.1 337.1 335.1 3 172.7 59 172.9 162.6 8

52 1,3-diCQA 515.2 353.1 335.1 2 173.0 4 190.9 179.0 60 135.1 6

53 1,4-diFQA 543.2 349.1 367.1 25 172.9 17 268.8 9 192.9 172.9 31 268.8 9 133.8 22

54 1C-3FQA 529.1 367.1 353.1 15 192.7 172.6 13 178.6 5 133.8 18

No. CGA MS1 MS4

Parent ion Base peak Secondary peak

m/z m/z int m/z int

13 3,4-diCQA 515.2 93.2 111.1 30

15 4,5-diCQA 515.2 93.1 111.0 20

18 4,5-diFQA 543.2 93.1 111.1 40

24 3F-5CQA 529.2 133.7 149.0 19

26 4F-5CQA 529.2 93.2

27 4C-5FQA 529.1 93.2 127.0 n.d.

40 3C-5pCoQA 499.0 85.2 93.0 70 126.9 99

41 4C-5pCoQA 499.3 93.2

42 4pCo-5CQA 499.1 93.2 111.1 98

52 1,3-diCQA 515.2 85.1 111.1 86 172.9 60

53 1,4-diFQA 543.2

54 1C-3FQA 529.1 133.7 149.0 16 127.0 6

2.3 LC-MSn identification of CGAs in roasted coffee As mentioned earlier, roasting decreases the CGAs content as well as it transforms CGAs to their isomers by epimerization or by acyl group migration and degrades them to corresponding cinnamic acid derivatives and quinic acid. Additionally, roasting gives rise to 16-21 the dehydrated CGA derivatives in the form of lactones or shikimates. C1- acylated mono- or di-acyl CGAs are not observed in green coffee since; they are essentially the products of roasting. The workgroup of Kuhnert has been able to develop LC-MSn method for the identification and distinction between cinnamoylshikimates and CGA lactones (CGLs) to their regio-isomeric level. 14 In present work, ten new chlorogenic acids have been identified in roasted coffee, in which organic acids were found conjugated with quinic acid. Detailed results are presented in Chapter 4. Table 2.4 and Table 2.5 summarize CGAs, CGLs and CSAs identified in the roasted coffee. CGLs and CSAs are characterized by their parent pseudo molecular ion at m/z 335 followed by characteristic fragment spectra (Table 2.5).14, 22

Table 2.4 Chlorogenic acids identified in roasted coffee 15

No. Compound Abbreviation R1 R3 R4 R5 1 3-O-caffeoylquinic acid 3-CQA H C H H 2 4-O-caffeoylquinic acid 4-CQA H H C H 3 5-O-caffeoylquinic acid 5-CQA H H H C 4 3-O-feruloylquinic acid 3-FQA H F H H 5 4-O-feruloylquinic acid 4-FQA H H F H 6 5-O-feruloylquinic acid 5-FQA H H H F 7 3-O-p-coumaroylquinic acid 3-pCoQA H pCo H H 8 4-O-p-coumaroylquinic acid 4-pCoQA H H pCo H 9 5-O-p-coumaroylquinic acid 5-pCoQA H H H pCo 10 3-O-dimethoxycinnamoylquinic acid 3-DQA H D H H 11 4-O-dimethoxycinnamoylquinic acid 4-DQA H H D H 12 5-O-dimethoxycinnamoylquinic acid 5-DQA H H H D 13 3,4-di-O-caffeoylquinic acid 3,4-diCQA H C C H 14 3,5-di-O-caffeoylquinic acid 3,5-diCQA H C H C 15 4,5-di-O-caffeoylquinic acid 4,5-diCQA H H C C 16 3,4-di-O-feruloylquinic acid 3,4-diFQA H F F H 17 3,5-di-O-feruloylquinic acid 3,5-diFQA H F H F 18 4,5-di-O-feruloylquinic acid 4,5-diFQA H H F F 22 3-O-feruloyl-4-O-caffeoylquinic acid 3F-4CQA H F C H 23 3-O-caffeoyl-4-O-feruloylquinic acid 3C-4FQA H C F H 25 3-O-caffeoyl-5-O-feruloylquinic acid 3C-5FQA H C H F 26 4-O-feruloyl-5-O-caffeoylquinic acid 4F-5CQA H H F C 28 3-O-dimethoxycinnamoyl-4-O-caffeoylquinic acid 3D-4CQA H D C H 29 3-O-dimethoxycinnamoyl-5-O-caffeoylquinic acid 3D-5CQA H D H C 30 4-O-dimethoxycinnamoyl-5-O-caffeoylquinic acid 4D-5CQA H H D C 33 4-O-caffeoyl-5-O-dimethoxycinnamoylquinic acid 4C-5DQA H H C D 36 4-O-dimethoxycinnamoyl-5-O-feruloylquinic acid 4D-5FQA H H D F

37 3-O-p-coumaroyl-4-O-caffeoylquinic acid 3pCo-4CQA H pCo C H 38 3-O-caffeoyl-4-O-p-coumaroylquinic acid 3C-4pCoQA H C pCo H 39 3-O-p-coumaroyl-5-O-caffeoylquinic acid 3pCo-5CQA H pCo H C 42 4-O-p-coumaroyl-5-O-caffeoylquinic acid 4pCo-5CQA H H pCo C 43 3-O-p-coumaroyl-4-O-feruloylquinic acid 3pCo-4FQA H pCo F H 45 4-O-p-coumaroyl-5-O-feruloylquinic acid 4pCo-FCQA H H pCo F 56 3-O-sinapoyl-4-O-caffeoylquinic acid 3Si-4CQA H Si C H 57 3-O-caffeoyl-4-O-sinapoylquinic acid 3C-4SiQA H C Si H 60 3-O-feruloyl-4-O-sinapoylquinic acid 3F-4SiQA H F Si H 64 3-O-trimethoxycinnamoyl-5-O-feruloylquinic acid 3T-5FQA H T H F 76 1-O-caffeoylquinic acid 1-CQA C H H H 77 1-O-feruloylquinic acid 1-FQA F H H H 78 1-O-p-coumaroylquinic acid 1-pCoQA pCo H H H 79 1-O-dimethoxycinnamoylquinic acid 1-DQA D H H H 80 1,3-di-O-caffeoylquinic acid 1,3-diCQA C C H H 81 1,4-di-O-caffeoylquinic acid 1,4-diCQA C H C H 82 1,5-di-O-caffeoylquinic acid 1,5-diCQA C H H C 83 1-O-caffeoyl-3-O-feruloylquinic acid 1C-3FQA C F H H 84 1-O-caffeoyl-4-O-feruloylquinic acid 1C-4FQA C H F H 85 1-O-caffeoyl-4-O-dimethoxycinnamoylquinic acid 1C-4DQA C H D H 86 4-O-feruloyl-5-O-dimethoxycinnamoylquinic acid 4F-5DQA H H F D 87 1-O-caffeoyl-3-O-sinapoylquinic acid 1C-3SiQA C Si H H 88 1-O-feruloyl-4-O-sinapoylquinic acid 1F-4SiQA F H Si H 89 1-O-feruloyl-3-O-sinapoylquinic acid 1F-3SiQA F Si H H 90 1-O-caffeoyl-3-O-trimethoxycinnamoylquinic acid 1C-3TQA C T H H C = caffeoyl; D = dimethoxycinnamoyl; F = feruloyl; pCo = p-coumaroyl; H = hydrogen; T = trimethoxycinnamoyl.

Table 2.5 Negative ion mode MS2, MS3 and MS4 fragmentation data for the cinnamoylshikimate esters and chlorogenic acid lactones 22

No. Compd. MS1 MS2 MS3

Parent ion Base peak Secondary peak Base peak Secondary peak

m/z m/z int m/z int m/z int m/z m/z int m/z int m/z int

91 3-CSA 335.1 178.9 178.5 50 134.9 7 85.3 127.0 71 172.9 67

92 4-CSA 335.1 178.9 178.9 60 190.8 20 135.0 9 93.2 111.0 48

93 5-CSA 335.1 178.9 178.5 5 135.0 15 85.2 126.9 66 172.9 27

94 1-CQL 335.1 160.8 172.8 67 132.8 14 132.8

95 3-CQL 335.1 160.8 134.8 82 132.8

96 4-CQL 335.1 160.8 134.8 17 132.9

97 3-FSA 349.1 192.9 155.0 10 148.9 177.9 73 134.0 78

98 4-FSA 349.1 192.9 174.9 24 154.9 24 137.0 13 148.9 177.9 63 134.0 75

99 5-FSA 349.0 192.9 155.0 27 148.9 177.9 81 134.0 71

100 1-FQL 349.0 172.7 192.7 56 175.0 88 159.7 19 93.1 159.7 22 110.9 32

101 3-FQL 349.0 174.7 192.7 42 148.7 65 133.8 32 159.7

102 4-FQL 349.1 174.7 192.7 41 148.7 13 159.7 20 159.7

103 3-DSA 363.1 206.8 154.8 70 136.8 45 294.7 35 115.0 130.8 93

104 4-DSA 363.1 154.8 206.8 15 136.7 50 136.8 111.0 12 93.0 15

105 5-DSA 363.1 154.8 206.8 50 136.7 50 110.9 16 136.8 111.0 15 93.0 20

106 1-DQL 363.1 206.8 154.8 17 132.8 191.8 19 162.8 66 148.9 20

107 3-DQL 363.1 206.8 148.8 190.8 23 162.8 32 134.8 34

108 4-DQL 363.1 206.8 148.8 191.8 52 130.8 67

References

1. Clifford, M.N.; Johnston, K.L.; Knight, S.; Kuhnert, N. Hierarchical Scheme for LC-MSn Identification of Chlorogenic Acids. J. Agric. Food Chem. 2003, 51, 2900-2911.

2. Clifford, M.N.; Knight, S.; Surucu, B.; Kuhnert, N. Characterization by LC-MSn of four new classes of chlorogenic acids in green coffee beans: Dimethoxycinnamoylquinic acids, diferuloylquinic acids, caffeoyl-dimethoxycinnamoylquinic acids, and feruloyl- dimethoxycinnamoylquinic acids. J. Agric. Food Chem. 2006, 54, 1957-1969.

3. Clifford, M.N.; Marks, S.; Knight, S.; Kuhnert, N. Characterization by LC-MSn of four new classes of p-coumaric acid-containing diacyl chlorogenic acids in green coffee beans. J. Agric. Food Chem. 2006, 54, 4095-4101.

4. Clifford, M.N.; Knight, S.; Kuhnert, N. Discriminating between the six isomers of dicaffeoylquinic acid by LC-MSn. J. Agric. Food Chem. 2005, 53, 3821-3832.

5. Alonso-Salces, R.M.; Guillou, C.; Berrueta, L.A. Liquid chromatography coupled with ultraviolet absorbance detection, electrospray ionization, collision-induced dissociation and tandem mass spectrometry on a triple quadrupole for the on-line characterization of polyphenols and methylxanthines in green coffee beans. Rapid Commun. Mass Spectrom. 2009, 23, 363-383.

6. Jaiswal, R.; Patras, M.A.; Eravuchira, P.J.; Kuhnert, N. Profile and Characterization of the Chlorogenic Acids in Green Robusta Coffee Beans by LC-MSn: Identification of Seven New Classes of Compounds. J. Agric. Food Chem. 2010, 58, 8722-8737.

7. Clifford, M.N.; Wang, Z.; Kuhnert, N. Profiling the chlorogenic acids of Aster by HPLC- MSn. Phytochem. Anal. 2006, 17, 384-393.

8. Clifford, M.N.; Wu, W.; Kuhnert, N. The chlorogenic acids of Hemerocallis. Food Chem. 2005, 95, 574-578.

9. Clifford, M.N.; Kirkpatrick, J.; Kuhnert, N.; Roozendaal, H.; Salgado, P.R. LC-MSn analysis of the cis isomers of chlorogenic acids. Food Chem. 2008, 106, 379-385.

10. Clifford, M.N.; Wang, Z.; Kuhnert, N. Profiling the chlorogenic acids of Aster by HPLC- MSn. Phytochem. Anal. 2006, 17, 384-393.

11. Clifford, M.N.; Stoupi, S.; Kuhnert, N. Profiling and Characterization by LC-MSn of the Galloylquinic Acids of Green Tea, Tara Tannin, and Tannic Acid. J. Agric. Food Chem. 2007, 55, 2797-2807.

12. Clifford, M.N.; Wu, W.; Kirkpatrick, J.; Kuhnert, N. Profiling the Chlorogenic Acids and Other Caffeic Acid Derivatives of Herbal Chrysanthemum by LC-MSn. J. Agric. Food Chem. 2007, 55, 929-936.

13. Clifford, M.N.; Wu, W.; Kirkpatrick, J.; Jaiswal, R.; Kuhnert, N. Profiling and characterisation by liquid chromatography/multi-stage mass spectrometry of the chlorogenic

51 acids in Gardeniae Fructus. Rapid Communications in Mass Spectrometry 2010, 24, 3109- 3120.

14. Jaiswal, R.; Sovdat, T.; Vivan, F.; Kuhnert, N. Profiling and Characterization by LC-MSn of the Chlorogenic Acids and Hydroxycinnamoylshikimate Esters in Mate (Ilex paraguariensis). J. Agric. Food Chem. 2010, 58, 5471-5484.

15. Jaiswal, R. Synthesis and Analysis of the Dietary Relevant Isomers of Chlorogenic Acids, Their Derivatives and Hydroxycinnamates. 2012. PhD Thesis.

16. Bennat, C.; Engelhardt, U.H.; Kiehne, A.; Wirries, F.M.; Maier, H.G. HPLC ANALYSIS OF CHLOROGENIC ACID LACTONES IN ROASTED COFFEE. Zeitschrift Fur Lebensmittel-Untersuchung Und-Forschung 1994, 199, 17-21.

17. Bicchi, C.P.; Binello, A.E.; Pellegrino, G.M.; Vanni, A.C. Characterization of Green and Roasted Coffees through the Chlorogenic Acid Fraction by HPLC-UV and Principal Component Analysis. J. Agric. Food Chem. 1995, 43, 1549-1555.

18. del Castillo, M.D.; Ames, J.M.; Gordon, M.H. Effect of Roasting on the Antioxidant Activity of Coffee Brews. J. Agric. Food Chem. 2002, 50, 3698-3703.

19. Farah, A.; De Paulis, T.; Trugo, L.C.; Martin, P.R. Effect of roasting on the formation of chlorogenic acid lactones in coffee. J. Agric. Food Chem. 2005, 53, 1505-1513.

20. Farah, A.; De Paulis, T.; Moreira, D.P.; Trugo, L.C.; Martin, P.R. Chlorogenic acids and lactones in regular and water-decaffeinated arabica coffees. J. Agric. Food Chem. 2006, 54, 374-381.

21. Hucke, J.; Maier, H.G. Quinic acid lactone in coffee. Z Lebensm Unters Forsch 1985, 180, 479-484.

22. Jaiswal, R.; Matei, M.F.; Ullrich, F.; Kuhnert, N. How to distinguish between cinnamoylshikimate esters and chlorogenic acid lactones by liquid chromatography-tandem mass spectrometry. Journal of Mass Spectrometry 2011, 46, 933-942.

CHAPTER 3: Acyl migration in mono- and di-caffeoylquinic acids under basic and aqueous acidic conditions and dry roasting conditions

3.1 Introduction Coffee is the most valued agricultural commodity in terms of the economic aspects of the exports from the developing coffee producing countries, accounting to ca. 8 million metric tonnes per year. Approximately, 2.3 billion cups of coffee are consumed worldwide per day.1 Coffea Arabica (known as Arabica coffee) and Coffea canephora (known as Robusta coffee) are the two types of coffee holding 70% and 30%, respectively of the total coffee market in the world.2 Chlorogenic acids are present in the range of 6-12% of the dry weight of the green coffee bean.3

Chlorogenic acids (CGAs) are a large group of esters formed between one or more cinnamic acid derivatives and D-(-)-quinic acid. CGAs are classified on the basis of the number of the cinnamoyl residues esterified with the quinic acid as well as the functional groups present on the aromatic moiety of the cinnamoyl residues. Out of the total content of the CGAs in green coffee, 5-O-caffoylquinic acid (3) comprises about 50%. Other subclasses like caffoylquinic acids, dicaffoylquinic acids, feruloylquinic acids and p-coumaroylquinic acids contribute to a large extent to the other 50% of the total CGAs present in coffee. CGAs are very important plant secondary metabolites due to their pharmacological properties, like antioxidant property,4 anti-hepatitis B virus activity,5 antispasmodic activity,3 anti-diabetic activity,6 inhibition of the HIV-1 integrase 7,8 and inhibition of the mutagenicity of carcinogenic compounds.3

The various roasting conditions affect the concentration and composition of the CGA content in the green coffee. For every 1% of the dry matter of the total CGA content in the green coffee beans, 8-10% of the original CGAs are transformed or decomposed into respective cinnamic acid derivatives and quinic acid. 1,9 The lightest drinkable roast (the so-called ‘Cinnamon’ roast) involves roasting of green coffee beans at around 180 0C until the coffee beans just encounter the ‘first crack’. It was reported by Clifford et al. that during the early stage of the roasting process, transformations such as isomerization (acyl migration) or hydrolysis of the ester bond, take place in the CGAs. Later, chemical transformations like decarboxylation in cinnamoyl moieties to produce a number of phenylindans, epimerization at the quinic acid and lactonization take place. 1,2 Clifford et al. also reported the base hydrolysis induced intramolecular isomerization and transesterification in 5-O-caffoylquinic acid (3), 3-

O-caffoylquinic acid (2), 4-O-caffoylquinic acid (4), 3,4-di-O-caffoylqunic acid (8), 3,5-di-O- caffoylqunic acid (9), 4,5-di-O-caffoylqunic acid (10), 5-O-p-coumaroylquinic acid (13) and 5-O-feruloylqunic acid (23), in which the identification of some of the transformed products was based on the putative conclusions acquired by analytical HPLC. 10,11 Dawidowicz et al. found nine transformation products of 5-O-caffoylquinic acid (3) after five hours of reflux in an acid-water solution including two water addition products. This study only incorporated 5- CQA. 12 No comprehensive mechanistic study has been previously reported, which comments on the intra- versus inter-molecular acyl migration under different conditions incorporating all major commercially available regio-isomers of mono- and di-caffeoyl chlorogenic acids.

The complexity in the data interpretation for the structural analyses of the CGAs present in the roasted coffee melanoidines arises from the regio- and stereoisomeric compounds in the natural sources. For this reason, model roasting experiments on the commercially available mono- and dicaffeoylquinic acids were attempted in the present work in order to study the transformations taking place in CGAs during the early roasting stages. Also, isomerization (acyl migration) was induced by both base hydrolysis and simple hydrolysis (brewing) in these reference standards to observe the isomeric transformations on the basis of relative quantification. Recently, tandem mass spectrometry has allowed accurate structural assignment and identification of the CGA regioisomers. The advantage of multi-dimensional speciphicity of LC-MSn enabled isomeric resolution and relative quantification of the early roasting transformations in the CGAs. 13

3.2 Materials and methods Chemicals and materials

All the chemicals (analytical grade) were purchased from Sigma-Aldrich (Bremen, Germany). Commercially available mono- and di-caffeoylquinic acids such as 5-O-caffoylquinic acid (3), 3-O-caffoylquinic acid (neo-chlorogenic acid) (2), 4-O-caffoylquinic acid (crypto-chlorogenic acid) (4), 1,3-di-O-caffoylqunic acid (cynarin) (5), 3,4-di-O-caffoylqunic acid (8), 3,5-di-O- caffoylqunic acid (9), 4,5-di-O-caffoylqunic acid (10) were purchased from PhytoLab GmbH & Co. KG, Germany.

Hydrolysis by tetramethylammonium hydroxide (TMAH)

All the seven CGAs reference standards were treated with aqueous TMAH (25 g/litre). The initial concentrations of standards were as shown in Table S1 in supplementary information. Each sample was diluted by 5 ml of aqueous TMAH and stirred at room temperature. 1 ml solution from each sample was taken out at 2, 5, 10, 30 and 60 minutes time intervals. Each sample was saturated with brine and extracted twice with ethyl acetate. Combined organic layers were concentrated in vacuo and each sample was prepared in 1 ml of methanol to analyse by LC-MSn for intramolecular acyl migration.

To study the intermolecular acyl migration (Cross-over experiment), 5-CQA (25 mg, 0.07062 mmol) was added to a round bottom flask containing ferulic acid (13.7 mg, 0.07062 mmol). 5 ml of 10 times diluted (25 g/litre) TMAH was added to the flask and the mixture was stirred at room temperature. 1 ml samples were taken out from the flask at 2, 5, 10, 15 and 30 minutes time intervals. Each sample was saturated with brine and extracted twice with ethyl acetate. The combined organic layers were concentrated in vacuo and each sample was prepared in methanol to be analysed by LC-MSn. The same procedure was repeated with p- coumaric acid (11.57 mg, 0.07062 mmol) and 5-CQA (25 mg, 0.07062 mmol).

Model roasting

All the seven CGAs reference standards were heated at 180 0C for 12 minutes separately to study the intramolecular acyl migration. Equimolar quantities of 5-CQA with p-coumaric acid and 5-CQA with ferulic acid were heated together at 180 0C for 12 minutes to study the intermolecular acyl migration (Cross-over experiment). All the samples were heated in a Buechi Glass Oven B-585 and prepared in 1 ml methanol for LC-MSn analysis.

Brewing of CGAs (2-5 and 8-10)

Commercially available chlorogenic acids standards (each sample 500 µg) were infused in 3 mL of hot water each and stirred for 5 h under reflux. pH of each sample was determined to be 5 with pH meter. The solvent was removed under low pressure and the samples were dissolved in 1ml MeOH and used for LC-MSn.

LC-MSn

The LC equipment (Agillent 1100 series, Bremen, Germany) comprised a binary pump, an auto sampler with a 100 µL loop, and a DAD detector with a light-pipe flow cell (recording at

254 and 320 nm and scanning from 200 to 600 nm). This was interfaced with an ion-trap mass spectrometer fitted with an ESI source (Bruker Daltonics HCT Ultra, Bremen, Germany) operating in full scan, auto MSn mode to obtain fragment ion m/z. Tandem mass spectra were acquired in Auto-MSn mode (smart fragmentation) using a ramping of the collision energy. Maximum fragmentation amplitude was set to 1 Volt, starting at 30% and ending at 200%. The MS operating conditions (negative mode) had been optimized using 5-caffeoylquinic acid with a capillary temperature of 365 oC, a dry gas flow rate of 10 L/min and a nebulizer pressure of 10 psi.

HPLC

Separation was achieved on a 150 x 3 mm i.d. column containing diphenyl 5 µm, with a 5 mm x 3 mm i.d. guard column (Varian, Darmstadt, Germany). Alternatively, separation was also achieved on a 250 mm x 3 mm i.d. column containing C18-amide 5 µm, with a 5 mm x 3 mm i.d. guard column of the same material (Varian, Darmstadt, Germany) for the cases of hydrolysis (brewing) of reference standards experiments. Solvent A was water/formic acid (1000:0.05 v/v) and solvent B was methanol. Solvents were delivered at a total flow rate of 500 µL/min. The gradient profile was from 10 % B to 70 % B linearly in 60 min followed by 10 min isocratic, and a return to 10 % B at 90 min and 10 min isocratic to re-equilibrate.

Preliminary assessment of data

All the data for the chlorogenic acids presented in this paper use the recommended IUPAC numbering system; 14 the same numbering system was adopted for chlorogenic acids, their cis- isomers, their acyl-migration isomers and water addition products (Figure 3.1). The relative concentrations of the transformed products are expressed here in terms of the peak areas obtained in their UV chromatograms assuming the relative response factor in UV close to one, based on identical absorptivity of all mono CGA. 13,15 In tables and figures, the peak area values are stated accordingly.

3.3 Results and discussion

3.3.1 Intramolecular acyl migration: hydrolysis by TMAH of 2-5 and 8-10 The concentration of the samples was chose to be sufficiently low (1-1.5 mg/ ml) to prevent intermolecular acyl migration or transesterification, therefore allowing observation of intra molecular acyl migration exclusively. In the hydrolysis of 3-CQA (2), 5-CQA (3) and 4-CQA (4), all the other mono-acyl derivatives were identified during the hydrolysis except for 1-

CQA (1). Also, we did not observe the formation of any di-acyl derivatives in the hydrolysis of the mono-acylquinic acids. This observation confirms that the acyl migration we observed in this study was in fact an intramolecular process. Figure 3.2 represents the UV chromatograms of the 5-CQA in basic solution at different time intervals. It should be noted that the UV response in mono- and di-acylquinic acids has been used in the past as a reliable relative response factor. 9 Hanson et al. used radiolabelled quinic acid to investigate the acyl migration pathway in cinnamoylquinic acids 16; in accordance to his findings, we assumed that the mechanism of the acyl migration follows the ortho ester intermediate formation. This assumption is also supported by the study reported by Hanson and Cen Xie et al. 17 The transformations of 5-CQA (3), 4-CQA (4) and 3-CQA (2) with time are presented in Figure 3.3. From the results, we can conclude that 5-CQA is much more stable than 4-CQA and 3- CQA and the order of the stability is 5>4>3 in terms of the hydrolysis of the caffeoyl ester. This stability was observed to provide the resistance to decomposition thus allowing 5-CQA to form the acyl migrated products over longer hydrolysis durations. On the other hand, while 3-CQA being the least stable of the three mono-acylquinic acids, 3-CQA decomposes to form caffeic acid and quinic acid even before acyl migration takes place.

HO O HO O HO O C C C O OR5 CH CH CH C OR HO 4 HC HC HC

OR1 OR3 CH OH O 3 Q OH OH OH

C pCo F

Number Name and abbreviation R1 R3 R4 R5

1 1 -O-caffoylquinic acid (1-CQA) C H H H 2 3-O-caffoylquinic acid (3-CQA) H C H H 3 5-O-caffoylquinic acid (5-CQA) H H H C 4 4-O-caffoylquinic acid (4-CQA) H H C H

5 1,3-di-O-caffoylqunic acid (1,3-diCQA) C C H H 6 1,4-di-O-caffoylqunic acid (1,4-diCQA) C H C H

7 1,5-di-O-caffoylqunic acid (1,5-diCQA) C H H C

8 3,4-di-O-caffoylqunic acid (3,4-diCQA) H C C H 9 3,5-di-O-caffoylqunic acid (3,5-diCQA) H C H C 10 4,5-di-O-caffoylqunic acid (4,5-diCQA) H H C C

11 1 -O-p-coumaroylqunic acid (1-pCoQA) pCo H H H 12 3-O-p-coumaroylqunic acid (3-pCoQA) H pCo H H 13 5-O-p-coumaroylqunic acid (5-pCoQA) H H H pCo 14 4-O-p-coumaroylqunic acid (4-pCoQA) H H pCo H

15 1,3 -di-O-p-coumaroylqunic acid (1,3-dipCoQA) pCo pCo H H 16 1,4-di-O-p-coumaroylqunic acid (1,4-dipCoQA) pCo H pCo H 17 1,5-di-O-p-coumaroylqunic acid (1,5-dipCoQA) pCo H H pCo

18 3,4 -di-O-p-coumaroylqunic acid (3,4-dipCoQA) H pCo pCo H 19 3,5-di-O-p-coumaroylqunic acid (3,5-dipCoQA) H pCo H pCo 20 4,5-di-O-p-coumaroylqunic acid (4,5-dipCoQA) H H pCo pCo

21 1 -O-feruloylquinic acid (1-FQA) F H H H 22 3-O-feruloylquinic acid (3-FQA) H F H H 23 5-O-feruloylquinic acid (5-FQA) H H H H 24 4-O-feruloylquinic acid (4-FQA) H H F H

25 1,3 -di-O-feruloylqunic acid (1,3-diFQA) F F H H 26 1,4-di-O-feruloylqunic acid (1,4-diFQA) F H F H 27 1,5-di-O-feruloylqunic acid (1,5-diFQA) F H H F

28 3,4 -di-O-feruloylqunic acid (3,4-diFQA) H F F H 29 3,5-di-O-feruloylqunic acid (3,5-diFQA) H F H H 30 4,5-di-O-feruloylqunic acid (4,5-diFQA) H H F F

Q- quinic acid, C- caffeic acid, pCo- p-coumaric acid, F- ferulic acid Figure 3.1 Structure of mono and di caffeoylquinic, p-coumaroylquinic and feruloylquinic acids

Considering the mechanism of the acyl migration proceeding through an ortho-ester intermediate formation as 5-CQA ⇋ 4-CQA ⇋ 3-CQA, the reverse equilibrium, through which the migrated acyl group reverts back to its original position appears to be slower. It is difficult to comment on the thermodynamic equilibrium of the acyl migration because of the ongoing simultaneous ester hydrolysis competing with the acyl migration process.

Intens. [mAU] 4 2 Min 2 100 3 32 75 50 25 0 5 Min 4 2 100 3 32 75 50 25 0 10 Min 4 2 3 20

10 32 0 30 Min 3

10.0 2 7.5 5.0 4 32 0 10.0 12.5 15.0 17.5 20.0 22.5 25.0 27.5 30.0 Time [min]

Figure 3.2 UV Chromatograms (318-322 nm) at 2, 5, 10 and 30 minutes of base hydrolysis of 5-CQA (3)

3000 5-CQA 2500 3-CQA 2000 4-CQA 5-CQA 1500

Peak Peak area 1000

500

0 2 Min. 5 Min. 10 Min. 30 Min. Time (Minutes)

30 4-CQA 25 3-CQA 20 4-CQA 15 5-CQA

10 Peak Peak area 5

0 2 Min. 5 Min. 10 Min. 30 Min. Time (Minutes)

8 3-CQA 7 6 5 3-CQA 4 4-CQA 3 Peak Peak area 5-CQA 2 1 0 2 Min. 5 Min. 10 Min. 30 Min. Time (Minutes)

Figure 3.3 Amount of the transformation products after base hydrolysis for different time intervals of 5-CQA (3), 4-CQA (4) and 3-CQA (2)

The equilibrium between 3-CQA (2) and 4-CQA (4) is readily achieved because the ortho- ester intermediate is more stable due to the cis geometry (Figure 3.4). Compound 2 shows a tendency to hydrolyse to generate caffeic acid and quinic acid rather than to undergo acyl migration presumably due to the 1,3-syn-diaxial arrangement between the C1 hydroxyl group and the C3 ester. The hydrogen bonding between C1-OH and the carbonyl oxygen on the ester on C3 results in steric hindrance preventing the nucleophilic attack on the carbonyl carbon by the C4 hydroxyl group and facilitates the hydrolysis of the ester bond thus dissociating the caffeoyl moiety in basic conditions. At the same time this hydrogen bonding presumably activates the ester at C3 for hydrolytic cleavage.

O O O -H O O O O OH O O R 2 R O O HO H HO O HO R OH OH OH OH O OH OH H

3 4 OH

-H2O

O OH O OH HO OH HO O OH O O OH O R O R

Figure 3.4 Mechanism of the acyl migration through an ortho-ester intermediate formation

Previous studies have shown that 1,5-diCQA (7) was converted into 1,3-diCQA (5) and 5- CQA (3) rapidly and extensively by TMAH treatment within 1 minute of hydrolysis. 18,19 In the present study however, base hydrolysis of 1,3-diCQA (5) did not show any presence of 1,5-diCQA (7) or 1,4-diCQA (6). 1,3-diCQA (5) decomposed largely into 1-CQA (1) rather than 3-CQA (2) in the ratio 2.2:1 after two minutes of base treatment. 1,3-diCQA (5) did not transform preferably into 3,5-diCQA (9) whereas 3,4-diCQA (8) and 4,5-diCQA (10) were formed in very small quantities (Figure 3.5). 1-CQA (1) was found to be present entirely as a decomposition product rather than a migrated derivative as it was not observed in any other mono- or di-acylated substrates; this fact also supports the ortho-ester propagation of acyl

61 migration process. Moreover, this observation confirms the hydrolytic lability of the esters in the 3-acylated position.

In the case hydrolysis of 4,5-diCQA (10), we observed that 4,5-diCQA transformed mainly into 3,4-diCQA (8) after two minutes of base treatment. 5-CQA (3) and 4-CQA (4) showed approximately the same peak area at two minutes and it was slightly larger than 3-CQA (2). This trend continued for five minutes during the base treatment, after which all of the derivatives were completely decomposed (Figure 3.5). 3,4-diCQA (8) was observed to be the least stable substrate during the base treatment study. At two minutes, 8 was found to be in equilibrium with 3,5-diCQA (9) however after five minutes of treatment, 3,5-diCQA (9) was observed to display a slightly larger peak area than 3,4-diCQA (8). 3-CQA (2) was not observed in any sample throughout the duration of the base treatment of 3,4-diCQA due to its tendency to decompose rapidly by hydrolysis.

Observations based on the results from the base treatment on 3,5-diCQA (9) were quite distinctive. After five minutes 3,5-diCQA (9) was observed to possess surprising stability to the hydrolysis of the ester as we did not detect any of the mono-CQA derivatives even after 60 minutes of base treatment (Figure 3.5). 3,5-diCQA (9) transformed mainly into 3,4- diCQA (8) followed by 4,5-diCQA (10). The ratio of 9:8:10 remained approximately constant throughout the 60 minutes of base treatment.

900 1, 3-diCQA 800 1, 3-diCQA 700 3, 4-diCQA 600 4, 5-diCQA 500 3-CQA

400 5-CQA Peak Peak area 300 4-CQA 200 1-CQA 100 0 2 Min. 5 Min. 10 Min. 30 Min. Duration (Min)

3, 5-diCQA 300

250 3, 5-diCQA 200 3, 4-diCQA 150 4, 5-diCQA

Peak Peak area 100

0 2 Min. 5 Min. 10 30 60 Min. Min. Min. Duration (Min)

3, 4-diCQA 16 15 14 14

12 2 Min. 5 Min. 10 10

8 8

Peak Peak area 6

4 3 2 2

0 0 0 3, 4-diCQA 3, 5-diCQA 4-CQA 5-CQA Figure 3.5 (Continued)

4, 5-diCQA 180 160 165 140 134 2 Min. 120 5 Min. 100 91

80 Peak Peak area 60 61 58 47 40 20 8 0 2 5 0 0 0 3,5-diCQA 3, 4-diCQA 4, 5-diCQA 3-CQA 4-CQA 5-CQA

Figure 3.5 Amount of the transformation products after base hydrolysis for different time intervals of di-acylated reference standards

3.3.2 Intermolecular acyl migration (Transesterification): hydrolysis by TMAH (Cross- over experiment) In this contribution, we studied intermolecular acyl migration by carrying out cross-over experiments, in which 5-CQA (3) was reacted with the free acids like ferulic and p-coumaric acids at 1:1 stoichiometry in presence of a base. The intermolecular acyl migration was found to be simultaneously competing with intramolecular acyl migration as well as hydrolysis of the CGA. The products identified in the reaction of 5-CQA with ferulic acid and p-coumaric acid at 2, 5, 10, 15 and 30 minutes of base experiment are summarized in Table 3.1. Along with the transesterification products of 5-CQA and respective free acid, formation of the cis isomers was also observed in ferulic, caffeic and p-coumaric acids. All products were identified according to the fragmentation schemes reported by Kuhnert et al., Jaiswal et al. and Clifford et al. 18-21

Table 3.1 Compounds identified after base treatment of CGA with p-coumaric acid and CGA with ferulic acid for various time intervals

5-CQA + p-coumaric acid 5-CQA + ferulic acid Reaction Product RT (min) m/z (M-H) Reaction Product RT (min) m/z (M-H) time(min) number time(min) number

2 33 26.70 162.6 2 31 27.3 192.6 32 19.30 178.6 34 29.6 192.6 37 25.70 334.8 22 19.2 366.7 38 29.40 334.9 23 24.1 366.9 2 11.50 352.8 24 26.4 366.9 3 17.10 352.8 49 3.0 352.8 4 20.80 352.8 2 11.2 352.9 50 13.1 352.8

5 37 25.80 335.0 3 16.9 352.8 38 29.80 335.1 4 20.5 352.8 14 27.60 337.0 37 23.6 334.9 1 10.70 353.1 38 25.7 334.9 2 12.10 353.0 32 7.4 178.6 3 17.30 353.1 35 19.2 178.6 4 21.70 353.1 40 30.60 367.1 5 23 26.9 367.1 32 19.50 178.6 40 28.7 367.1 33 26.60 162.6 1 11.10 353.1 37 25.9 335.1

10 2 12.20 353.1 38 29.6 335.1 37 25.90 335.1 2 12.5 353.1 38 29.80 335.0 3 17.7 353.0 1 10.80 353.1 4 22.2 353.0 3 17.50 353.0 51 30.8 529.0 4 21.90 353.1 33 26.40 162.9 10 37 25.9 335.1 40 30.60 367.1 1 10.9 353.0

13 24.30 336.8 2 12.2 353.0 14 28.40 337.0 3 17.6 353.0 32 19.70 178.6 4 22.1 353.1 23 26.8 367.1

15 38 29.80 335.1 40 28.9 367.1 2 12.40 353.1 51 31.0 529.0 3 17.70 353.0 32 19.60 178.9 15 37 26.2 335.1 4 22.00 352.8 2 12.4 353.0 3 17.7 353.0

30 3 18.00 353.1 4 22.3 353.1 4 22.60 352.8 23 27.2 367.1 33 26.60 162.9 40 28.9 367.1

51 31.2 529.0

30 40 28.9 367.1

3 18.0 353.0

Intens. 9 [mAU] 2 Min 8 8 6 4 10 2 0 5 Min 9 10 8 8 6 4 10 2 0 10 Min 9 10 8 8 6 4 10 2 0 9 6 30 Min

4 8

2 10 0 60 Min 9 3 8 2 1 10 0 10 15 20 25 30 35 40 45 Time [min]

Figure 3.6 UV Chromatograms (318-322 nm) at 2, 5, 10, 30 and 60 minutes of base hydrolysis of 3, 5-diCQA (9)

O OH OH OH O O OH HO O OH O

O O O

OH OH OH HO OH 31 32 33 36

OH OH OH O O OH O O OH HO O O O O HO OH OH OH OH

34 35 36 37

O OH OH HO OH OH O O

O O O O O O HO OH O O OH OH OH HO OH OH

38 39 40

OH OH OH HO OH OH O O O O O O O O O HO OH OH OH OH HO HO O OH OH OH OH O

41 42 43

Figure 3.7 Compounds identified during acyl migration studies (Continued)

OH OH OH OH O

HO O O O O OH O OH O O O O O OH HO OH O HO OH O O O O O O

HO OH

HO HO 46 OH OH 44 45

OH OH O OH OH OH O OH OH HO O HO O O OH OH HO HO O O OH

O O OH O HO O OH HO O O O OH O O 49 O OH OH 47 48 (regio-chemistry randomly selected)

O OH O OH O OH OH HO O OH OH O O O O HO O O O OH OH O O HO OH OH OH HO HO O OH OH HO O O O OH 50 51 (regio-chemistry randomly selected) 52 (regio-chemistry randomly selected)

Figure 3.7 Compounds identified during acyl migration studies

12000 5-CQA + pCoA 3-CQA 10000 5-CQA 4-CQA 8000 Caffeic acid 6000

Peak Peak Area 4000

2000

0 2 5 10 15 30 Duration (Min)

Figure 3.8 Comparison between the peak areas of compounds formed during TMAH treatment of 5-CQA (3) with p-coumaric acid (pCoA)

In the case of the base treatment of an equimolar mixture of p-coumaric acid and 5-CQA (3), the intramolecular acyl migration within 5-CQA seemed to be dominating the hydrolysis and the intermolecular acyl migration. According to the peak areas observed, the formation of the intermolecular acyl migrated species (transesters) found to be least favoured (Figure 3.8). For example, 3-CQA (2), 5-CQA (3) and 4-CQA (4) formed predominantly over caffeic acid (32) during two minutes of TMAH treatment of 5-CQA. p-coumaric acid (33) did not esterify with the quinic acid generated from the hydrolysis of the 5-CQA (3) after two minutes hence, no p- coumaroylqunic acids were observed. However, after 5 minutes of the base treatment first transesterification product appeared in the form of 4-pCoQA (14) and after ten minutes both 5-pCoQA (13) and 4-pCoQA (14) were observed. Unfortunately, due to very low concentration we could not compare the peak areas of compounds 13 and 14 in the UV chromatogram and hence cannot comment on the kinetics of the acyl migration. Although it was clear that 4-pCoQA (14) was formed earlier than 5-pCoQA (13), but it was not obvious whether 13 was an acyl migration product of 14. Formation of intramolecular acyl migration products takes place according to the conclusions established earlier in this paper. After two minutes the rate of hydrolysis of 5-CQA was very low and the amounts of 3-CQA (2), 5-CQA (3) and 4-CQA (4) were the highest. Between five to ten minutes of base hydrolysis, equilibrium was reached where 5-CQA was found to be predominant. Formation of the cis derivatives is supposed to follow a water molecule addition-elimination to the double bond in the cinnamoyl moiety during the base hydrolysis.

Intens. 7 EIC 353.0 x10 3 4 3

50 2 2 49 1

0 5 10 15 20 25 Time [min]

[%] MS2(353) 49 191 100

0 MS3(353>191) 126.7 172.5 100 110.7 0 MS4(353 >191>172) 170.6 100 154.6 110.6 126.6 0 110 120 130 140 150 160 170 180 190 m/z

Intens. x107 EIC 529.0

0.8

0.6

0.4 51 0.2

0.0 22 24 26 28 30 32 34 Time [min]

Figure 3.9 (Continued)

[%] MS2(529.0) 51 334.9 100 290.9 0 MS3(529>335) 290.9 100 192.9 148.9 0 MS4(529>335>291) 148.9 100

0 50 100 150 200 250 300 350 400 450 500 m/z

Figure 3.9 EIC and fragmentation patterns for 1-cis-caffeoylquinic acid (49) at m/z 353 and caffeoyl-feruloylquinic acid (51) at m/z 529 in transesterification induced by TMAH

When an equimolar mixture of ferulic acid (31) and 5-CQA (3) was treated with base we observed three regio-isomers of feruloylquinic acid resulting from intermolecular transesterification: 3-FQA (22), 5-FQA (23) and 4-FQA (24) (Figure 3.10). 22 The same was not observed in the case of the p-coumaric acid experiment. After two minutes, 3-FQA (peak area = 3215) was formed predominantly over 5-FQA and 4-FQA. 5-FQA (23) and 4-FQA (24) showed negligible peak areas. Only 5-FQA remained stable enough to be detected after 10 and 15 minutes of base hydrolysis. After 30 minutes all the transesters and the substrate 5- CQA found to be decomposed completely as we could identify caffeic acid only. Formation of 3-FQA was found to be kinetically favoured. This was found to be consistent with the fact that during the first few minutes of the base treatment of 5-CQA to study intramolecular acyl migration, 3-CQA dominates the acyl migration product spectrum. We also identified cis-1- O-caffeoylquinic acid (49) in the EIC of m/z 353 and the UV chromatogram on the basis of its early elution and fragmentation (Figure 3.9). 23 We suspect the formation of the compound 49 is the product of the two step procedure involving water molecule addition-elimination at the double bond of the caffeic acid moiety. Since, we strictly did not observe any 1-acylated caffeoylquinic acid when 5-CQA was subjected to intramolecular acyl migration through base treatment as discussed previously in this paper. 4-cis-CQA (50) was also identified as another isomerised caffeoylquinic acid derivative, which is assumed to be formed by an addition elimination of the water molecule across the double bond in caffeic acid. 4-cis-CQA (50) was observed to be in equilibrium with 4-CQA (4) after two minutes but with an increase in the

72 reaction time only 4-CQA was observed to survive the base treatment. Unexpectedly, we identified 3-CQL (37) and 4-CQL (38) in this sample having negligible peak areas since the dehydrated product of caffeoylquinic acid must be in continuous equilibrium with caffeoylquinic acid itself in basic aqueous conditions. Hetero-diacyl chlorogenic acid was identified in the form of caffeoyl-feruloylquinic acid (51). It was observed after 5 minutes to 15 minutes of base hydrolysis (Figure 3.10). Figure 3.9 shows the fragmentation pathway for compound 51, in which it loses the ferulic acid moiety and undergoes simultaneous dehydration to give m/z 335 as a base peak in MS2. Furthermore, in MS3 the dehydrated caffeoylquinic acid entity undergoes decarboxylation to give a base peak at m/z 291 and also showing the presence of ferulic acid as a secondary peak at m/z 193. This fragmentation pathway for a caffeoyl-feruloylquinic acid was found to be inconsistent with the fragmentation pathways of 1-caffeoyl-3-feruloylquinic acid, 3-feruloyl-5-caffeoylquinic acid, cis-4-feruloyl-5-caffeoylquinic acid, 4-feruloyl-5-caffeoylquinic acid, 4-caffeoyl-5- feruloylquinic acid and cis-3-feruloyl-5-caffeoylquinic acid previously reported by Jaiswal et al. 21 Hence, the regio-chemistry of the acyl groups in caffeoyl-feruloylquinic acid (51) remains unknown.

16000 5-CQA + FA 14000 3-CQA 12000 5-CQA

10000 4-CQA 4-cis-CQA 8000 1-CQA

6000 Caffeic acid Peak Peak Area 4000

2000

0 2 5 10 15 30 Duration (Min)

Figure 3.10 Comparison between the peak areas of compounds formed during TMAH treatment of 5-CQA (3) with ferulic acid (FA)

3.3.3 Intramolecular acyl migration: model roasting of 2-5 and 8-10 Compounds 3-CQA (2), 5-CQA (3), 4-CQA (4), 1,3-diCQA (5), 3,4-diCQA (8), 3,5-diCQA (9) and 4,5-diCQA (10) were heated at 180 0C for 12 minutes separately to study the

73 intramolecular acyl migration in the absence of other under conditions mimicking coffee roasting.

Table 3.2 Compounds identified after heating (model roasting) reference standards

Starting Peak material Compound Product RT(min) Area(UV) 3 5-CQA 3 23.7 11035

2 3-CQA 2 18.9 06034

4 4-CQA 37 31.8 02418 38 35.1 01543 41 32.9 00396 48 44.9 00210 48 52.3 00197 52 46.3 00080 48 49.5 00521

5 1,3-diCQA 5 30.2 16245

8 3,4-diCQA 37 31.7 00098 38 35.2 00074 46 51.2 07099 46-cis 52.4 00896

9 3,5-diCQA 46 56.2 03034 46-cis 56.9 00119

10 4,5-diCQA 10 46.3 26560

From the data summarized in Table 3.2, we clearly see that only 4-CQA (4), 3,4-diCQA (8) and 3,5-diCQA (9) undergo transformations to generate various dehydrated products mainly in the form of caffeoyl lactones. In case of mono-acylated chlorogenic acids reference standards, only 4-CQA undergoes acyl migration with simultaneous dehydration. 3-CQL (37) was found to be the predominant transformation product in the heat treatment of 4-CQA. 5-

74 caffeoyshikimic acid was also identified but was found to be the least favoured dehydration product after 4-CQL. In this experiment, we observed that 3-CQL (37) was forming predominantly over 4-CQL (38) irrespective of the substrate speculatively because of the additional stability awarded by the equatorial position 3-CQL obtains in the inverted chair conformation, therefore we can conclude that the dehydration processes such as lactone and shikimic acid formation at the quinic acid moiety follow acyl migration. i.e. in the simulated roasting environment, acyl migration takes place before dehydration at quinic acid moiety. Additionally, lactonization dominates over the alternative shikimate formation in the model roasting of all the substrates (Table 3.2). A peculiar fragmentation was observed for a product yielding a pseudomolecular ion at m/z 671 leading to structures 48 and 52 (regio-chemistry unknown). Although, several peaks in the EIC at m/z 671 are observed, they are supposedly isomers of the two compounds 48 and 52 since only two distinct fragmentation patterns were observed shown in Figure 3.11, which suggests the possibility of the two different structures. Compound 48 is present in the form of three different isomers, which are represented in EIC at m/z 671 as 48a, 48b and 48c (Figure 3.11). All the three isomers generate a base peak at m/z 335 in MS2 suggesting the presence of di-caffeoyl quinide in the original structure. The other acyl group is believed to be sharing an ether linkage with an extra unit of quinic acid on one of its phenolic groups, which disappears as a neutral loss in MS2. On the other hand, in the MS2 of the compound 52, we see a base peak at m/z 509 and a secondary peak at m/z 353 clearly indicating the presence of di-caffeoylquinic acid as a central structure in which, one of the caffeoyl moieties is fused with a dehydrated quinic acid unit via ether bond on one of its phenolic groups. 1,3-diCQA (5) did not undergo noticeable transformation by the heat treatment whereas 3,4-diCQA (8) generated most of the transformation products among all four di-CQAs. It is likely that the observed products 3-CQL (37) and 4-CQL (38) resulting from 3,4-diCQA (8) were formed by lactonization and loss of one of the two acyl moieties following the order of the processes as dissociation first and dehydration at the quinic acid moiety later; attributed to the fact that both of the caffeoyl lactones possess similar peak areas. 3,4-di-O-caffoyl-1,5-quinide (46) was preferably formed in the heat treatment of 3,4-diCQA. The presence of the two different peaks eluting at 51.2 and 52.4 minutes having the same fragmentation pattern as 3,4-diCQL (46) suggested the presence of a cis isomer of compound 46 (Figure 3.12). Both isomers of 3,4-diCQL (46) generate a base peak at m/z 335 with virtually non-existent secondary peaks confirming the regio-chemistry of 3,4-diCQL.

Among the unchanged substrates throughout the heat treatment at 180 0C for 12 minutes such as, 5-CQA, 3-CQA, 1,3-diCQA and 4,5-diCQA the stability of 5-CQA (3) and 1,3-diCQA (5) to high temperatures was confirmed by heating them at 200 0C for 10 minutes.

Intens. EIC at m/z 671 x10 48b 48c 1.5 48a 1.0

0.5 37 41 38 52

0.0 30 35 40 45 50 55 Time [min]

[%] MS2(671) 48 (a/b/c) 335.7 100 509.1 671.1 0 MS3(671>336) 160.9 100 135.1 254.9 0 MS4(671>336>161) 133 100

0 100 200 300 400 500 600 m/z

[%] MS2(671.5) 52 509.2 100 353.3 191.4 0 MS3(671 >509) 190.8 100 353.0

0 MS4(671>509>191) 93 100

0 100 200 300 400 500 600 m/z

Figure 3.11 EIC and fragmentation patterns for m/z 671 observed during model roasting

Intens. EIC at m/z 497 x10 46 6

46-cis 2

0 42 44 46 48 50 52 54 56 Time [min]

[%] MS2(497) 46/ 46-cis 335.1 100

0 MS3(497>335) 160.9 100 135.3 0 MS4(497>335>161) 133 100

0 100 200 300 400 500 m/z

Figure 3.12 EIC and fragmentation patterns for m/z 497 observed during model roasting

3.3.4 Transesterification: model roasting (Cross-over experiment) In this contribution, we explored the possibilities of the transesterification or intermolecular acyl migration by carrying out cross-over experiments, in which 5-CQA was subjected to the heat treatment with the free acids like ferulic acid and p-coumaric acid at 1:1 stoichiometry. In the first case where equimolar mixture of pCoA and 5-CQA (3) was heated together, it was observed that 5-CQA (3) mostly remained unchanged by the heat treatment this experiment since the peak area under 5-CQA alone is more than 5 times the peak area of all the transformation products combined (Table 3.3). It is clear from the list that the caffeoylactones dominated all of the other transformation products. Since it was established

78 earlier by Clifford et al. that acyl migration takes place before dehydration in chlorogenic acid when there is still some water present in the sample, 2 it can be assumed that the 5-acyl group migrates to positions C3 and C4 first and then both products undergo dehydration to yield the corresponding lactones in the form of 3-CQL and 4-CQL. 4-CQL (38) was formed in larger quantities than 3-CQL (37). Only transesterification product identified was 4-pCoQA (14), which formed during model roasting conditions shows the peak area of 330 in UV chromatogram (Table 3.3 and Figure 3.13). It is speculated that the formation of the 1,5- quinide takes place earlier making C5 on quinic acid unavailable for the possible condensation with p-coumaric acid hence, we do not observe 5-pCoQA (13) but 4-pCoQA (14). Also, as mentioned earlier, 1,3-syn-diaxial arrangement between C1 and C3 hydroxyl groups in quinic acid generated by the hydrolysis of caffeoylquinic acid sterically hinders ester bond formation at C3, with additional H-bond activation of the C3 ester by the C1-OH. Hence, the overall absence of C3 transesterification products in model roasting experiment can be explained.

Table 3.3 Compounds identified after heating (Model roasting) of 5-CQA (3) and p-coumaric acid

Product number RT (min) Peak area 37 25.9 01093 38 29.6 01538 14 28.0 00330 48 36.9 00145 52 39.8 00642 52 41.6 02137 3 17.5 30503

Model roasting experiment between 5-CQA (3) and ferulic acid (31) generated a larger number of transformation products compared to the number of transformation products detected from the same experiment with pCoA (33) and 5-CQA (3). Similar trend in terms of number of the transformed products was observed in the cross-over experiment by TMAH treatment. The list of the transformation products in this experiment looks very similar to the list in Table 3.3. Additional two products are found to be the acyl migration products of 5- CQA. The acyl group in 5-CQA (3) was found to migrate intramolecularly to produce 3-CQA (2) and 4-CQA (4) but the corresponding caffeoyllactones 3-CQL (37) and 4-CQL (38) were formed more in quantity. The ratio of 3-CQL (37) to 3-CQA (2) was 11:1 whereas, the ratio

79 of 4-CQL (38) to 4-CQA (4) was found to be 1:1.3 since, 4-CQL formed in more quantity 3- CQL by comparison of corresponding peak areas (Table 3.4). In this experiment it was observed that 5-CQA (3) remained unchanged to even greater extent than previous experiment by the heat treatment since the peak area under 5-CQA alone is more than 16 times the peak area of all the transformation products combined (Table 3.4). 5-FQA (23) was formed in considerable quantity confirming intermolecular acyl migration between ferulic acid and 5-CQA (3) (Figure 3.13).

[%] MS, 28.0min 4-pCoQA (14) 337 100

0 MS2(337), 28.0min 173 100

0 MS3(337>173), 28.0min 110.9 100 93.1 154.7 0 100 150 200 250 300 350 400 450 m/z

[%] MS2(366.7), 26.8min 5-FQA (23) 191.1 100

0 MS3(367 >191), 26.8min 127 173 100 110.7

0 MS4(367>191 >173), 26.9min 109 100

0 100 125 150 175 200 225 250 275 300 m/z

Figure 3.13 MS3 and MS4 of 4-pCoQA (14) and 5-FQA (23) respectively observed during cross-over experiment by model roasting

Table 3.4 Compounds identified after heating (Model roasting) 5-CQA (3) and ferulic acid

Product number RT (min) Peak area 37 26.0 00484 38 29.4 00535 3 16.8 44224 2 12.2 00045 4 21.9 00421 23 26.8 00484 48 36.8 00176 52 39.9 00138 52 41.9 00500

Formation of the transesterification products in the cross-over experiments by the heat treatment was highly unexpected since; we did not observe both caffeic acid (32) and quinic acid in the model roasting of the mono- and di-caffeoylquinic acids (Table 3.2). It seems that the dissociation products remain stable enough to be esterified with another entity. Problem is, we can prove the existence of the hydrolysed products only if there is a chance to form transesterification products. Hence, we cannot supply any evidence of the hydrolysis of the esters in mono- and di-caffeoylquinic acids when they are heated separately because, even if free caffeic acid or quinic acid condenses with each other, it will be counted as acyl migration product in roasting experiments.

3.3.5 Intramolecular acyl migration: Brewing of CGAs In this contribution we also studied the acyl migration products of mono- and di-acyl CGAs (2-5 and 8-10), which were formed during the brewing process (Table 3.5). Typically, the pH of a coffee brew is at 4.7-5.1, in our case the pH of the brew was measured at 5.0. We observed that the hot water also serves as a reactive reagent other than just a simple solvent in coffee brewing similar to the previous work on tea fermentation, where water was shown to be the key reagent in thearubigin formation.24 Apart from the acyl migration products and trans-cis isomerization (cis-caffeoylquinic acids) products; the resulting chromatograms showed transformation products referred here as hydroxy-dihydrocaffeoylquinic acids arising through conjugate addition of water molecule to the olefinic cinnamoyl moiety.25

Table 3.5 Compounds identified after hydrolysis of reference standards (Brewing of CGAs)

Starting material Product name RT (Min) Peak area(UV)

5-CQA(3) 5 -CQA 20.1 17610 cis-5-CQA 23.0 00347 5-hCQA I 7.3 NA 5-hCQA II 7.9 NA

4-CQA(4) 3 -CQA 13.1 17576 4-CQA 20.6 06098 cis-3-CQA 11.9 00017 cis-4-CQA 16.5 00179 4-hCQA I 6.9 NA 4-hCQA II 7.9 NA

3-CQA(2) 3 -CQA 13.1 00639 4-CQA 20.6 00954 cis-3-CQA 11.9 00150 3-hCQA I + II 5.6 00017

1, 3-diCQA(5) 1, 3-diCQA 22.9 03422

3, 4-diCQA(8) 3,4 -diCQA 35.2 02558 cis-4,5-diCQA I 38.0 00099 4,5-diCQA 36.9 00520 cis-3,4-diCQA I 34.2 00393 cis-3,4-diCQA II 35.9 00345 3-CQA 12.3 00020 4-CQA 19.3 00027 5-CQA 15.6 00002 3-C-4-hCQA 26.4 00057

3, 5-diCQA(9) 3,4 -diCQA 36.5 01406 3,5-diCQA 37.3 01057 4,5-diCQA 41.4 01107

3-C-5-hCQAI 23.8 00014 3-hC-5-CQA I 27.8 00022 3-hC-cis-5-CQA 28.3 00004 3-hC-5-CQA II 31.7 00004 3-CQA 12.9 00080 5-CQA 20.3 00126

4, 5-diCQA(10) 3,4 -diCQA 43.7 01825 3,5-diCQA 42.6 00512 4,5-diCQA 45.7 02148 cis-4,5-diCQA I 47.2 00036 cis-4,5-diCQA II 52.0 00007 3-CQA 16.8 00019 4-CQA 27.9 00014 5-CQA 23.1 00018 4-hC-5-CQA 34.7 00018 3-hC-5-CQA II 35.5 00018 3-C-5-hCQA II 63.1 00023

NA- Insignificant peak area

Similar to the model roasting experiment, 5-CQA (3) did not show any acyl migrated products in hydrolysis (brewing). The acyl moiety in 4-CQA (4) and 3-CQA (2) did not migrate to C5 of the quinic acid but acyl moiety interchange between C3 and C4 was observed due to the stability of the ortho-ester intermediate arising from the cis geometry (Figure 3.4). Acyl migration to C5 from C3 and C4 was found to be highly pH dependent as we only observed it in case of base hydrolysis.

Cynarin (1,3-diCQA) did not show any transformation products after 5 h. of refluxing with water. In rest of the di-acylated reference standards 3,4-di-acylated esters were preferably formed irrespective of the substrate. This observation can possibly be attributed to the fact that the parallel displaced 휋- 휋 stacking arrangement of the cinnamoyl benzene rings provide the added stability to the 3,4-diCQA in the minimum energy chair conformation of quinic acid moiety. By comparing the peak areas in UV chromatogram it was observed that the decomposition of the di-CQAs to produce mono-CQAs was taking place in minute quantity as

83 compared to the base hydrolysis experiment to study intermolecular acyl migration. In the case of mono-acylated chlorogenic acids, up to 1.5-2.0% of the chlorogenic acids were transformed into their hydroxylated derivatives and the diacylated chlorogenic acids up to 4- 4.5% if the relative peak areas in EIC are considered.25 But their peak areas in UV chromatograms were found to be negligible. The structures for the water addition compounds can be found in in our previous publication 25 and in Figure 3.14.

3.4 Conclusions In this work we observed that that the acyl migration phenomena occur before dehydration takes place in quinic acid moiety. Acyl migration is facilitated in presence of the liquid media as compared to the roasting process. Therefore, the lower temperature roasts like the ‘cinnamon roast’ produce more number of acyl migration products than higher temperature roasts, which generate dehydration products like lactones and shikimates high in numbers.

Esters present on C3 position of the quinic acid are prone to hydrolysis of the ester bond than undergoing acyl migration in any experimental condition. In both cross-over experiments (TMAH and model roasting), 5-CQA (3) and ferulic acid (31) when treated together generated a large number of transesterification products than 5-CQA (3) treated with p- coumaric acid (33). Dehydration products in cross-over experiment by roasting are observed 10 times more in abundance than acyl migration or transesterification products. The amount of esters present on C3 and C4 positions of quinic acid moiety in a cup of coffee after roasting and brewing processes is highly contributed by C5 positioned esters in case of mono- caffeoylquinic acids content. In contrast to this observation we found that acyl migration to C5 position from C3 and C4 is only possible in base hydrolysis i.e. it is highly pH dependant. 1,3- diCQA (5) and 5-CQA (3) were observed to be more stable than the rest of the reference standards in both roasting and brewing conditions.

OH OH HO OH OH HO HO HO

HO HO OH OH O O OH O O OH O O HOOC O HOOC OH HOOC OH HOOC O OH OH OH OH OH OH OH OH

5-hCQA I 5-hCQA II 4-hCQA I O 4-hCQA II OH OH O HOOC OH HOOC OH HO O

OH O O OH O O OH O

OH OH HO OH

3-hCQA I 3-hCQA II 1-hCQL OH

HO HO OH OH

HO OH OH HO HOOC OH

OH O O OH HO O O OH O O cis-3-CQA HOOC OH OH 4-C-cis-5-CQA

HO OH OH HO HO

HO O O OH OH HOOC OH O HOOC O O OH O O O OH O OH O O OH HOOC O OH OH OH OH

HO cis-4-C-5-CQA 3-hC-cis-5-CQA cis-3-C-4-CQA OH

Figure 3.14 Structures identified after brewing of the reference standards (Continued)

OH HO HO HO OH OH

OH O OH OH O O O HOOC O HOOC O HOOC OH OH OH O O O O OH O O

OH OH

3-hC-4-CQA 3-C-4-hCQA 3-hC-5-CQA I

HO HO HO OH OH OH OH OH HO HO HO

HO O O HO HOOC OH O OH O O O O O OH HOOC OH OH HOOC OH OH O O O O 3-hC-cis-5-CQA OH HO 3-hC-5-CQA II OH 3-C-5-hCQA I

HO HO OH OH

OH HO

OH OH HO HO HO HO HO OH OH O O HO HOOC OH OH OH O O O O O O O O HOOC O HOOC O

OH OH OH OH

HO OH 4-hC-5-CQA 4-C-5-hCQA 3-C-5-hCQA II

Figure 3.14 Structures identified after brewing of the reference standards

References

1. Clifford, M.N. Chlorogenic acids and other cinnamates - nature, occurrence, dietary burden, absorption and metabolism. J. Sci. Food Agric. 2000, 80, 1033-1043.

2. Jaiswal, R.; Matei, M.F.; Golon, A.; Witt, M.; Kuhnert, N. Understanding the fate of chlorogenic acids in coffee roasting using mass spectrometry based targeted and non-targeted analytical strategies. Food Funct. 2012, 3, 976-984.

3. Crozier, A.; Jaganath, I.B.; Clifford, M.N. Dietary phenolics: chemistry, bioavailability and effects on health. Nat. Prod. Rep. 2009, 26, 1001-1043.

4. Farah, A.; De Paulis, T.; Trugo, L.C.; Martin, P.R. Effect of roasting on the formation of chlorogenic acid lactones in coffee. J. Agric. Food Chem. 2005, 53, 1505-1513.

5. Kweon, M.H.; Hwang, H.J.; Sung, H.C. Identification and antioxidant activity of novel chlorogenic acid derivatives from bamboo (Phyllostachys edulis). J. Agric. Food Chem. 2001, 49, 4646-4655.

6. Wang, G.; Shi, L.; Ren, Y.; Liu, Q.; Liu, H.; Zhang, R.; Li, Z.; Zhu, F.; He, P.; Tang, W.; Tao, P.; Li, C.; Zhao, W.; Zuo, J. Anti-hepatitis B virus activity of chlorogenic acid, quinic acid and caffeic acid in vivo and in vitro. Antiviral Res. 2009, 83, 186-190.

7. Hemmerle, H.; Burger, H.; Below, P.; Schubert, G.; Rippel, R.; Schindler, P.W.; Paulus, E.; Herling, A.W. Chlorogenic Acid and Synthetic Chlorogenic Acid Derivatives: Novel Inhibitors of Hepatic Glucose-6-phosphate Translocase. J. Med. Chem. 1997, 40, 137-145.

8. Robinson, W.E.; Reinecke, M.G.; AbdelMalek, S.; Jia, Q.; Chow, S.A. Inhibitors of HIV-1 replication that inhibit HPV integrase. Proc. Natl. Acad. Sci. U. S. A. 1996, 93, 6326-6331.

9. Kwon, H.C.; Jung, C.M.; Shin, C.G.; Lee, J.K.; Choi, S.U.; Kim, S.Y.; Lee, K.R. A new caffeoyl quinic acid from Aster scaber and its inhibitory activity against human immunodeficiency virus-1 (HIV-1) integrase. Chem. Pharm. Bull. 2000, 48, 1796-1798.

10. Clifford, M.N.; Kellard, B.; Birch, G.G. Characterisation of chlorogenic acids by simultaneous isomerisation and transesterification with tetramethylammonium hydroxide. Food Chem. 1989, 33, 115-123.

11. Clifford, M.N.; Kellard, B.; Birch, G.G. Characterization of caffeoylferuloylquinic acids by simultaneous isomerization and transesterification with tetramethylammonium hydroxide. Food Chem. 1989, 34, 81-88.

12. Dawidowicz, A.L.; Typek, R. Thermal Stability of 5-o-Caffeoylquinic Acid in Aqueous Solutions at Different Heating Conditions. J. Agric. Food Chem. 2010, 58, 12578-12584.

13. Trugo, L.C.; Macrae, R. A study of the effect of roasting on the chlorogenic acid composition of coffee using HPLC. Food Chem. 1984, 15, 219-227.

14. Anonymous IUPAC Commission on the Nomenclature of Organic Chemistry (CNOC) and IUPAC-IUB Commission on Biochemical Nomenclature (CBN). Nomenclature of cyclitols. Recommendations, 1973. Biochem J 1976, 153, 23-31.

15. Clifford, M.N.; Williams, T.; Bridson, D. Chlorogenic acids and caffeine as possible taxonomic criteria in Coffea and Psilanthus. Phytochemistry 1989, 28, 829-838.

16. Hanson, K.R. Chlorogenic acid biosynthesis Chemical synthesis and properties of the mono-O-cinnamoylquinic acids. Biochemistry 1965, 4, 2719-2731.

17. Xie, C.; Yu, K.; Zhong, D.; Yuan, T.; Ye, F.; Jarrell, J.A.; Millar, A.; Chen, X. Investigation of Isomeric Transformations of Chlorogenic Acid in Buffers and Biological Matrixes by Ultraperformance Liquid Chromatography Coupled with Hybrid Quadrupole/Ion Mobility/Orthogonal Acceleration Time-of-Flight Mass Spectrometry. J. Agric. Food Chem. 2011, 59, 11078-11087.

18. Clifford, M.N.; Johnston, K.L.; Knight, S.; Kuhnert, N. Hierarchical Scheme for LC-MSn Identification of Chlorogenic Acids. J. Agric. Food Chem. 2003, 51, 2900-2911.

19. Clifford, M.N.; Knight, S.; Kuhnert, N. Discriminating between the six isomers of dicaffeoylquinic acid by LC-MSn. J. Agric. Food Chem. 2005, 53, 3821-3832.

20. Jaiswal, R.; Matei, M.F.; Ullrich, F.; Kuhnert, N. How to distinguish between cinnamoylshikimate esters and chlorogenic acid lactones by liquid chromatography-tandem mass spectrometry. Journal of Mass Spectrometry 2011, 46, 933-942.

21. Jaiswal, R.; Sovdat, T.; Vivan, F.; Kuhnert, N. Profiling and Characterization by LC-MSn of the Chlorogenic Acids and Hydroxycinnamoylshikimate Esters in Mate (Ilex paraguariensis). J. Agric. Food Chem. 2010, 58, 5471-5484.

22. Kuhnert, N.; Jaiswal, R.; Matei, M.F.; Sovdat, T.; Deshpande, S. How to distinguish between feruloyl quinic acids and isoferuloyl quinic acids by liquid chromatography/tandem mass spectrometry. Rapid Commun. Mass Spectrom. 2010, 24, 1575-1582.

23. Clifford, M.N.; Kirkpatrick, J.; Kuhnert, N.; Roozendaal, H.; Salgado, P.R. LC-MSn analysis of the cis isomers of chlorogenic acids. Food Chem. 2008, 106, 379-385.

24. Kuhnert, N.; Drynan, J.W.; Obuchowicz, J.; Clifford, M.N.; Witt, M. Mass spectrometric characterization of black tea thearubigins leading to an oxidative cascade hypothesis for thearubigin formation. Rapid Commun. Mass Spectrom. 2010, 24, 3387-3404.

25. Matei, M.F.; Jaiswal, R.; Kuhnert, N. Investigating the Chemical Changes of Chlorogenic Acids during Coffee Brewing: Conjugate Addition of Water to the Olefinic Moiety of Chlorogenic Acids and Their Quinides. J. Agric. Food Chem. 2012, 60, 12105-12115.

CHAPTER 4: Synthesis, structure and tandem MS investigation of diastereomers of quinic acid

4.1 Introduction (-)-Quinic acid (1) is distributed naturally in a variety of plant materials ranging from coffee to cinchona bark to tobacco leaves and cranberries in its free form or in the form of its depsides, chlorogenic acids. Quinic acid was first isolated in 1790 and was given an empirical formula in 1838. 1 In a single coffee bean, up to 4.0mg free quinic acid is found. 2 In Colombian Arabica green coffee, up to 7.0 g/kg of quinic acid is present, which increases up to 10.0 g/kg upon roasting.3 Quinic acid provides characteristic astringent taste to the beverage hence; it is also used as a flavor enhancer in certain beverages. 4 Quinic acid is considered as a primary metabolite in most living organisms, being an intermediate in the shikimic acid pathway in the biosynthesis of aromatic compounds. 5,6 This paper will use the nomenclature defined by the IUPAC system for (-)-quinic acid as, 1L-1(OH),3,4/5- tetrahydroxycyclohexane carboxylic acid. 7

O OH O OH O OH O OH OH OH OH OH HO OH HO HO OH HO OH OH OH OH OH OH (-)-quinic acid (1) (-)-epi-quinic acid (2) muco-quinic acid (3) cis-quinic acid (4) C inverted C4 inverted 3 C5 inverted

O OH O OH OH OH OH HO OH HO OH OH OH HO OH OH OH OH HO OH O O OH OH (+)-quinic acid (+)-epi-quinic acid scyllo-quinic acid (5) neo-quinic acid (6) C and C inverted C and C inverted 4 5 3 4

Figure 4.1 Stereoisomers of quinic acid

Quinic acid has eight possible stereoisomers: four meso forms and two pairs of enantiomers (Figure 4.1). While the esters of the 3R, 4S, 5R isomers of the quinic acid dominates the esters with other diastereomers, they has been reported to be found naturally or as products of the food processing. As reported by Kuhnert and co-workers, 80 different chlorogenic acid

90 derivatives have been identified in green coffee beans. After roasting and brewing, this number is increased to 120 derivatives identified on the basis of the presence of fragment ions corresponding to quinic acid and quinic acid lactones in MSn. 8,9 After ingestion of foods containing quinic acid esters, metabolism in humans may also give rise to the esters of the diastereomers of the quinic acid. 8,10 This fact supports the assumption that the roasting or food processing in general facilitates the isomerization at the stereogenic centers in 3R, 4S, 5R esters of the quinic acid. 11,12 Having stated this, we found that a number of esters of the diastereomers of the quinic acid are already reported in literature as plant secondary metabolites. For example, in Lactuca indica L., Asimina triloba and Aster scaber 3,5- dicaffeoyl-muco-quinic acid was identified. In Asimina triloba 3-caffeoyl-muco-quinic acid was also identified. 13-15 In Chrysanthemum morifolium 3,5-dicaffeoyl-epi-quinic acid and 1,3-dicaffeoyl-epi-quinic acid was identified. 16 3,5-dicaffeoyl-epi-quinic acid esters were also reported in Ilex kudingcha. 17 These muco, epi and scyllo esters of diastereomers of quinic acid are reported to show important biological activities like, hepatoprotectivity, antioxidant activity and anti HIV-1 integrase activity. 13-17

Considering the fact that regiosomers as well as the esters of the diastereomers of quinic acids are readily distinguishable by their fragmentation pattern in tandem MS experiments, it is very important to acquire the authentic synthetic standards for the diastereomers of the quinic acid to obtain selective diastereomers of chlorogenic acids by organic synthesis. Development of the mass spectrometrical methods for diastereomers of chlorogenic acids will enable us to identify the novel chlorogenic acid derivatives present in the biological samples even in very low concentrations. Since, these biological samples emerge biosynthetically, either through various processes such as, roasting, brewing, cooking etc. or as products of metabolism; their mass spectrometrical study will help us understand the changes in the chlorogenic acid profile through biosynthetic processes in great details. Additionally, this study will contribute immensely to answer if diastereomeric compounds possess biological activities and if they are easily distinguishable by mass spectrometry. By this, the role of the tandem MS as a powerful tool for the structural elucidation will be emphasized.

Almost all of the studies involving analysis of the diastereomers of the quinic acid until now are either based on the non-selective isomerization of the quinic acid approach, which involves slightly ambiguous assignments of the diastereomers isolated by chromatographic techniques or by extraction of the quinic acid derivatives from plant material. In this contribution we have selectively synthesized muco-, epi-, cis- and scyllo-quinic acids using

91 appropriate functional group protection and deprotection strategies confirmed by analytical techniques like NMR or single crystal XRD. We also report the behavior of the diastereomers in the reverse phase HPLC in terms of the retention times and elution order and basic features of their tandem mass spectra using mass spectrometry as a reliable and predictive tool to assign the stereochemistry to even comparatively smaller molecules like (-)-quinic acid.

4.2 Experimental Chemicals

All the chemicals (analytical grade) were purchased from Sigma-Aldrich (Bremen, Germany) and were used without further purification.

LC/MSn

The LC equipment (Agillent 1100 series, Bremen, Germany) comprised a binary pump, an auto sampler with a 100 µL loop, and a DAD detector with a light-pipe flow cell (recording at 254 and 320 nm and scanning from 200 to 600 nm). This was interfaced with an ion-trap mass spectrometer fitted with an ESI source (Bruker Daltonics HCT Ultra, Bremen, Germany) operating in full scan, auto MSn mode to obtain fragment ion m/z. As necessary, MS2, MS3, and MS4 fragment-targeted experiments were performed to focus only on compounds producing a parent ion at m/z 191, 173. Tandem mass spectra were acquired in Manual-MSn mode using fixed collision energy. The fragmentation amplitude was set to 0.75 V. Also, direct injection experiments targeting the fragments in MS2, MS3, and MS4 were performed on all diastereomers of quinic acid keeping the fragmentation amplitude constant at 1.0 volts. MS operating conditions (negative mode) was optimized using (-)-quinic acid with a capillary temperature of 365oC, a dry gas flow rate of 10 L/min, and a nebulizer pressure of 10 psi.

HPLC Separation was achieved on a 250 × 4.6 mm i.d. column containing diphenyl 5 µm and 5 × 4.6 mm i.d. guard column of the same material (Varian, Darmstadt, Germany). Solvent (water: formic acid 1000:0.05 v/v) was delivered at a total flow rate of 800 µL/min by 30 min isocratic.

1.2.1 Synthesis of the mixture of the epimers of (-)-quinic acid The mixture of the epimers of (-)-quinic acid was obtained non-selectively by the process already described by Maier and co-workers. 6

4.2.2 Synthesis of the epi-quinic acid (2) Methyl quinate was prepared by refluxing quinic acid (1) (5000 mg, 26.02 mmol) with MeOH (100 ml) and Amberlite IR120 acidic resin (5000 mg) for 12 hours. The reaction mixture was then filtered and concentrated in vacuo. The product was obtained in more than 95 % yield and the purity of the product was confirmed by NMR. It was then subjected to selective silyl protection on C3 and C5 of the methyl quinate by tert-Butyldimethylsilyl chloride (TBDMSCl). 18

O O O HO HO OH HO O O a, b c

Si Si Si HO OH Si O O O O OH OH O S 1 e 7 O O 8

O O HO O HO O HO O f d 11 Si Si Si O O Si O O OH O O O HO g 12 HO O O O O + HO HO OH OH g O OH O 5 1 + OH + O O HO OH HO OH OH OH

2 1 9 10

Reagents and conditions: (a) MeOH, reflux, 12 h, Amberlite IR 120, 100%; (b) 2.6 equiv 0 TBDMSCl, 2.6 equiv Et3N, DMF 0 C, 2 h and then to RT, 16 h 75%; (c) 1.5 equiv MsCl, Py, RT, overnight, 95%; (d) 11, CsF, DMF, 90 0C, 24 h 2% 9 and 10; (e) 2 equiv Dess- 0 Martin periodinane, CH2Cl2 , overnight, RT, 87%; (f) 1.5 equiv NaBH4, Ethanol, -30 C, 40 min.; (g) 2M HCl, H2O, 1 h.

Figure 4.2 Reaction scheme for obtaining scyllo-quinic acid (5) and epi-quinic acid (2)

Methyl quinate (2650 mg, 12.86 mmol) was stirred with triethylamine (5 ml) in anhydrous dimethylformamide (26 ml) in an inert atmosphere and to this solution, tert- Butyldimethylsilyl chloride (5034 mg, 33.436 mmol) was added and the mixture was stirred for 2 hours at 0 0C and 16 hours at room temperature. EtOAc was added and the residue was

93 filtered. Concentrated filtrate was purified using flash chromatography (gradient eluent: 20% EtOAc in petroleum ether) to afford white crystalline methyl 3,5-Di-O-(tert- butyldimethylsilyl)quinate (7) in 75% yield, which was confirmed by crystal XRD (see

Figure 4.3 and 4.4) to have free C4 on quinic acid skeleton. Compound 7 was used as a precursor for the synthesis of epi-quinic acid (2) and scyllo-quinic acid (5) as shown in the Figure 4.2.

Figure 4.3 X-ray structure of methyl 3,5-Di-O-(tert-butyldimethylsilyl)quinate (7) Conformer A

Figure 4.4 X-ray structure of methyl 3,5-Di-O-(tert-butyldimethylsilyl)quinate (7) Conformer B

Dess-Martin periodinane (390 mg, 0.92 mmol) to the solution of compound 7 (200 mg, 0.46 mmol) and dichloromethane (15 ml) at room temperature. Reaction was stirred overnight and diluted with Et2O. Then to this, 1:1 saturated mixture of Na2S2O3 and NaHCO3 solution was added and stirred until reaction becomes almost clear. Organic layer was collected and the aqueous layer was extracted with 20 ml EtOAc thrice. Combined organic layers were collected, dried and concentrated in vacuo. Crude product was subjected to flash chromatography (gradient eluent: 24% EtOAc in petroleum ether), which afforded methyl 3,5-Di-O-(tert-butyldimethylsilyl)-4-oxoquinate (12) in the form of sticky colorless solid in 13 87%. See Figure 4.2. C NMR (100 MHz, CDCl3): 206.51, 173.15, 75.98, 75.33, 69.81, 52.94, 46.71, 41.22, 25.79, 25.61, 18.45, 17.88, -4.73, -5.04, -5.25, -5.30

Compound 12 was subjected to reduction at C4 by the action of NaBH4 and L-selectride. Out of these two reduction procedures, reduction with NaBH4 showed higher selectivity towards the formation of epi- derivative as compared to L-selectride. 177 mg of compound 12 (0.4074 mmol) was dissolved in ethanol and the reaction flask was immersed in an acetone bath. 0 Liquid nitrogen was slowly added until the bath temperature reaches to -30 C. NaBH4 (23 mg, 0.6111 mmol) was added and the mixture was stirred for 40 minutes. Solvent was removed immediately under reduced pressure at 30 0C in a rotary evaporator. The residue was then extracted with water and EtOAC mixture three times. Organic layers were collected and 13 dried over Na2SO4 and concentrated. C NMR of the crude product confirmed reduction at C4 indicated by the loss of a peak at 206 ppm. The crude product was directly subjected to hydrolysis by 2M HCl and water without further purification. Product of the hydrolysis was diluted with water and extracted with water and EtOAC mixture thrice. Aqueous layers were collected and concentrated in vacuo. Resulting white product was used in HPLC/MS analysis.

4.2.3 Synthesis of the muco-quinic acid (3) muco-quinic acid (3) was obtained from the methyl TMB-muco-quinate, which was synthesized by the procedure already reported by Jaiswal et al.. 8 Methyl TMB-muco-quinate (500 mg, 1.56 mmol) was subjected to hydrolysis by 70% aqueous trifluoroacetic acid for 1 hour. Resulting methyl muco-quinate was stirred with 2M KOH for 20 minutes followed by neutralization by Amberlite IR 120 acidic resin for 10 minutes. The mixture was filtered and concentrated under reduced pressure. The resulting yellowish solid was analyzed by NMR. 1H 3 3 NMR (400 MHz, D2O): 1.73 (dd, 1H, H6ax, JHH 12.4), 1.75 (dd, 1H, H2ax, JHH 12.4), 1.86 2 3 2 3 (dd, 1H, H6eq, JHH 4.1, JHH 11.2), 1.89 (dd, 1H, H2eq, JHH 4.6, JHH 11.4), 3.21 (t, 1H, H4, 3 2 3 13 JHH 9.6), 3.62 (ddd, 2H, JHH 4.6, H3 and H5, JHH 11.9, 9.1). C NMR (100 MHz, D2O): 180.9, 79.5, 74.7, 69.8, 69.7, 40.5, 40.4

4.2.4 Synthesis of the cis-quinic acid (4) The synthetic scheme is shown in Figure 4.5. 3,4-O-Cyclohexylidene-1,5-quinide (13) was synthesized by adding quantities of 10.00 g (52.04 mmol) of quinic acid and 200 mg (1.05 mmol) of p-toluenesulfonic acid monohydrate (PTSA·H2O) to 100 mL of cyclohexanone to give a white suspension. The reaction was then refluxed for 24 h to give a yellow solution, which was cooled to 50 ºC and neutralized with a solution of NaOEt (71.5 mg) in EtOH (5 mL) to give a yellow clear solution. The solvents were removed under reduced pressure and to the resulting yellow viscous liquid a volume of 100 mL of EtOAc was added. The organic phase was washed with 50 mL of H2O and the aqueous phase was back-extracted with 30 mL

EtOAc. The combined organic layers were washed with a half-saturated NaHCO3 solution, dried on Na2SO4, filtered and evaporated. The resulting yellow viscous liquid was recrystallized successively from a 1:1 n-heptane:EtOAc solution to afford white crystals of 1 compound 13 (9.26 g, 36.43 mmol, 70%); H-NMR (CDCl3): δH 4.73 (dd, 1H, J = 5.6, 2.8 Hz), 4.48 (td, 1H, J = 6.8, 2.8 Hz), 4.29 (ddd, 1H, J = 6.4, 2.8, 1.4 Hz), 2.66 (d, 1H, J = 11.9 Hz), 2.38-2.31 (ddd, 1H, J = 14.7, 7.8, 2.3 Hz), 2.31-2.25 (ddt, 1H, J = 11.9, 6.4, 2.3 Hz), 2.17 (dd, 1H, J = 14.7, 3.2 Hz), 1.73-1.68 (m, 2H), 1.67-1.60 (m, 2H), 1.57-1.51 (m, 4H), 13 1.43-1.36 (m, 2H); C-NMR (CDCl3): δC 178.9 (-COOR), 110.7 (C-1'), 76.1 (C-4), 71.8 (C- 1), 71.6 (C-3), 71.2 (C-5), 38.6 (C-6), 37.0 (C-2), 34.5 (C-6'), 33.7 (C-2'), 25.1 (C-5'), 24.0 (C-3'), 23.6 (C-4').

O O O O HO HO HO HO OH a b O c O O O O OH O O HO OH O O O OH

1 13 14 15 d

O O HO HO O OH f

O OH HO OH O OH

4 16 e

O HO O

HO OH OH

Reagents and conditions: (a) cyclohexanone, PTSA·H2O, reflux, 24 h; (b) 21% NaOMe/MeOH, MeOH, rt, overnight; (c) Dess-Martin periodinane, DCM, rt, 18 h; (d) NaBH4, MeOH/THF (1:1), -30 ˚C, 1 h; (e) KOH, THF, rt, 45 min; (f) HCl (trace amounts in CDCl3).

Figure 4.5 Reaction scheme for obtaining cis-quinic acid (4) and Methyl-cis-quinate (17)

3,4-O-Cyclohexylidene-1,5-quinide (13) (8.75 g, 34.41 mmol) was dissolved in 100 mL MeOH and a 21% solution NaOMe/MeOH was added (187 mg NaOMe). The clear solution was stirred overnight, the mixture was then quenched with glacial acetic acid (232 µL) and the volatile components were removed under vacuum. The resulting mixture was dissolved in

EtOAc and washed 3 times (3x40 mL). The organic layer was dried over Na2SO4, filtered and the solvent was removed under low pressure. The crude product was purified by column chromatography on silica gel (20-50% EtOAc/petroleum ether) to give methyl 3,4-O- 1 cyclohexylidene-quinate (14) as a white solid (5.87 g, 20.51 mmol, 60%). H-NMR (CDCl3):

δH 4.46 (dt, 1H, J = 5.9, 3.7 Hz), 4.14-4.07 (m, 1H), 3.97 (t, 1H, J = 5.9 Hz), 3.79 (s, 3H), 2.26 (m, 1H), 2.25 (d, 1H, J = 4.1 Hz), 2.08 (ddd, 1H, J = 13.4, 4.1, 3.0 Hz), 1.86 (dd, 1H, 13 13.7, 11.0 Hz), 1.74-1.68 (m, 2H), 1.68-1.52 (m, 6H), 1.44-1.33 (m, 2H); C-NMR (CDCl3):

δC 175.6 (-COOCH3), 110.1 (C-1'), 79.4 (C-4), 74.1 (C-1), 73.1 (C-3), 68.7 (C-5), 53.0 (-

CH3), 39.0 (C-6), 38.0 (C-2), 34.9 (C-6'), 34.8 (C-2'), 25.0 (C-5'), 24.1 (C-3'), 23.7 (C-4').

Figure 4.6 X-ray structure 3,4-O-Cyclohexylidene-1,5-quinide (13)

To a suspension of Dess-Martin periodinane (2963 mg, 6.99 mmol) in anhydrous CH2Cl2 (65 mL) compound 14 (1.82 g, 6.36 mmol) was added. The reaction mixture was stirred at room temperature for 18 h, was then diluted with Et2O (100 mL) and a 1:1 mixture (v/v) of saturated aqueous Na2S2O3 and NaHCO3 solution (100 mL). The mixture was stirred until the

solids were dissolved (20 min). The aqueous layer was extracted with Et2O and the combined organic layers were dried over Na2SO4, filtered and concentrated. Product methyl 3,4-O- cyclohexylidene-5-oxoquinate (15) (1.81 g, 6.36 mmol, 100%) was used for the next step 1 without further purification. H-NMR (CDCl3): δH 4.72 (m, 1H), 4.41 (d, 1H, J = 5.5 Hz), 3.81 (s, 3H), 2.89 (d, 1H, J = 14.2 Hz), 2.80 (dd, 1H, J = 14.7, 2.3 Hz), 2.56 (t, 1H, J = 2.3 Hz), 2.55 (d, 1H, J = 4.1 Hz), 1.74-1.64 (m, 2H), 1.64-1.56 (m, 4H), 1.56-1.47 (m, 2H), 1.43- 13 1.32 (m, 2H); C-NMR (CDCl3): δC 204.4 (C-5), 173.0 (-COOCH3), 111.7 (C-1'), 78.2 (C-4),

76.9 (C-1), 75.9 (C-3), 53.5 (-CH3), 49.1 (C-6), 37.0 (C-2), 35.3 (C-6'), 34.7 (C-2'), 24.9 (C- 5'), 23.9 (C-3'), 23.8 (C-4').

To obtain methyl 3,4-O-cyclohexylidene-epi-quinate (16), compound 15 (1.53 g, 5.33 mmol) was dissolved in a 1:1 mixture (v/v) MeOH/THF (100 mL) and was cooled to -30 ˚C with an acetone/liquid nitrogen bath. NaBH4 (222 mg, 5.86 mmol) was added and the mixture was stirred at -30 ˚C for 1 h. The solvents were removed in vacuum and the residue was extracted three times with a water/EtOAc mixture. The organic layers were dried over Na2SO4, filtered and the solvent was removed under reduced pressure. The product was purified by column 1 chromatography (40% EtOAc/petroleum ether). H-NMR (CDCl3): δH 4.52 (dt, 1H, J = 7.6, 5.0 Hz), 4.30 (dd, 1H, J = 7.6, 4.1 Hz), 3.90 (dt, 1H, J = 9.6, 4.1 Hz), 3.78 (s, 3H), 2.18-2.06 (m, 3H), 2.02 (dd, 1H, J = 14.2, 10.1 Hz), 1.75 (m, 2H), 1.69-1.59 (m, 4H), 1.59-1.51 (m, 13 2H), 1.44-1.35 (m, 2H); C-NMR (CDCl3): δC 175.4 (-COOCH3), 110.0 (C-1'), 73.8 (C-4),

73.4 (C-1), 72.5 (C-3), 66.1 (C-5), 53.0 (-CH3), 38.1 (C-6), 36.4 (C-2), 35.6 (C-6'), 34.0 (C- 2'), 25.2 (C-5'), 24.1 (C-3'), 23.7 (C-4').

Figure 4.7 X-ray structure of cis-quinic acid (4)

Crystals of cis-quinic acid (4) (Figure 4.7) suitable for single crystal XRD were obtained in an NMR tube containing demethylated 16 dissolved in CDCl3 by removal of the acid-labile cyclohexylidene protection promoted by the trace amounts of HCl present in the deuterated 1 solvent. H-NMR (D2O): δH 3.76 (br, 1H), 3.67 (t, 1H, J = 3.2 Hz), 3.64 (t, 1H, J = 3.2 Hz), 13 1.97 (dd, 2H, J = 12.4, 4.1 Hz), 1.62 (t, 2H, J = 12.4 Hz); C-NMR (D2O): δC 177.1 (-

COOCH3), 72.4 (C-4), 71.1 (C-1), 66.9 (C-3, C-5), 35.9 (C-2, C-6).

4.2.5 Synthesis of the scyllo-quinic acid (5) Methyl 3,5-Di-O-(tert-butyldimethylsilyl)quinate (7) (500 mg, 1.15 mmol) was mixed and stirred with 12 ml of anhydrous pyridine and cooled to 0 0C. Methanesulfonyl chloride was added drop wise to the mixture. Reaction was stirred at room temperature for overnight. 10 ml

NaHCO3 added and after stirring for 10 minutes, aqueous phase was washed two times with

Et2O. Organic layers were collected, dried over Na2SO4 and concentrated. Crude product was subjected to flash chromatography (gradient eluent: 22% EtOAc in petroleum ether), which afforded methyl 3,5-Di-O-(tert-butyldimethylsilyl)-4-O-methanesulfonylquinate (8) in 95% 1 yield. H NMR (400 MHz, CDCl3): 0.06 (s, 3H, MeSi), 0.08 (s, 3H, MeSi), 0.11 (s, 3H,

MeSi), 0.12 (s, 3H, MeSi), 0.86 (s, 9H, Me3CSi), 0.88 (s, 9H, Me3CSi), 1.94 (dd, 1H, H6ax, 2 3 JHH 13.74, JHH 10.07, Hz), 2.00-2.10 (m, 1H, H2ax), 2.12-2.31 (m, 2H, H6eq, H2eq), 3.04 (s, 3 2 3H, MeS), 3.77 (s, 3H, -OMe), 4.25 (dd, 1H, H4, JHH 8.7, JHH 2.3 Hz), 4.34 (m, 1H, H5) and

4.57 (m, 1H, H3).

100

Compound 8 (520 mg, 1.014 mmol) was dissolved in anhydrous dimethylformamide and to this was added cesium fluoride (790 mg, 5.2 mmol). The mixture was stirred for 30 minutes and cinnamic acid (770 mg, 5.2 mmol) was added. The reaction was heated to 90 0C and allowed to stir for 36 hours. DMF was removed under low pressure and the obtained crude product was subjected to column chromatography (gradient eluent: 45% EtOAc in petroleum ether). One fraction collected from the column chromatography confirmed the formation of the desired product in the NMR analysis and formed a crystal in the NMR tube. The crystal was analyzed by crystal XRD without detailed assignment of the NMR signals. XRD showed a rare phenomenon of co-crystallization of 3-O-cinnamoyl-1,4-scyllo-quinide (9) and 3-O- cinnamoyl-1,5-quinide (10) as can be seen in Figure 4.8.

The mixture of compound 9 and compound 10 was subjected to hydrolysis through the same procedure as described earlier. Product of the hydrolysis was diluted with water and extracted with water and EtOAC mixture thrice. Aqueous layers were collected and concentrated in vacuo. Resulting white product was used in HPLC/MS analysis.

Figure 4.8 X-ray structures of 3-O-cinnamoyl-1,4-scyllo-quinide (9) and 3-O-cinnamoyl-1,5- quinide (10)

101

4.2.6 Hydrolysis of the CGAs in roasted coffee 10 grams of finely grounded roasted coffee from Guatemala was boiled in distilled water for 1 hour. The mixture was then filtered and the water in the filtrate was removed under reduced pressure. Resulting residue was subjected to base hydrolysis by the treatment of 25 ml 2M NaOH for 2 hours. The reaction was neutralized by addition of 2M HCl. The mixture was diluted with 25 ml water and the aqueous layer was extracted with EtOAC thrice. The aqueous layers were collected and concentrated in vacuo. Resulting yellowish product was used directly in HPLC/MS analysis.

4.2.7 Synthesis of the methyl esters of epi-, muco-, cis-, scyllo-quinic acids and (-)-quinic acid As described earlier, methyl quinates of all the synthesized diastereomers except the methyl cis-quinate (17) were prepared by refluxing respective diastereomers in MeOH with equal amount of Amberlite IR120 acidic resin for 12 hours. The reaction mixture was then filtered and concentrated in vacuo. The resulting product was used directly in HPLC/MS analysis in case of methyl epi-quinate, methyl scyllo-quinate and methyl muco-quinate. Crystals of methyl cis-quinate (17) suitable for single crystal XRD (Figure 4.9) were obtained in an

NMR tube containing methyl 3,4-O-cyclohexylidene-epi-quinate (16) dissolved in CDCl3 by removal of the acid-labile cyclohexylidene protection promoted by the trace amounts of HCl present in the deuterated solvent. The methyl quinate of (-)-quinic acid was confirmed by the NMR and then utilized in HPLC/MS.

102

Figure 4.9 X-ray structure of methyl cis-quinate (17)

4.2.8 X-ray crystallography Crystals were mounted on a Hampton cryoloop in light oil for data collection at 100 K. Indexing and data collection were performed on a Bruker D8 SMART APEX II CCD diffracto-meter with κ geometry and Mo Kα radiation (graphite mono- chromator, λ = 0.71073 Å). Data integration was performed using SAINT. Routine Lorentz and polarization corrections were applied. The SHELX package was used for structure solution and refinement. Refinements were full-matrix least-squares against F2 using all data. In the final refinement, all non-hydrogen atoms were refined anisotropically and hydrogen atoms were either found directly and refined isotropically or placed in calculated positions. Crystallographic data are summarized in Table 4.2.

4.3 Results and discussions Several methods for the synthesis of the epi-quinic acid (2) have been reported, 18 out of which two methods were selected for further modification. Most of the other methods rely on formation of 1,5-quinide followed by selective protection of C3 by silyl protecting group and then achieve inversion of configuration at C4 by oxidation-reduction reactions. This approach was found out to be unreliable and highly problematic, since the protection was observed to be non-selective. In the present work, once the methodology to obtain compound 7 in pure

103 form was developed, the ambiguity in the selective protection was eliminated in order to leave

C4-OH free for further treatment (Figure 4.2). Methyl 3,5-Di-O-(tert- butyldimethylsilyl)quinate (7) served as a precursor for both of the diastereomers, epi-quinic acid (2) and scyllo-quinic acid (5). Out of which, diastereomer 2 was obtained by oxidation- 2 reduction pathway and compound 5 was synthesized by incorporating SN reaction strategy, in which cinnamic acid served the role of nucleophile attacking C4 from backside to achieve inversion of the configuration. Acids such as, benzoic acid and 3,4-dimethoxycinnamic acid were also unsuccessfully used to serve as nucleophiles. However, in fact the desired product was found out to have rearranged itself in the form of 3-O-cinnamoyl-1,4-scyllo-quinide (9) possibly due to the traces of the HCl present in CDCl3 in the NMR sample tube. 3-O- cinnamoyl-1,4-scyllo-quinide(9) and 3-O-cinnamoyl-1,5-quinide (10) were co-crystallized and were obtained in very low yield. Hence, attempts towards the isolation of compound 9 were avoided and the crystals were directly hydrolyzed to get the mixture of scyllo-quinic acid (5) and quinic acid.

Implementation of the protection strategies to obtain cis-quinic acid (4) was observed to be less complicated if compared to epi-quinic acid (2) and scyllo-quinic acid (5). Protection on

C3 and C4 in 1, 2 acetal formation and formation of 1,5-quinide can be achieved simultaneously by using cyclohexanone as previously described by Fernandez et al.. 19 Once the C1- C5 lactone was cleaved with sodium methoxide, C5-OH becomes available for oxidation-reduction reactions to achieve inversion at C5 (Figure 4.7). From the synthetic products and intermediates, a total of five single crystal x-ray structures could be obtained. cis-quinic acid (4) was obtained as a single crystal. scyllo-quinic acid (5) was obtained by hydrolyzing the co-crystal of Compounds 3-O-cinnamoyl-1,4-scyllo-quinide (9) and 3-O- cinnamoyl-1,5-quinide (10). muco-quinic acid (3) was obtained in quantitative yield from the hydrolysis of Methyl TMB-muco-quinate. The co-eluting mixture of epi-quinic acid (2) and quinic acid (1) was purified by column chromatography and was analyzed as a mixture in LC- MS. We have investigated all of the above mentioned compounds by chromatography as single compounds and as a mixture by optimization of the separation methods on reversed phase HPLC packings. We also performed the non-selective synthesis of all the diastereomers from (-)quinic acid using prolonged heating in sulfuric acid as described previously 6 in an attempt of the identification and assignment of the remaining diastereomers of quinic acid, which have not been obtained in this study by synthetic methods.

104

Table 4.1 MS2 data of quinic acid diastereomers in negative ion mode at 75% collision energy

Compound MS1 MS2 Parent Base No. ion peak Secondary peak m/z m/z int % m/z int % m/z int % m/z int % 1 191 127 173 82 85 56 93 51 111 23 2 191 173 127 86 145 30 85 24 111 17 3 191 127 173 64 85 45 109 20 145 19 4 191 93 173 37 111 34 155 7 61 5 5 191 173 127 91 85 50 111 48 93 46

If we observe the retention times of all the diastereomers, we see that they elute very close to each other (Figure 4.10). The elution order can be given as, 1>5>2>3=4. Where quinic acid (1) elutes at 3.9 minutes and muco- and/or cis-quinic acid (3/4) elute at 3.4 minutes. Although diastereomers 3 and 4 elute at the same retention time, their tandem MS shows clear distinction as can be seen in Figure 4.10 and Figure 4.13. Diastereomers 2 and 5 elute very close to each other and they fragment almost identically in MS2 however in MS3, scyllo- quinic acid (5) produces a base peak at m/z 93 whereas epi-quinic acid (2) gives a base peak at m/z 111 in MS3. In other case, (-)-quinic acid (1) and muco-quinic acid (3) show similar fragmentation pattern giving base peaks at m/z 127 and m/z 109 in MS2 and MS3 respectively but, they differ in the retention times with compound 3 eluting half a minute earlier than compound 1. Also, to find a difference between diastereomers 1 and 3 on the basis of their tandem MS, we had to fragment them further into MS4, which will be discussed in detail later in this paper. Next to tandem LC-MSn experiments, direct infusion tandem MS experiments were carried out wherever necessary to see the further fragmentation of a particular base peak ion. The MS2 fragmentation all of the diastereomers was confirmed by keeping collision energy constant in the LC-MS runs at 0.75 volts (see Table 4.1). Optimum value for the collision energy was standardized by taking (-)-quinic acid (1) as a standard. Roasted coffee hydrolysis experiment shows us that diastereomers epi-quinic acid (2), muco-quinic acid (3) and cis-quinic acid (4) are present along with quinic acid (1) in the roasted coffee as products of the food processing (Figure 4.12). Based on the fragmentation behavior and retention times learned from the earlier experiments, we could only identify diastereomers 2, 5 and 1 from the EIC of the non-selectively isomerized quinic acid as shown in Figure 4.11.

105

Intens. 7 EIC 191.0 x10 (-)-quinic acid (1)

0 2 4 6 8 10 Time [min]

[%] -MS2(190.7), 3.9min (1) 126.8 100 85.2 172.7 61.5 93.0 110.9 188.6 0 154.9 -MS3(190.8>126.9), 4.0min 108.9 100

85.5 0 40 60 80 100 120 140 160 180 200 m/z

Figure 4.10 EIC at m/z 191 in negative mode and MSn of quinic acid (1), epi-quinic acid (2), muco-quinic acid (3), cis-quinic acid (4) and scyllo-quinic acid (5) (Continued)

106

Intens. 6 EIC 191.0 x10 1 epi-quinic acid (2) 2 1.5

1.0

0.5

0.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 Time [min]

[%] -MS2(190.7), 3.6min (2) 126.8 100 172.7

85.1 144.8 108.8 59.5 0 -MS3(190.8->172.5), 3.7min 100 154.8

111.0

0 50 100 150 200 250 m/z

Figure 4.10 EIC at m/z 191 in negative mode and MSn of quinic acid (1), epi-quinic acid (2), muco-quinic acid (3), cis-quinic acid (4) and scyllo-quinic acid (5) (Continued)

107

Intens. 7 EIC 191.0 x10 muco –quinic acid (3)

0 2.5 5.0 7.5 10.0 12.5 15.0 17.5 Time [min]

[%] -MS2(190.7), 3.4min 126.8 100 (3)

85.1 144.8 172.7 108.9 154.7 0 -MS3(190.8->126.8), 3.5min 108.8 100

0 85.1 40 60 80 100 120 140 160 180 200 m/z

Figure 4.10 EIC at m/z 191 in negative mode and MSn of quinic acid (1), epi-quinic acid (2), muco-quinic acid (3), cis-quinic acid (4) and scyllo-quinic acid (5) (Continued)

108

Intens. EIC 191.0 x107 cis-quinic acid (4) 8

0 2 4 6 8 10 Time [min]

[%] -MS2(190.7), 3.4min 93.1 100 (4)

60 172.7 110.9 40

20 59.1 136.8 86.1 154.7 0 40 60 80 100 120 140 160 180 200 m/z

Figure 4.10 EIC at m/z 191 in negative mode and MSn of quinic acid (1), epi-quinic acid (2), muco-quinic acid (3), cis-quinic acid (4) and scyllo-quinic acid (5) (Continued)

109

Intens. 6 EIC 191.0 x10 scyllo-quinic acid (5) 4

2 (1) 1

0 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 Time [min]

[%] -MS2(190.7), 3.7min (5) 172.6 100 126.8 85.1

110.9 154.7 188.7 71.3 0 -MS3(190.8->173.3), 3.8min 93.0 100

127.8 110.9 152.7 0 50 100 150 200 250 300 m/z

Figure 4.10 EIC at m/z 191 in negative mode and MSn of quinic acid (1), epi-quinic acid (2), muco-quinic acid (3), cis-quinic acid (4) and scyllo-quinic acid (5)

110

Intens. Acid fraction: TIC -All MS x107 2

1.5 1 5 1.0

0.5

3.00 3.50 4.00 4.50 Time [min]

[%] -MS2(190.7), 3.5min 100 (2) 126.8 172.7

144.8 108.9 85.1 0 -MS3(190.8->172.8), 3.6min 100 93.1 71.4 126.7 154.8 110.8

0 50 100 150 200 250 m/z

[%] -MS2(190.7), 3.7min 172.7 100 (5) 126.8

85.1 108.9 144.8 0 71.3 -MS3(190.8->172.6), 3.8min 100 111.0 93.1 127.0 154.7 0 50 100 150 200 250 m/z

Figure 4.11 MSn of the acidic fraction of non-selectively isomerized quinic acid (Continued)

111

[%] -MS2(190.7), 4.0min 126.8 100 (1) 172.7 85.1 110.9 71.4 144.8 0 -MS3(190.9->126.9), 4.0min 100 108.9

73.3 0 50 100 150 200 250 m/z

Figure 4.11 EIC at m/z 191 in negative mode and MSn of the acidic fraction of non- selectively isomerized quinic acid

Intens. x106 EIC 191.0 Hydrolyzed roasted coffee 1.0

0.8

0.6

0.4

0.2 1 3 5 2 0.0 1 2 3 4 5 6 Time [min]

Figure 4.12 Diastereomers identified in the EIC of the hydrolyzed roasted coffee in negative ion mode (Continued)

112

[%] -MS2(190.7), 3.5min 126.9 100 3

172.6 73.3 108.8 144.7 0 -MS3(190.9>126.8), 3.5min 100 108.8

0 50 100 150 200 250 300 m/z

[%] -MS2(190.8), 4.0min 126.8 100 1 172.7 144.7 85.2110.9 0 -MS3(190.9>127.0), 4.1min 108.9 100

0 50 100 150 200 250 300 m/z

[%] -MS2(190.7), 3.6min 93.1 100 5 80 172.7 60 126.8 40 110.9 20 144.7 71.3 0 50 100 150 200 250 300 m/z

Figure 4.12 Diastereomers identified in the EIC of the hydrolyzed roasted coffee in negative ion mode (Continued)

113

[%] -MS2(190.7), 3.7min 172.7 100 4 126.8 85.1 110.9 144.8 188.7 0 -MS3(190.8>172.6), 3.8min 100 110.9 93.1 152.7 65.8 0 50 100 150 200 250 300 m/z

Figure 4.12 Diastereomers identified in the EIC of the hydrolyzed roasted coffee in negative ion mode

114

[%] -MS2(191.0) 1 100 126.8 172.8 85.1 108.9 144.8 188.6 0 -MS3(191.0>127.0) 100 108.9 57.8 85.1 0 -MS4(191.0>127.0>109.0) 80.8 100

0 40 60 80 100 120 140 160 180 200 220 m/z

[%] -MS2(191.0) 2 126.8 100 172.7

82.4 144.8

0 -MS3(191.0>173.0) 111 100

93 127.8

0 60 80 100 120 140 160 m/z

Figure 4.13 MSn of the diastereomers of the quinic acid and reference compounds obtained from the direct infusion experiments (Continued)

115

[%] -MS3(191.0->127.0) 100 3 108.9

99.0 0 -MS4(191.0->127.0>109.0) 81.2 100

0 50 60 70 80 90 100 110 120 130 140 m/z

[%] -MS2(173.0) Shikimic acid 154.7 100 93.1 110.9

99.0 142.8 0 73.383.2 -MS3(173.0>155.0) 110.9 100 93.1 83.2 136.8 0 -MS4(173.0>155.0>111.0) 93.1 100 83.2

0 40 60 80 100 120 140 160 180 m/z

Figure 4.13 MSn of the diastereomers of the quinic acid and reference compounds obtained from the direct infusion experiments (Continued)

116

[%] 2 OH -MS (109.0) 107.9 100

OH Hydroquinone 0 -MS3(109.0->108.0) 100

0 40 60 80 100 120 140 160 180 200 m/z

[%] -MS2(109.0) 81.3 100 OH OH

Catechol

0 50 60 70 80 90 100 110 120 130 140 m/z

Figure 4.13 MSn of the diastereomers of the quinic acid and reference compounds obtained from the direct infusion experiments (Continued)

117

[%] -MS2(109.0) 100 65.5 OH

OH Resorcinol

107.0

0 40 60 80 100 120 140 160 180 200 220 m/z

[%] -MS2(97.0) 81.2 100

Cyclohexene oxide

0 20 40 60 80 100 120 140 m/z

Figure 4.13 MSn of the diastereomers of the quinic acid and reference compounds obtained from the direct infusion experiments

We observed that the tandem MS of each of the diastereomers is different therefore allow unambiguous identification. In the following mechanisms of fragmentation, rationalizing different fragment spectra are discussed and proposed. As shown in scheme A in Figure 4.14, (-)-quinic acid (1) undergoes dehydration and decarboxylation to produce a base peak at m/z 127 to give fragment Q1 in MS2. For dehydration in quinic acid, there are in principle three alternate mechanistic pathways available. Firstly, formation of lactone, secondly, elimination of water leading to a cyclohexene and finally, epoxide formation. To confirm the mechanistic pathways, wherever possible, fragment ions generated from the precursor ions were compared to reference substances or derivatives to assign the structure of the fragment ion as lactone,

118 cyclohexene or epoxide. We have to assume that these two fragmentations take place simultaneously as we see no secondary peak at m/z 145, which signifies decarboxylated quinic acid. Carboxyl group on C1 leaves as a formic acid and the dehydration takes place in between C2 and C3 due to the fact that the proton on C2 and hydroxyl group on C3 share an anti-periplanar arrangement. Q1 at m/z 127 further fragments in MS3 to give base peak at m/z

109, which suggests the presence of dihydroxybenzene-type structure. But, since C4 and C5 do not possess the anti-periplanar arrangement between either protons and hydroxyl groups all three 1,3-, 1,4- and 1,2-dihydroxybenzenes do not show fragmentation similar to the MS4 of quinic acid (1). In fact, the fragmentation of cyclohexene oxide in MS2 gives only peak at m/z 81 similar to MS4 of (-)-quinic acid (1) as can be seen in Figure 4.13 suggests the presence of an epoxide between C4 and C5 (Q2).

Scheme B in Figure 4.14 shows the proposed fragmentation pathway for epi-quinic acid (2).

The fragmentation can be explained if we consider that the elimination of water through an E2 route takes place between C1 and C2 to give shikimic acid with a base peak at m/z 173 (Q3). A comparison of the MS3 spectrum of shikimic acid with the MS3 spectrum of 2 with Q3 at m/z 173 as a precursor ion, confirms this assignment. Q3 shows a fragment spectrum similar to that of shikimic acid. The direct injection experiment shows that m/z 111 is a primary peak in the MS3 of epi-quinic acid (2) (Figure 4.13). It can be only explained through the assumption of the presence of Q4 arising through the dehydration in between C4 and C5, which possess the anti-periplanar arrangement of a proton and –OH group as can be seen in the Figure 4.14.

Fragmentation of the muco-quinic acid (3) is explained in scheme C. Carboxyl group on C1 leaves as a formic acid and the dehydration takes place in between C2 and C3 due to anti- periplanar arrangement similar to the fragmentation in (-)-quinic acid (1) to give base peak in MS3 at m/z 127 (Q5). As shown in the scheme, when we invert the cyclohexane ring we see the anti-periplanar geometry between C2-H and C3-OH due to which, the dehydration between

C2 and C3 takes place. As can be observed in the direct injection experiment in Figure 4.13, m/z 109 fragments further to give m/z 81.2 similar to the MS4 of (-)-quinic acid (1). This means that the fragments Q2 and Q6 are in fact same molecules. Only difference in the fragmentation of the diastereomers 1 and 3 is that in the MS2 of 3, we observe a secondary peak at m/z 145, which signifies the decarboxylated molecule (Figure 4.10). In the structural arrangement of the diastereomer 3, absence of the hydrogen bonding between 1,3-di axially positioned oxygen anion on C1 and hydrogen on C3-OH does not facilitate the elimination of

C3-OH as observed in case of (-)-quinic acid (1).

119

O OH OH O O H H O A -HCO H -H O -H O O OH 2 OH OH 2 OH 2 H O OH MS2 O OH O OH O MS3 O H m/z 127 m/z 109 (M-H) of (-)-quinic acid (1) Q1 Q2

OH OH OH O H OH O OH OH OH 5 4 -H O -H2O -HCO2 6 1 5 2 HO B O O 3 OH 2 6 3 3 OH 2 1 4 MS OH OH MS OH OH 2 m/z 173 OHH m/z 111 (M-H) of epi-quinic acid (2) Q3 Q4

OH OH OH O O O H OH H O OH -H O -HCO2H OH -H2O OH 2 OH OH H 2 OH C HO OH 3 H 4 O OH MS2 O MS3 O O MS H OH m/z 127 m/z 109 (M-H) of muco-quinic acid (3) Q5 Q6 O O H O H H

D O OH O OH OH O H OH OH OH O OH OH OH OH H MS2 m/z 173 m/z 93 (M-H) of cis-quinic acid (4) m/z 111 Q7 Q8 Q9

Figure 4.14 Proposed mechanisms for the fragmentation of 1, 2, 3 and 4

120

MS2 of the cis-quinic acid (4) gives base peak at m/z 93 along with the secondary peaks at m/z 173 and m/z 111. Base peak at m/z 93 suggests that the decarboxylation and dehydration processes in the molecule had taken place simultaneously to attain aromaticity in the molecule. Unfortunately, phenol as an external standard and m/z 93 in compound 4 does not fragment further as observed in direct injection experiments where the collision energy was kept constant at 1.0 volts. But, we can propose mechanistic evidence as presented in scheme D in Figure 4.14 to prove that m/z 93 can be in fact, a phenolic ion. In the first step of the simultaneous fragmentation phenomena, intermediate Q7 is formed due to the anti-periplanar arrangement similar to the MS2 of epi-quinic acid (2). We assume that the dehydration takes place between C1 and C2 before decarboxylation due to the fact that the elimination of C1-OH is assisted by the hydrogen bonding provided by 1,3-diaxially positioned hydroxyl groups on

C3 and C5. Decarboxylation takes place leaving negative charge on C6. Possibility of the hydrogen bonding between the oxygen on C3-OH and hydrogen on the C5-OH considering the

1,3-diaxial arrangement, C3-OH leaves leaving a double bond between C6 and C5. The anti- periplanar geometry between C4-H and C3-OH helps the molecule to achieve aromaticity through dehydration to give Q8, which shows up as a secondary peak at m/z 111 in MS2. scyllo-quinic acid (5) shows unique fragmentation pattern if compared to the other diastereomers in present work. In MS2 it shows the pattern similar to the MS2 of the epi- quinic acid (2) and in MS3, m/z 173 directly fragments into m/z 93 similar to the MS2 of the cis-quinic acid (4) as can be seen in Figure 4.10. It is unclear at this point of time that how the fragmentation scheme in case of 5 can be explained mechanistically.

4.4 Discussion of the X-ray structures In the crystal of cis-quinic acid (4) (Figure 4.7), we observed two molecules in the asymmetric unit representing the same structure. Compounds 3-O-cinnamoyl-1,4-scyllo- quinide (9) and 3-O-cinnamoyl-1,5-quinide (10) were co-crystallized along with one molecule of chloroform in one asymmetric unit. The crystal structure of 3,5-Di-O-(tert- butyldimethylsilyl)quinate (7) shows conformational disorder in one of the silyl groups (Si1A and Si1B), both conformations are shown in Figure 4.3 (Conformer A) and Figure 4.4 (Conformer B). The disorder was modeled and showed a value of 50% for each conformation. Table 4.2 illustrates the crystal data and structure refinement for compounds methyl cis- quinate (17), cis-quinic acid (4), 3-O-cinnamoyl-1,4-scyllo-quinide(9), 3-O-cinnamoyl-1,5- quinide (10), 3,4-O-Cyclohexylidene-1,5-quinide (13) and methyl 3,5-Di-O-(tert- butyldimethylsilyl)quinate (7).

121

Table 4.2 Crystal data and structure refinement for compounds 17, 4, 9, 10, 15 and 7

Compound 17 Compound 4 Compounds Compound 15 Compound 7 9 and 10

Formula Unit C8H14O6 C14H24O12 C33H33Cl3O12 C13H18O5 C19H27O6Si2 Formula 206.19 384.33 727.94 254.27 407.59 weight (g/mol) Crystal Orthorhombic Monoclinic Monoclinic Orthorhombic Orthorhombic System

Space group P212121 P21/c C2 P212121 P212121 a (Å) 6.3490(2) 6.2971(2) 26.5045(12) 5.5977(6) 7.3775(5) b (Å) 9.6841(3) 19.7482(6) 5.9464(3) 10.6426(11) 11.1129(7) c (Å) 14.8574(5) 12.8471(4) 22.5590(12) 19.6591(19) 31.4879(19) α (°) 90 90 90.00 90 90 β (°) 90 94.218(2) 112.380(3) 90 90 δ (°) 90 90 90.00 90 90 Volume (Å3) 913.50(5) 1593.29(9) 3287.6(3) 1171.2(2) 2581.5(3) Z 4 4 4 4 4 3 Dcalc (g/cm ) 1.499 1.602 1.471 1.442 1.049 Abs. Coeff. µ 0.129 0.142 0.344 0.110 0.163 (mm-1) Temperature 100(2) K 100(2) K 100(2) K 100(2) K 100(2) K Total 35008 95659 36946 45621 66496 reflections Min-max θ 3.49 – 30.51 3.40 – 30.50 3.50 – 24.71 3.65 – 30.51 3.56 – 27.48 (°) Unique 2762 4848 5559 3521 5552 reflections Calculated 2532 4111 4229 3432 4739 reflection (I > 2σ) Final R1* 0.0308 0.0355 0.0565 0.0289 0.0883 wR2* 0.0813 0.1010 0.1500 0.0787 0.2501

Rint 0.0525 0.0557 0.1044 0.0527 0.0733 Goodness of 1.006 1.001 1.018 1.016 1.019 Fit Parameters 183 331 434 166 249 Restraints 0 0 1 0 0 Largest Peak/ 0.291/-0.229 0.543 /-0.210 0.467/-0.533 0.365/-0.199 1.262/-0.506 Deepest Hole Flack 0.0(6) n/a -0.02(9) -0.3(5) 0.0 (3) Parameter

2 2 2 2 2 1/2 * R1 = Σ║F(obs)│- │F(calc) ║/ Σ │F(obs) │; wR2 = { Σ[w(Fo – Fc ) ] / Σ[w(Fo ) ]}

122

4.5 Conclusions In conclusion, we have selectively synthesized all the diastereomers of the quinic acid except for the neo-quinic acid (6) to prove that they can be chromatographically resolved and can be identified by their characteristic fragmentation behavior in tandem MS spectra. This study also illustrates the importance of the tandem mass spectrometry in the area of identification and confirmation of the stereochemistry of comparatively smaller molecules like the diastereomers of the quinic acid. It is also worth mentioning that it is the first time the crystal structures of 3-O-cinnamoyl-1,4-scyllo-quinide (9) , 3-O-cinnamoyl-1,5-quinide (10), cis- quinic acid (4), methyl cis-quinate (17) and 3,4-O-Cyclohexylidene-1,5-quinide (13) are being reported. The fragmentation mechanisms are explained on the basis of the conformational differences between the diastereomers of the quinic acid. Presence of the diastereomers such as, muco-quinic acid (3), scyllo-quinic acid (5) and epi-quinic acid (2) in hydrolyzed roasted coffee sample proves the existence of the diastereomerized chlorogenic acids in roasted coffee. However, it still remains unclear if the epimerization occurs after the esterification with cinnamoyl functionality or the cinnamoyl group esterifies with epimerized quinic acid during the food processing.

123

References

1. E.L. Eliel, M.B. Ramirez. (-)-quinic acid: configurational (stereochemical) descriptors Tetrahedron- Asymmetry. 1997, 8, 3551.

2. Y. Koshiro, M.C. Jackson, R. Katahira, M. Wang, C. Nagai, H. Ashihara. Biosynthesis of chlorogenic acids in growing and ripening fruits of Coffea arabica and Coffea canephora plants. Z. Naturforsch., C: J. Biosci. 2007, 62, 731.

3. M. Weers, H. Balzer, A. Bradbury, O.G. Vitzthum. Analysis of acids in coffee by capillary electrophoresis. Colloq. Sci. Int. Cafe, [C. R.]. 1995, 16th, 218.

4. O. Frank, G. Zehentbauer, T. Hofmann. Bioresponse-guided decomposition of roast coffee beverage and identification of key bitter taste compounds. Eur. Food Res. Technol. 2006, 222, 492.

5. A. Crozier, I.B. Jaganath, M.N. Clifford. Dietary phenolics: chemistry, bioavailability and effects on health. Nat. Prod. Rep. 2009, 26, 1001.

6. B.M. Scholz-Boettcher, L. Ernst, H.G. Maier. New stereoisomers of quinic acid and their lactones. Liebigs Ann. Chem. 1991, 1029.

7. Anonymous. IUPAC Commission on the Nomenclature of Organic Chemistry (CNOC) and IUPAC-IUB Commission on Biochemical Nomenclature (CBN). Nomenclature of cyclitols. Recommendations, 1973 Biochem J. 1976, 153, 23.

8. R. Jaiswal, M.H. Dickman, N. Kuhnert. First diastereoselective synthesis of methyl caffeoyl- and feruloyl-muco-quinates. Org. Biomol. Chem. 2012, 10, 5266.

9. M.F. Matei, R. Jaiswal, N. Kuhnert. Investigating the Chemical Changes of Chlorogenic Acids during Coffee Brewing: Conjugate Addition of Water to the Olefinic Moiety of Chlorogenic Acids and Their Quinides. J. Agric. Food Chem. 2012, 60, 12105.

10. A. Crozier. Recent studies on the bioavailability of dietary (poly)phenolics. Abstracts of Papers, 245th ACS National Meeting & Exposition, New Orleans, LA, United States, April 7-11, 2013. 2013, AGFD.

11. C. Bennat, U.H. Engelhardt, A. Kiehne, F.M. Wirries, H.G. Maier. HPLC ANALYSIS OF CHLOROGENIC ACID LACTONES IN ROASTED COFFEE Zeitschrift Fur Lebensmittel- Untersuchung Und-Forschung. 1994, 199, 17.

124

12. K. Schrader, A. Kiehne, U.H. Engelhardt, H.G. Maier. Determination of chlorogenic acids with lactones in roasted coffee J.Sci.Food Agric. 1996, 71, 392.

13. K.H. Kim, Y.H. Kim, K.R. Lee. Isolation of quinic acid derivatives and flavonoids from the aerial parts of Lactuca indica L. and their hepatoprotective activity in vitro Bioorg.Med.Chem.Lett. 2007, 17, 6739.

14. M. Haribal, P. Feeny, C.C. Lester. A caffeoylcyclohexane-1-carboxylic acid derivative from Asimina triloba Phytochemistry. 1998, 49, 103.

15. H.C. Kwon, C.M. Jung, C.G. Shin, J.K. Lee, S.U. Choi, S.Y. Kim, K.R. Lee. A new caffeoyl quinic acid from Aster scaber and its inhibitory activity against human immunodeficiency virus-1 (HIV-1) integrase. Chem. Pharm. Bull. 2000, 48, 1796.

16. H.J. Kim, Y.S. Lee. Identification of new dicaffeoylquinic acids from Chrysanthemum morifolium and their antioxidant activities Planta Med. 2005, 71, 871.

17. P.T. Thuong, N.D. Su, T.M. Ngoc, T.M. Hung, N.H. Dang, N.D. Thuan, K. Bae, W.K. Oh. Antioxidant activity and principles of Vietnam bitter tea Ilex kudingcha Food Chem. 2009, 113, 139.

18. L. Sanchez-Abella, S. Fernandez, N. Armesto, M. Ferrero, V. Gotor. Novel and Efficient Syntheses of (-)-Methyl 4-epi-Shikimate and 4,5-Epoxy-Quinic and -Shikimic Acid Derivatives as Key Precursors to Prepare New Analogues. J. Org. Chem. 2006, 71, 5396.

19. S. Fernandez, M. Diaz, M. Ferrero, V. Gotor. New and efficient enantiospecific synthesis of (-)- methyl 5-epi-shikimate and methyl 5-epi-quinate from (-)-quinic acid. Tetrahedron Lett. 1997, 38, 5225.

125

CHAPTER 5: Transesterification of chlorogenic acids with small organic acids present in the coffee bean

5.1 Introduction Coffee is one of the most traded agricultural commodities in the world. More than 70 countries in the world cultivate coffee, which grows in the form of ‘cherries’ on a coffee plant. Brewed coffee is the third most consumed beverage after water and black tea.1 In countries like USA and Germany coffee is second most consumed beverage. Approximately, 450 million cups of coffee are consumed every day in United States only. The conventional coffee whether instant, filter or freshly ground is made from roasting, grinding and brewing the green beans of coffee obtained from cherry fruits of either Arabica or Robusta varieties of the coffee plant.

The perceived taste of the coffee is largely contributed by the acids present in the coffee. In fact, acidity in coffee is considered as one of the important parameters for the quality of the coffee variety.2 11 % of the total weight of the green coffee bean is contributed by the acid content, which decreases upon roasting to 6 %.3 This acid content is contributed by various volatile and non-volatile acids. In this work we will discuss about the non-volatile fraction of the total acid content. As Clifford reported earlier, in brewed coffee citric acid, phosphoric acid, phytic acid, quinic acid (1), chlorogenic acids and malic acid are the most important acids contributing to the perceived acidity.2 Other free organic acids present in roasted coffee such as oxalic acid (4), malonic acid (5), glutaric acid (7), adipic acid (8), tartaric (11) and succinic acid (6) 4-15 need to be considered as factors affecting the acidic taste as well. These acids do not contribute to the titrable acidity of the coffee as established by Engelhard and Maier 16 but they might exist as anions in a coffee brew or in a coffee extract providing protons. Apart from small organic acids, various authors have also studied the presence of free fatty acids (FFA) in coffee.17-19 FFA content of various green arabica and robusta coffee ranges from 0.8 to 3.0 g/100 g lipid.16 In the commercially available roasted coffees in German market it was found to be in the ranges of 0.8-1.8 g/100 g lipid, whereas 1.2-2.5 g/100 g lipid in French market.16 So, it is clear that although upon roasting FFA decrease slightly, they are present in coffee in roasted coffee in considerable amount. Among the reported FFA constituents in the roasted coffee, linoleic acid and palmitic acid are prominently present.16, 20

126

Approximately 10% of the total composition of the processed seeds of green Coffea canephora (robusta coffee) comes from chlorogenic acids on a dry weight basis.21 50% of the chlorogenic acids are lost through decomposition during roasting by producing half of the decomposition products such as, quinic acid and hydroxycinnamic acids through hydrolysis.1, 21, 22 Quinic acid (1) ranges from 3 to 6 g/kg in green robusta and Arabica coffee beans from various origins in the free form as reported earlier.15, 23 After steaming the green coffee bean, original quinic acid content rises by up to 15% as shown by Hucke and Maier.24 Similarly, roasting also helps to elevate the quinic acid content.3, 25, 26 Although chlorogenic acid lactones and quinic acid lactones are among the degradation products from chlorogenic acid, free quinic acid in roasted coffee is still found out in roasted coffee to maintain its existence between the ranges of 6.63 to 9.47 g/kg.27 Among the other non-volatile organic acids, citric and malic acid (9 and 10) are present in green coffee in the ranges of 5 to 15 g/kg in arabica and 3 to 10 g/kg in robusta respectively16 12 to 18 % of these acids are degraded during roasting process. Among many other decomposition products of citric acid occurring as a result of the various processes, succinic acid (6) and glutaric acid (7) are included in this work to study their significance towards the formation of transesterification products produced in coffee roasting.28 Despite considerably large amount of free non-volatile organic acids, degradation products and chlorogenic acids themselves exist in the roasted coffee, the understanding about the fate of these so-called melanoidine products remains hugely obscure. The green or roasted coffee beans contain a number of compounds such as, caffeine, trigonelline, lipids, phytosterols and series of small organic acids, which needs to be accounted for in analytical studies in future. Kuhnert et al. used mass spectrometry to characterize the chemical components of coffee melanoidines. They observed that the CGA derivatives undergo further condensation during coffee roasting to produce transesterified products such as, isomers of acetyl-caffeoyl quinic acid. They also found CGA derivatives condensed or transesterified with quinic acid and shikimic acid.1 Additionally, their work indicated the presence of the further esterification products of chlorogenic acids with small organic acids, which are present in the roasted coffee such as, maloyl caffeoyl quinic acid (at m/z 469.09876, C20H21O13) by studying the data generated by FT-ICR-MS measurements of the roasted coffee samples.1 This finding stands as an inspiration for the present work. Considering the fact that relatively large amount of the degradation products of CGAs such as, quinic acid (1), caffeic acid along with the most prominent member of the CGAs profile in

127

roasted coffee, 5-caffeoylquinic acid itself are present in the roasted coffee along with free small, non-volatile organic acids, we must entertain the possibility of the further esterification phenomenon among themselves. To investigate transesterification in roasted coffee in details we designed a thorough analytical plan involving four experiments. We have taken a set of small organic acids and heated each of them individually with 5-CQA (2) to check if simulated roasting conditions facilitate the formation of the transesterification products. Same experimental conditions were used incorporating caffeic acid and quinic acid as well. Also, we heated 5-CQA (2), caffeic acid (3) and quinic acid (1) with the mixture of all the organic acids separately to check, which of the organic acid show greater affinity towards the formation of the condensed esters. The set of eight organic acids contained oxalic acid, malonic acid, succinic acid, glutaric acid, adipic acid, citric acid, malic acid and dextrotartaric acid (Figure 5.1). They were chosen on the basis of their reported occurrence in green or roasted coffee and solid state. In each of the experiment, the roasting conditions were simulated by keeping the temperature at 200 oC for the duration of 12 minutes. All the samples acquired from these experiments were analyzed by high resolution ESI-TOF-MS. Four green coffee samples were also roasted in the conditions described earlier and then analyzed by ESI-FT-ICR-MS to identify the transesterification products in roasted coffee samples. In addition to these experiments, to study the possibility of the transesterification between quinic acid and free fatty acids present in the roasted coffee, we performed similar experiments involving quinic acid and linoleic and/or palmitic acid. Finally, mass spectral data obtained for the model roasting was compared to those obtained early on FT-ICR-MS data on roasted coffee extract and the presence of tentative transesterification products was thus confirmed in roasted coffee beans.

5.2 Materials and methods

5.2.1 Chemicals and materials All the chemicals (analytical grade) were purchased from Sigma-Aldrich (Bremen, Germany). The range of small organic acids was selected to purchase by their reported occurrence in coffee and solid state. Roasted coffee (Robusta) samples (powder) were obtained from the Kraft Foods Bremen (Germany).

128

O O HO HO OH OH OH OH HO OH HO O HO OH OH O OH OH O 1 2 3 (-)-qunic acid 5-Caffeoylquinic acid Caffeic acid

O OH O O O O O HO HO O HO OH OH HO OH O 4 5 6 7 Oxalic acid Malonic acid Succinic acid Glutaric acid

O O OH O OH OH O O O HO OH HO OH HO OH HO OH O OH O O OH

8 9 10 11 Adipic acid Citric acid Malic acid Tartaric acid

Figure 5.1 Structures of all the reactants involved in the model roasting experiments

5.2.2 Model roasting Two different roasting experiments were carried out taking 5-CQA (2), quinic acid and caffeic acid as substrates and oxalic acid (4), malonic acid (5), succinic acid (6), glutaric acid (7), adipic acid (8), citric acid (9), malic acid (10) and dextrotartaric acid (11) (Figure 1) as reactants. 5 mg (0.014 mmol) of 5-CQA was heated with equimolar quantity of each organic acid (4- 11) separately. 75 mg (0.212 mmol) of 5-CQA was heated with 0.125 equivalents of each organic acid collectively. 10 mg (0.055 mmol) of caffeic acid was heated with equimolar amount of each organic acid separately and then 100 mg (0.55 mmol) of caffeic acid was heated with 0.125 equivalents of each organic acid collectively. Quinic acid (1) (10 mg, 0.052 mmol) was heated with equimolar amount of each organic acid separately and then 100 mg (0.55 mmol) of quinic acid (1) was heated with 0.125 equivalents of each organic acid collectively. Lastly, quinic acid (5 mg, 0.026 mmol) was heated with equimolar amounts of linoleic acid and palmitic acid separately and then quinic acid (50 mg, 0.26

129

mmol) was heated with 0.5 equivalents of linoleic and palmitic acid collectively. All of the samples were heated in a Buechi Glass Oven B-585 at 200 oC for 12 minutes and then dissolved in methanol for LC-MS/MSn analysis.

5.2.3 Aqueous extract of roasted coffee Four different samples according to their origin of roasted coffee powder (5 g of each) were extracted with water by Soxhlet extraction using distilled water for 5 h. The extract was treated with Carrez reagents (1 mL of reagent A plus 1 mL of reagent B) to precipitate colloidal material and subsequently filtered through a Whatman no. 1 filter paper. The water was removed in vacuo and the residue was stored at -20 oC until required, thawed at room temperature, dissolved in methanol (6 g/L), filtered through membrane filter and used for LC-MS.

5.2.4 Roasted coffee samples for ESI-FT-ICR-MS analysis The samples were prepared by the procedure already reported by Jaiswal et al.1 LC-MSn The LC equipment (Agillent 1100 series, Bremen, Germany) comprised a binary pump, an auto sampler with a 100 µL loop and a DAD detector with a light-pipe flow cell (recording at 320 and 254 nm and scanning from 200 to 600 nm). This was interfaced with an ion-trap mass spectrometer fitted with an ESI source (Bruker Daltonics HCT Ultra, Bremen, Germany) operating in full scan, auto MSn mode to obtain fragment ion m/z. Tandem mass spectra were acquired in Auto-MSn mode (smart fragmentation) using a ramping of the collision energy. Maximum fragmentation amplitude was set to 1 Volt, starting at 30% and ending at 200%. MS operating conditions (negative mode) had been optimized using 5-caffeoylquinic acid with a capillary temperature of 365 oC, a dry gas flow rate of 10 L/min and a nebulizer pressure of 10 psi. LC-TOF-MS High Resolution LC-MS experiments were carried out using the same HPLC equipped with a MicrOTOF Focus mass spectrometer (Bruker Daltonics, Bremen, Germany) fitted with an ESI source. An internal calibration was achieved with 10 mL of 0.1 M sodium formate solution injected through a six port valve prior to each chromatographic runs. Calibration was carried out using the enhanced quadratic calibration mode. HPLC Separation was achieved on a 150 x 3 mm i.d. column containing diphenyl 5 µm, with a 5

130

mm x 3 mm i.d. guard column (Varian, Darmstadt, Germany). Solvent A was water/formic acid (1000:0.05 v/v) and solvent B was methanol. Solvents were delivered at a total flow rate of 500 µL/min. The gradient profile was from 10% B to 70% B linearly in 70 min followed by 10 min isocratic, and a return to 10% B at 90 min and 10 min isocratic to re-equilibrate.

ESI-FT-ICR mass spectrometry

Ultra high resolution mass spectra were acquired using a Bruker solarix Fourier transform Ion Cyclotron Resonance mass spectrometer (FTICR-MS) with a 12 T refrigerated superconducting cryo-magnet. The instrument was equipped with a Dual electrospray ion source with ion funnel technology. Spectra of the coffee samples were acquired in electrospray ionization positive and negative ion mode using direct infusion with a syringe pump with a flow rate of 120 μl/h. Stock solutions (1 mg/mL in MeOH) of the samples were diluted 1:40 in 50:50 MeOH: water for negative ion mode and 1:20 in 49.95:49.95 MeOH : water and 0.1% formic acid for positive ion mode. The drying gas temperature of the ion source was set to 200 °C. The instrument was externally calibrated with arginine cluster using a 10 μg/ml arginine solution containing 50% methanol. Spectra were acquired with 4 MW resulting in a resolving power of 500000 at m/z 400. Data were zero filled once to a data size of 8 MW. A single sine apodization was performed prior to Fourier transformation of the time domain signal. After spectra acquisition the mass spectra were internally calibrated with known polyphenolic pseudo-molecular ions. The ion accumulation time was set to 0.1 s and 300 scans were accumulated and added up for each mass spectrum in a mass range between m/z 200 and 3000.

131

O O O O HO HO HO HO O OH O OH

O O O O OH O O O O OH OH OH OH OH

O O O O OH O H O O O OH OH O OH OH 12 13 14 15

O O O OH OH HO HO OH OH O O O O OH O O OH O O HO OH O OH OH O O HO O O O O O O O OH OH OH O O O O O O OH O OH O OH O 16 17 18 19

O O O HO O HO HO O OH O O O O OH O O O O OH O O

O O OH OH O OH OH O OH OH OH O OH 20 21 22 O O OH HO O O O OH OH O OH OH O OH O O OH O OH O OH O HO O HO O HO O O O HO O O O OH OH OH O OH O 25 23 24

Figure 5.2 Tentative structures of transesterification products (Continued)

132

OH OH OH OH O HO O O OH HO O HO OH O O OH OH O O O O OH O O O O O O O OH HO O O O OH OH OH O O HO O OH OH 26 27 28

O O OH O OH O O OH OH OH OH O O OH O OH O O O O OH O O O O O O HO OH HO O HO O HO O

O O OH O O O O O O OH OH OH OH

29 30 31 32

O O O HO HO HO O OH O OH OH O O O O O O O O OH HO O OH OH O O HO OH

O OH O OH O OH O O OH OH OH OH O O OH OH O OH OH O

33 34 35 36

O O O O HO HO HO HO OH O OH OH O O O HO OH O HO HO OH O OH HO O O O O OH HO HO O O

37 38 39 40

Figure 5.2 Tentative structures of transesterification products (Continued)

133

O O O O O HO HO OH O HO HO HO O OH O HO OH HO HO HO OH HO O O O O O O O O O O O O HO O HO OH OH OH OH HO OH HO O O O O 41 42 O 43 44 45 O HO O O HO O OH HO HO OH O HO HO OH O HO OH O HO O O O O O O HO HO OH O OH O HO OH HO OH O O 48 O 47 O 45* 46 O OH O OH O O HO HO O O HO O OH O O HO HO O O O O O OH OH HO O O O O O OH O OH 49 50 51 52 OH O OH OH O O OH O O OH OH

HO O O

O O O O O O O OH HO O OH O O O O O O 53 54 55 OH 56 OH O O OH OH OH O OH OH O O O OH O

O O O O O OH HO HO O OH OH HO 59 57 58 O O Figure 5.2 Tentative structures of transesterification products (Continued)

134

O OH O O OH O HO OH OH

O O OH HO OH OH O OH O HO O O OH O OH H O O OH OH O O O H 62 60 61 OH 63

O O OH HO OH OH HO O OH O O O OH HO O O OH O O O OH O O OH HO OH HO OH 66 OH 64 65

O O O HO HO OH HO HO OH O O

HO HO OH O O O O O O OH O O O O

HO HO OH 68 69 70 67 OH HO HO OH OH

O O HO HO O O O O OH O OH OH OH O

72 71

Figure 5.2 Tentative structures of transesterification products of 1, 2 and 3 from LC-MSn, LC-TOF-MS and FT-ICR-MS data (regiochemistry of the condensation products is randomly selected).

135

5.3 Results and discussion Aim of this work is to achieve a better understanding of the chemical transformations occurring at very fine level in chlorogenic acids profile during food processing, which gives rise to interesting melanoidine reaction products. Analysis of melanoidines fraction in roasted coffee is a complicated venture, which was undertaken in this contribution with the help of the so called ‘Domino-Tandem MS’ approach proposed by Kuhnert.1 This approach combines two powerful modern mass spectrometry tools developed in recent times- high resolution MS and LC-tandem MS. High resolution MS provides ultimate resolution in the molecular weight dimension and provides lists of molecular formulas corresponding to melanoidine compounds, whereas LC-tandem MS provides resolution at isomeric level combined with gas phase isolation of targeted ions, adding structural information via fragment spectra. In the first step of the work, potential transesterification products of the respective substrates with each of the organic acid were postulated with the help of basic chemical reactivity principles. Roasting conditions were simulated to acquire the condensation products of 5-CQA (2), quinic acid (1) and caffeic acid (3) with the range of selected small organic acids, which were reported to be present in various green and roasted coffee samples in the literature. Secondly, identification of these transesterification products was undertaken in the samples generated by the model roasting experiments and in the roasted coffee samples. Tandem LC-MS was employed to obtain further structural information and likely reaction mechanisms involved in the formation of the transesterification products, which may have formed in the model roasting experiments as well as in the coffee samples. The analytical strategy involving high resolution FT-ICR-MS measurements as direct infusion experiments to obtain ultimate resolution in the molecular weight dimension and generate molecular formula lists of all compounds present in the coffee sample was also incorporated. The absence of the reference standards is a limitation encountered by presented work in the area of the assignment of the regio- and stereo-chemistry of the focused group of melanoidine fraction. However, employment of the combined analytical strategies in this work has enabled us to open up to the new possibilities of exploring roasted coffee melanoidine fraction for condensation products of acidic profile.

136

5.3.1 Transesterification of 5-CQA (2) in model roasting and in roasted coffee samples Figure 5.3 shows the representative UV chromatogram of the model roasting experiment sample generated by heating 5-CQA (2) with succinic acid (6). In the model roasting experiment where oxalic acid and 5-CQA were heated together we found four peaks, out of which three were identified as the isomers of caffeoyllactones as, formoyl- caffeoylquinide (12a-12c) and the fourth peak was assigned as formoyl-caffeoylshikimate (63) in LC-TOF-MS (Table 5.1). Malonic acid when heated with 5-CQA generated five peaks of isomeric esters, two of caffeoyllactones and three of caffeoylquinic acids as follows, acetoyl-caffeoylquinide (14a, 14b) and acetoyl-caffeoylquinic acid (15a-15c). Succinic acid gave eight peaks containing four isomers of succinoyl-caffeoylquinic acid (16a-16d), two isomers of succinoyl-caffeoylquinide (17a-17b) and two isomers of di-O- succinoyl-caffeoylquinic acid (18a, 18b). In case of glutaric acid with 5-CQA, we observed nine peaks generated by, six isomers of gluteroyl-caffeoylquinic acid (19a-19f) and three isomers of gluteroyl-caffeoylquinide (20a-20c). Adipic acid+5-CQA formed six isomeric compounds, out of which four were identified as the isomers of adipoyl- caffeoylquinic acid (21a-21d) and two as the isomers of adipoyl-caffeoylquinide (22a, 22b). Citric acid when heated with 5-CQA formed the highest number of transesterification products totaling to sixteen, which contained gluteroyl-caffeoylquinide (20b), three isomers of citroyl-caffeoylquinic acid (23a-23c), four isomers of 24, which is possibly an ester of degraded form of citric acid with quinic acid lactone, four isomers of citroyol-caffeoylquinde (25a-25d) and four isomers of methy-citroyl-caffeoylquinide (26a-26d). Fourteen peaks were observed in the EIC’s of respective m/z in negative mode of malic acid+ 5-CQA roasting. Four were generated by the isomers of maloyl- caffeoylquinic acid (28a-28d), two by isomeric methyl-maloyl-caffeoylquinic acid (29a, 29b), three by the isomers of maloyl-caffeoylquinide (30a-30c), three by isomeric methyl- maloyl-caffeoylquinide (31a-31c) and two peaks were generated by fumaroyl- caffeoylquinide (32). Lastly, tartaric acid showed five peaks in model roasting with 5- CQA contributed by, two isomers of tarteroyl-caffeoylquinide (34a, 34b) and three isomers of tarteroyl-caffeoylquinic acid (35a-35c). In the model roasting experiment of 5-CQA with the range of organic acids, we identified the total of 67 transesterification products in LC-TOF-MS listed in Table 5.1. Most of the transesterification products are formed with the lactones of the chlorogenic acid. Formoyl- caffeoyl-shikimate (63) and fumaroyl-caffeoyl-1,5-quinide (32b) were identified as only

137

two shikimate based dehydrated transesterification products arising from chlorogenic acid. The fragmentation pattern of these two compounds is shown in Table 5.2. In these, we can see that m/z 335, which is identical for the lactones and shikimates in MS2; fragments into m/z 161 and m/z 179 in MS3 only in the case of 63 and 32b respectively.

138

Table 5.1 Transesterification products of 4-11 with 5-CQA (2) identified with LC-TOF-MS in the samples generated by heating each acid separately with 5-CQA

No. Compounds Retention Condensation Mol. Formula Theoretical m/z Experimental m/z Error involved time (min) product No. (M-H) (M-H) (ppm)

1 Oxalic acid+5-CQA 35.3 12a C17H15O9 363.0722 363.0726 1.1

2 36.3 63 C17H15O9 363.0722 363.0724 0.7

3 38.1 12c C17H15O9 363.0722 363.0706 4.2

4 51.3 12d C17H15O9 363.0722 363.0732 2.8

5 Malonic acid+5-CQA 36.6 14a C18H17O9 377.0878 377.0886 2.2

6 39.8 14b C18H17O9 377.0878 377.0874 1.1

7 25.3 15a C18H19O10 395.0984 395.0967 4.2

8 33.0 15b C18H19O10 395.0984 395.0976 1.8

9 37.1 15c C18H19O10 395.0984 395.0972 3.0

10 Succinic acid+5-CQA 27.3 16a C20H21O12 453.1038 453.1024 3.1

11 28.4 16b C20H21O12 453.1038 453.1036 0.6

12 32.2 16c C20H21O12 453.1038 453.1036 0.7

13 33.6 16d C20H21O12 453.1038 453.1043 1.0

139

14 38.7 17a C20H19O11 435.0933 435.0942 2.0

15 41.4 17b C20H19O11 435.0933 435.0949 3.7

16 49.3 18a C24H23O14 535.1093 535.1070 4.3

17 50.6 18b C24H23O14 535.1093 535.1109 2.9

18 Glutaric acid+5-CQA 31.2 19a C21H23O12 467.1195 467.1204 2.0

19 32.6 19b C21H23O12 467.1195 467.1201 1.2

20 34.0 19c C21H23O12 467.1195 467.1209 3.1

21 35.0 19d C21H23O12 467.1195 467.1203 1.8

22 35.8 19e C21H23O12 467.1195 467.1200 1.0

23 39.1 19f C21H23O12 467.1195 467.1185 2.2

24 41.7 20a C21H21O11 449.1089 449.1094 1.0

25 44.0 20b C21H21O11 449.1089 449.1101 2.6

26 44.8 20c C21H21O11 449.1089 449.1097 1.8

27 Adipic acid+5-CQA 35.1 21a C22H25O12 481.1351 481.1344 1.6

28 38.0 21b C22H25O12 481.1351 481.1359 1.5

29 38.8 21c C22H25O12 481.1351 481.1357 1.1

140

30 39.7 21d C22H25O12 481.1351 481.1351 0.2

31 44.9 22a C22H23O11 463.1246 463.1254 1.8

32 47.9 22b C22H23O11 463.1246 463.1268 4.8

33 Citric acid+5-CQA 43.0 20b C21H21O11 449.1089 449.1096 4.2

34 26.6 23a C22H23O15 527.1042 527.1056 2.6

35 28.6 23b C22H23O15 527.1042 527.1050 1.4

36 29.0 23c C22H23O15 527.1042 527.1053 2.0

37 42.8 24a C21H19O11 447.0933 447.0924 1.9

38 43.6 24b C21H19O11 447.0933 447.0928 1.1

39 44.0 24c C21H19O11 447.0933 447.0920 2.9

40 45.3 24d C21H19O11 447.0933 447.0935 0.4

41 33.4 25a C22H21O14 509.0937 509.0933 0.8

42 36.1 25b C22H21O14 509.0937 509.0927 2.0

43 40.4 25c C22H21O14 509.0937 509.0929 1.5

44 41.0 25d C22H21O14 509.0937 509.0924 2.6

45 42.2 26a C23H23O14 523.1093 523.1080 2.5

141

46 42.7 26b C23H23O14 523.1093 523.1088 1.0

47 44.1 26c C23H23O14 523.1093 523.1079 2.6

48 45.1 26d C23H23O14 523.1093 523.1082 2.1

49 Malic acid+5-CQA 20.8 28a C20H21O13 469.0988 469.0983 0.9

50 22.3 28b C20H21O13 469.0988 469.0991 0.6

51 23.4 28c C20H21O13 469.0988 469.0999 2.4

52 26.3 28d C20H21O13 469.0988 469.0994 1.4

53 33.5 29a C21H23O13 483.1144 483.1141 0.6

54 35.8 29b C21H23O13 483.1144 483.1142 0.8

55 32.5 30a C20H19O12 451.0882 451.0887 1.1

56 33.7 30b C20H19O12 451.0882 451.0899 3.7

57 36.5 30c C20H19O12 451.0882 451.0894 2.7

58 37.4 31a C21H21O12 465.1038 465.1038 0.1

59 38.8 31b C21H21O12 465.1038 465.1036 0.5

60 40.5 31c C21H21O12 465.1038 465.1031 1.6

61 46.1 32 C20H17O11 433.0776 433.0771 1.3

142

62 33.5 33 C21H23O13 483.1144 483.1141 0.6

63 Tartaric acid+5-CQA 33.4 34a C20H19O13 467.0831 467.0830 0.2

64 37.3 34b C20H19O13 467.0831 467.0827 1.0

65 16.8 35a C20H21O14 485.0937 485.0945 1.7

66 18.5 35b C20H21O14 485.0937 485.0954 3.5

67 23.7 35c C20H21O14 485.0937 485.0953 3.4

Table 5.2 Transesterification products of 4-11 with 5-CQA (2) identified with LC-MSn in the samples generated by heating each acid separately with 5-CQA

No. Product Parent ion Characteristic m/z of ions in negative ion mode No. (M-H) 1 12a 363 MS2→ 335 (100), 173 (5); MS3 → 173 (100), 161 (11), 179 (19); MS4 → 110 (100), 92 (76), 136 (35), 71 (35) 2 63 363 MS2→ 335 (100), 317 (46); MS3 → 161 (100), 173 (75), 179 (70) ; MS4 →132 (100) 3 12c 363 MS2→ 335 (100); MS3 → 173 (100), 211 (65), 255 (45), 291 (34) 4 14a 377 MS2→ 335 (100), 317 (72); MS3 → 173 (100); MS4 →155 (100), 110 (58) 5 14b 377 MS2→ 335 (100), 317 (80); MS3 → 173 (100); MS4 → 110 (100), 92 (69), 71 (64), 129 (46) 6 14c 377 MS2→ 335 (100), 317 (66), 255 (29), 179 (37); MS3 → 173 (100), 161 (20), 179 (40); MS4 →110 (100), 137 (55), 71 (58), 129 (52), 86 (27) 7 14d 377 MS2→ 335 (100), 289 (5), 179 (9); MS3 → 173 (100), 179 (45), 161 (31); MS4 →155 (100), 110 (92), 92 (40)

143

8 17a 435 MS2→ 335 (100); MS3 → 173 (100), 179 (16), 161 (15); MS4 →110 (100), 92 (78), 71 (55), 155(34) 9 17b 435 MS2→ 335 (100); MS3 → 173 (100), 179 (42), 161 (43), 155 (8); MS4 →110 (100), 154 (65), 86 (47) 10 17c 435 MS2→ 335 (100); MS3 → 255 (100), 211 (83), 179 (60), 229 (30), 161 (58); MS4 →211 (100) 11 18 535 MS2→ 335 (100), 435 (36); MS3 → 173 (100); MS4 →154 (100), 110 (71), 92 (97) 12 20a 449 MS2→ 335 (100); MS3 → 173 (100), 179 (18), 161 (12); MS4 →110 (100), 92 (60), 155 (42) 13 20b 449 MS2→ 335 (100), 287 (20); MS3 → 173 (100), 179 (18), 161 (12); MS4 →155 (100), 127 (76), 110 (65), 92 (64) 14 20c 449 MS2→ 335 (100), 287 (34), 161 (20); MS3 → 255 (100), 210 (98), 173 (73), 291 (40), 228 (30); MS4 →210 (100) 15 19 467 MS2→ 305 (100), 353 (83), 335 (12), 191 (40), 406 (63); MS3 → 191 (100) 16 21a 481 MS2→ 353 (100), 319 (58), 191 (20); MS3 → 191 (100); MS4 →127 (100), 110 (88), 173 (68), 92 (79) 17 21b 481 MS2→ 319 (100), 353 (6), 460 (20); MS3 → 191 (100); MS4 →127 (100) 18 22a 463 MS2→ 335 (100), 300 (52), 161 (40); MS3 → 173 (100); MS4 →110 (100), 92 (76), 137 (48), 81 (43) 19 22b 463 MS2→ 301 (100), 335 (17), 161 (14); MS3 → 145 (100), 173 (84), 137 (71), 127 (57); MS4 →83 (100), 127 (22) 20 22c 463 MS2→ 301 (100), 335 (17), 161 (14); MS3 → 145 (100), 173 (87), 127 (37); MS4 →127 (100), 83 (22) 21 25 509 MS2→ 335 (100), 481 (17), 191 (27); MS3 → 255 (100), 291 (27), 227 (33), 211 (38) 22 24a 447 MS2→ 335 (100); MS3 → 173 (100); MS4 →110 (100), 154 (47), 100 (45), 92 (44) 23 24b 447 MS2→ 335 (100); MS3 → 173 (100), 255 (77), 211 (40), 291 (25) 24 30a 451 MS2→ 335 (100); MS3 → 173 (100), 179 (15); MS4 →110 (100), 92 (93), 71 (100), 85 (50), 155 (48) 25 30b 451 MS2→ 335 (100); MS3 → 173 (100), 179 (25), 161 (12), 134 (6); MS4 →155 (100), 110 (77), 92 (70), 71 (72) 26 32a 433 MS2→ 335 (100), 317 (20), 389 (32); MS3 → 173 (100), 179 (15); MS4 →110 (100), 92 (41), 71 (27), 59 (75) 27 64 433 MS2→ 335 (100), 389 (37); MS3 → 179 (100), 317 (12); MS4 →255 (100), 211 (63), 179 (96), 173 (53) 28 27a 671 MS2→ 335 (100), 353 (32), 191 (58); MS3 → 173 (100); MS4 → 110(100), 155 (16), 93 (42), 81 (27)

144

29 27b 671 MS2→ 509 (100), 353 (82), 191 (14); MS3 → 353 (100), 191 (47), 335 (10); MS4 → 191(100)

Table 5.4 Transesterification products of 4-11 with 5-CQA (2) identified with LC-TOF-MS in the roasted coffee samples

No. Compounds Retention Condensation Mol. Formula Theoretical m/z Experimental m/z Error involved time (min) product No. (M-H) (M-H) (ppm)

1 Oxalic acid+5-CQA 44.4 13 C18H17O12 425.0725 425.0709 3.9

2 Malonic acid+5-CQA 25.4 15a C18H19O10 395.0984 395.0990 1.7

3 33.2 15b C18H19O10 395.0984 395.0980 1.1

4 38.9 15c C18H19O10 395.0984 395.1002 4.7

5 Succinic acid+5-CQA 28.2 17c C20H19O11 435.0933 435.0916 3.9

6 32.6 17d C20H19O11 435.0933 435.0914 4.4

7 Glutaric acid+5-CQA 31.6 19a C21H23O12 467.1195 467.1184 2.4

8 Adipic acid+5-CQA 29.3 21a C22H25O12 481.1351 481.1344 1.6

9 Citric acid+5-CQA 29.8 27 C32H31O16 671.1618 671.1622 0.7

145

Although only one condensation product was identified involving 5-CQA and adipic acid in LC-MSn, which is adipoyl-caffeoylquinic acid (21a) in Figure 5.2, we found many condensation compounds having chlorogenic acid skeleton in LC-TOF-MS (Table 5.1). We were able to identify 29 transesterification products with their structural information by LC-MSn (Table 5.2, Figure 5.2).

Intens. UV Chromatogram, 254 nm [mAU] 15

5 10 15 20 25 30 35 Time [min]

Figure 5.3 UV chromatogram at 254 nm of the model roasting experiment sample generated by heating 5-CQA (2) with succinic acid (6)

Difference in the roasting conditions may have an effect on the integrity of the smaller organic acids more than the acids having higher molecular weight. In the model roasting experiments, oxalic acid (4) condenses with 5-CQA in its degraded form giving rise to the formates of the chlorogenic acid lactones and shikimates such as formoyl-caffeoyl-1,5- quinide (12) and Formoyl-caffeoyl-shikimate (63). But, the FT-ICR data of the roasted coffee confirms the presence of oxaloyl-caffeoylquinic acid (13), which is an oxalate of 5- CQA. Malonic acid was found to be condensed in its degraded form to produce acetates with either 5-CQA or with 5-CQA lactones such as, acetoyl-caffeoyl-1,5-quinide (14) and acetoyl-caffeoylquinic acid (15) in model roasting and in roasted coffee samples as well. Model roasting showed three transesterification products with succinic acid, succinoyl- caffeoylquinic acid (16), succinoyl-caffeoyl-1,5-quinide (17) and di-O-succinoyl- caffeoylquinic acid (18), all of which are identified in roasted coffee samples analyzed by FT-ICR-MS and LC-TOF-MS. Six isomers of compound 19 (glutaric acid+ caffeoylquinic acid) and three isomers of 20 (glutaric acid+ caffeoyl-1,5-quinide) were observed in LC-

146

TOF-MS and LC-MSn. Fragmentation pattern of 19 in tandem MS suggests that unlike most of the other transesterification products, glutaric acid unit is condensed at quinic acid part rather than the caffeoyl moiety. The fragmentation scheme is shown in Figure 5.4.

OH OH OH OH O O HO OH O 2 + HO OH MS O O HO O O O HO O O O O O O OH O O OH OH OH Parent ion at m/z 467 Base peak at m/z 305 Sec. peak at m/z 353 19 MS3

O HO O

HO OH OH

Base peak at m/z 191

Figure 5.4 Fragmentation scheme for compound 19 (glutaric acid+ caffeoylquinic acid)

Reaction of glutaric acid with 5-CQA gives isomeric products 19 and 20, which both are glutaric acid+caffeoylquinic acid derivatives are observed to be present in the roasted coffee also. The fragmentation patterns of some of the isomers of the transesters of adipic acid, which are adipoyl-caffeoylquinic acid and adipoyl-cafeeoyl-1,5-quinides (21b, 22b and 22c) are fully consistent with the fact that the transesterification is occurring at the quinic acid moiety similar to the previous case. As shown earlier, aliphatic dicarboxylic acids fragment preferentially if compared to hydroxycinnamic acid if esterified to quinic acid. However, as found here, this earlier finding cannot be generalized. (Figure 5.5). From the FT-ICR-MS data of the roasted coffee, presence of the products 21 and 22 in coffee can be confirmed, unfortunately we cannot comment on the position of the adipic acid transesterification. On the other hand, LC-TOF-MS data of the roasted coffee samples suggests that the esterification of the adipic acid is taking place at the caffeoyl moiety of 5-CQA (Table 5.3 and 5.4).

147

Table 5.3 Compounds transesterified with 5-CQA identified in FT-ICR-MS data of roasted coffee samples

No. Compounds Condensation Mol. Formula Theoretical m/z Experimental m/z Error involved product No. (M-H) (M-H) (ppm)

1 Oxalic 13 C18H17O12 425.072550 425.072438 0.3 acid+5-CQA

2 Malonic 14 C18H17O9 377.087618 377.087608 0.5 acid+5-CQA

3 15 C16H16O8 395.098370 395.098254 0.3

4 Succinic 16 C20H21O12 453.103850 453.103741 0.2 acid+5-CQA

5 17 C20H19O11 435.093285 435.091115 5.0

6 18 C24H23O14 535.109329 535.106990 4.4

7 Glutaric 19 C21H23O12 467.119500 467.119502 0.0 acid+5-CQA

8 20 C21H21O11 449.108935 449.108940 0.0

9 Adipic 21 C22H25O12 481.135150 481.135189 -0.1 acid+5-CQA

10 22 C22H23O11 463.124585 463.124635 -0.1

11 Citric acid+5- 23 C22H23O15 527.104244 527.102236 3.0 CQA

12 24 C21H19O11 447.093285 447.093547 -0.6

13 Malic acid+5- 28 C20H21O13 469.123403 469.123705 -0.6 CQA

14 29 C21H23O13 483.114414 483.114466 0.1

15 Tartaric 34 C20H19O13 467.083114 467.080960 4.6 acid+5-CQA

The highest number of the transesterification products in model roasting experiment is contributed by citric acid transesters (Table 5.1). We observed 15 condensation products with citric acid followed by the number of transesters generated by malic acid with 5-

148

CQA. In contradiction to this fact, the roasted coffee analysis by LC-TOF-MS and FT- ICR-MS we were able to identify only four transesterification products, which observed to be sourced from citric and malic acid together in FT-ICR-MS and one in LC-MS-TOF (Table 5.3 and Table 5.4). This can be attributed to the degradation of the citric acid during roasting process, which gives rise to the glutaric acid and succinic acid. Evidently, in roasted coffee, transesters arising from glutaric acid succinic acid totals to five, which is comparatively higher. Additionally, the identification of a gluteroyl-caffeoyl-1,5- quinide (20b) in LC-TOF-MS analysis (Table 5.1) of the sample generated by heating citric acid with 5-CQA proves that the degraded products of citric acid such as, gluatric acid generate transesterification products in model roasting conditions. LC-MSn does not provide as a reliable identification technique in the case of analysis of citric acid condensation products due to its inability to provide the molecular formula considering the fact that the molecular weight of the citric acid and quinic acid is the same. This on the other hand, clears the confusion in the structural assignment for m/z 671 by eliminating the possibility of the involvement of any citric acid moiety. We did not observe any tartates in the tandem MS but, we identified tarteroyl-caffeoyl-1,5-quinide (34) in roasted coffee in FT-ICR-MS analysis (Table 5.3).

149

Intens. -MS2(481), 35.3min [%] 319 100 191 353 461 0 -MS3(481->319), 35.3min 191 100

0 -MS4(481->319->191), 35.4min 127 100

0 50 100 150 200 250 300 350 400 450 m/z

Intens. -MS2(463), 48.9min [%] 301 100 161 335 0 -MS3(463->301), 48.9min 145 100 173

0 -MS4(463->301->145), 49.0min 83 100

0 50 100 150 200 250 300 350 400 450 m/z

Figure 5.5 Fragmentation patterns for m/z 481 (21) and m/z 463 (22)

If we observe the trend of the transesterification product formation when all the acids are heated in collective equimolar quantity with 5-CQA from Table 5.5 and Table 5.6 we see that the majority of the products are formed by the esters generated by glutaric and succinic acid (product numbers 16-20). This elevated selectivity towards transesterification products sourced from succinic and glutaric acid can be a result of the increased quantity of these acids in the mixture due to the degradation of the citric acid during model roasting experiment. This assumption is also supported by the fact that only one citrate transester is identified at m/z 509 in Table 5.6.

150

Table 5.5 Transesterification products of 4-11 with 5-CQA (2) identified with targeted LC- MSn in the samples generated by heating all of the acids collectively with 5-CQA

No. Product Parent Characteristic m/z of ions in negative ion mode No. ion (M-H) 1 18 535 MS2→ 335 (100), 435 (36); MS3 → 173 (100); MS4 →154 (100), 110 (71), 92 (97) 2 19 467 MS2→ 305 (100), 353 (83), 335 (12), 191 (40), 406 (63); MS3 → 191 (100) 3 27a 671 MS2→ 335 (100), 353 (32), 191 (58); MS3 → 173 (100); MS4 → 110(100), 155 (16), 93 (42), 81 (27) 4 27b 671 MS2→ 509 (100), 353 (82), 191 (14); MS3 → 353 (100), 191 (47), 335 (10); MS4 → 191(100)

Table 5.6 Transesterification products of 4-11 with 5-CQA (2) identified with LC-TOF-MS in the samples generated by heating all of the acids collectively with 5-CQA

No. Retention Condensation Mol. Formula Theoretical m/z Experimental Error time (min) product No. (M-H) m/z (M-H) (ppm)

1 38.5 14a C18H17O9 377.0878 377.0888 2.3

2 39.9 14b C18H17O9 377.0878 377.0876 0.6

3 25.4 15a C18H19O10 395.0984 395.0985 0.4

4 33.8 15b C18H19O10 395.0984 395.0989 1.4

5 37.3 15c C18H19O10 395.0984 395.0995 2.8

6 32.4 16c C20H21O12 453.1038 453.1031 1.6

7 39.9 17c C20H19O11 435.0933 435.0943 2.3

8 41.4 17b C20H19O11 435.0933 435.0995 3.9

9 30.9 19a C21H23O12 467.1195 467.1199 0.8

10 32.8 19b C21H23O12 467.1195 467.1179 3.4

11 34.1 19c C21H23O12 467.1195 467.1183 2.5

12 35.3 19d C21H23O12 467.1195 467.1203 1.8

151

13 40.0 19f C21H23O12 467.1195 467.1196 0.2

14 40.9 20a C21H21O11 449.1089 449.1105 3.5

15 44.1 20b C21H21O11 449.1089 449.1099 2.1

16 44.9 20c C21H21O11 449.1089 449.1102 2.9

17 43.0 20d C21H21O11 449.1089 449.1095 1.2

18 35.1 21a C22H25O12 481.1351 481.1350 0.3

19 38.1 21b C22H25O12 481.1351 481.1369 3.7

20 38.9 21c C22H25O12 481.1351 481.1361 1.9

21 45.0 22a C22H23O11 463.1246 463.1247 0.3

22 48.0 22b C22H23O11 463.1246 463.1262 3.5

23 41.0 25d C22H21O14 509.0937 509.0954 3.4

24 46.2 32 C20H17O11 433.0776 433.0793 4.0

25 16.7 35a C20H21O14 485.0937 485.0935 0.3

26 18.4 35b C20H21O14 485.0937 485.0945 1.8

5.3.2 Transesterification of quinic acid (1) in model roasting and in roasted coffee samples In the model roasting experiment of quinic acid (1) with the range of all organic acids (4-11) 37 isomeric transesterification products were identified by LC-TOF-MS listed in the Table 5.7. In the EIC’s at the respective m/z in negative mode of the structures we speculated to be forming in this experiment, we found that in case of oxalic acid heated with quinic acid; only one peak of oxaloylquinide was observed (36). Two peaks of two esters of quinic acid with malonic acid were identified to be the esters of quinic acid (37, 38). Succinic acid was found to be forming eight esterification products giving eight peaks, which were assigned to three isomeric esters of quinic acid (39) and five isomeric esters if quinic acid lactone (40). Glutaric acid when heated with QA was observed to be forming only one peak generated by an ester of quinic acid lactone (41). Adipic acid+ QA showed six peaks assigned to an ester of quinic acid (42) and five isomeric esters of quinic acid lactone (43). Surprisingly, citric acid generated only three peaks when heated with QA, out of which one was assigned to be an

152

ester of quinic acid (44) and the rest two as esters of quinic acid lactone (45, 46). Malic acid formed the highest number of esters with QA showing twelve peaks generated by two isomeric esters of quinic acid (47) and the rest ten of the isomeric esters were identified as esters of quinic acid lactone (48, 49). Four peaks were observed in the case of tartaric acid+QA, all of which were assigned to be the isomers of esters of quinic acid lactone (50). Figure 5.6 shows the representative total ion chromatogram in the negative mode of the model roasting experiment sample generated by heating QA (1) with glutaric acid (7). Similar to the case of model roasting experiment with 5-CQA, here we also observe that the majority of the transesterification products are formed with dehydration at the quinic acid part. Unfortunately due to the lack of the analytical data by LC-MSn, at the moment we cannot comment on whether the dehydrated products are shikimates or lactones. Hence, in the Figure 5.2, we have presented them as lactones for the purpose of simplification.

Intens. QA + Glutaric acid, TIC - x10 6

1.0

0.5

5 10 15 20 25 30 35 Time [min]

Figure 5.6 Total ion chromatogram in the negative mode of the model roasting experiment sample generated by heating QA (1) with glutaric acid (7)

In the model roasting experiment, we identified oxaloylquinide, which is oxalic acid condensed with the quinic acid lactone (36) (Table 5.7) but, in the roasted coffee sample oxalic acid appears in the form of its degraded product condensed with quinic acid to give formoylquinic acid (62) (Table 5.8). Except for the tartrates of the quinic acid or quinic acid lactones, all the other acids were found as the transesterification products in roasted coffee LC-MS analysis. Due to the unavailability of the structural information provided by the tandem MS, the position of the dehydration taking place i.e. either at QA part or at citric acid part in structure 45 is open for alternate interpretation. Twelve isomeric esters were identified

153

in the model roasting of malic acid with QA, which is highest number of the esters sourced from a single acid if compared to of other acids. Continuing the trend, malates of QA were found to have highest selectivity compared to other acids attributed to the number of the maloyl-quinic acid derivatives formed shown in Table 5.9.

154

Table 5.7 Transesterification products of 4-11 with quinic acid (1) identified with LC-TOF-MS in the samples generated by heating each acid separately with quinic acid (QA)

No. Compounds Retention Condensation Mol. Formula Theoretical m/z Experimental m/z Error involved time (min) product No. (M-H) (M-H) (ppm)

1 Oxalic acid+ QA 5.2 36 C9H9O8 245.0303 245.0308 2.3

2 Malonic acid+ QA 8.9 37a C9H13O7 233.0667 233.0677 4.5

3 5.3 38 C10H13O9 277.0565 277.0556 3.2

4 Succinic acid+ QA 6.6 39a C11H15O9 291.0722 291.0717 1.6

5 18.4 39b C11H15O9 291.0722 291.0714 2.7

6 28.1 39c C11H15O9 291.0722 291.0715 2.1

7 7.4 40a C11H13O8 273.0616 273.0607 3.2

8 10.9 40b C11H13O8 273.0616 273.0607 3.4

9 22.3 40c C11H13O8 273.0616 273.0604 4.2

10 23.0 40d C11H13O8 273.0616 273.0604 4.4

11 24.0 40e C11H13O8 273.0616 273.0608 2.9

12 Glutaric acid+ QA 16.4 41 C12H15O8 287.0772 287.0788 5.4

13 Adipic acid+ QA 14.1 42 C13H19O9 319.1035 319.1017 5.5

155

14 14.9 43a C13H17O8 301.0929 301.0916 4.4

15 16.2 43b C13H17O8 301.0929 301.0920 2.9

16 17.9 43c C13H17O8 301.0929 301.0922 2.2

17 18.6 43d C13H17O8 301.0929 301.0919 3.2

18 23.3 43e C13H17O8 301.0929 301.0914 4.8

19 Citric acid+ QA 7.0 44 C13H17O12 365.0725 365.0712 3.6

20 7.5 45 C13H15O11 347.0620 347.0602 5.2

21 12.1 46 C13H13O10 329.0514 329.0496 5.5

22 Malic acid+ QA 5.0 47a C10H15O10 307.0671 307.0670 0.1

23 4.3 47b C10H15O10 307.0671 307.0664 2.1

24 3.9 48a C11H13O9 289.0565 289.0563 0.7

25 4.4 48b C11H13O9 289.0565 289.0561 1.3

26 4.7 48c C11H13O9 289.0565 289.0563 1.8

27 6.0 48d C11H13O9 289.0565 289.0557 2.8

28 3.6 49a C11H11O8 271.0459 271.0457 0.8

29 3.9 49b C11H11O8 271.0459 271.0464 1.5

156

30 4.7 49c C11H11O8 271.0459 271.0457 1.0

31 9.0 49d C11H11O8 271.0459 271.0458 0.5

32 9.6 49e C11H11O8 271.0459 271.0460 0.4

33 11.5 49f C11H11O8 271.0459 271.0453 2.5

34 Tartaric acid+ QA 3.9 50a C11H13O10 305.0514 305.0505 3.1

35 4.7 50b C11H13O10 305.0514 305.0517 1.0

36 5.6 50c C11H13O10 305.0514 305.0510 1.3

37 7.0 50d C11H13O10 305.0514 305.0519 1.5

Table 5.11 Transesterification products of 4-11 with CA (3) identified with LC-TOF-MS in the roasted coffee samples

No. Compounds Retention Condensation Mol. Formula Theoretical m/z Experimental m/z Error involved time (min) product No. (M-H) (M-H) (ppm)

1 Oxalic acid+ CA 14.0 58 C20H13O10 413.0514 413.0530 3.8

2 Adipic acid+ CA 19.0 57 C15H15O7 307.0823 307.0816 2.5

3 Glutaric acid+ CA 41.4 56 C23H19O10 455.0984 455.0984 0.1

4 Succinic acid+ CA 45.3 54 C17H15O10 379.0671 379.0665 4.5

5 Malic acid+ CA 28.4 59 C22H17O11 457.0776 457.0759 6.7

157

Table 5.8 Compounds transesterified with quinic acid (QA) identified in FT-ICR-MS data of roasted coffee samples

No. Compounds Condensation Mol. Formula Theoretical m/z Experimental m/z Error involved product No. (M-H) (M-H) (ppm)

1 Oxalic 62 C8H11O7 219.051026 219.051024 0.0 acid+QA

2 Malonic 37 C9H13O7 233.066676 233.066700 0.1 acid+QA

3 Glutaric 41 C12H15O8 287.077241 287.077105 0.5 acid+QA

4 Adipic 43 C13H17O8 301.092891 301.092844 0.2 acid+QA

5 Citric 44 C13H17O12 365.072550 365.072629 0.2 acid+QA

6 45 C13H15O11 347.061985 347.062109 0.4

8 Malic 49 C11H11O8 271.045941 271.047040 4.1 acid+QA

Aliphatic esters of the quinic acid and its lactones can be termed as chlorogenic acids according to the modern definition (717, 726). Accordingly, we have found ten new chlorogenic acids in the analysis of the roasted coffee samples (Table 5.8 and Table 5.4) done in this work by LC-TOF-MS and FT-ICR-MS. Malonic acid generates two esters with quinic acid (QA) in the form of acetylquinic acid (37) and malonoylquinic acid (38), both of which were found to be present in the roasted coffee (Table 5.4). Seven isomers of succinoylquinic acid (39) and succinoyl-quinide (40) were identified in model roasting, out of which only one isomer (39a) was identified in roasted coffee by LC-TOF-MS. Glutaroyl- quinide (41) was identified in the roasted coffee in Table 5.8 similar to the observation from the model roasting experiment. Furthermore, adipoyl-qunide (43), citroylquinic acid (44), citroyl-qunide (45), fumaroylquinic acid (49) and formoylquinic acid (62) were identified in FT-ICR-MS data (Table 5.8) and adipoyl-qunide (43) and maloyl-quinide (48) were identified in LC-TOF-MS data (Table 5.4).

158

Table 5.9 Transesterification products of 4-11 with quinic acid (1) identified with LC-TOF- MS in the samples generated by heating all of the acids collectively with quinic acid (QA)

No. Retention Condensation Mol. Theoretical Experimental Error time (min) product No. Formula m/z (M-H) m/z (M-H) (ppm)

1 7.4 40a C11H13O8 273.0616 273.0607 3.1

2 16.4 41a C12H15O8 287.0772 287.0761 4.0

3 12.2 41b C12H15O8 287.0772 287.0763 3.2

4 14.2 42 C13H19O9 319.1035 319.1023 3.6

5 18.7 43d C13H17O8 301.0929 301.0913 5.4

6 23.2 43e C13H17O8 301.0929 301.0914 4.7

7 8.5 45 C13H15O11 347.0620 347.0635 4.5

8 4.5 48b C11H13O9 289.0565 289.0559 2.2

9 4.7 48c C11H13O9 289.0565 289.0557 2.8

10 3.6 49a C11H11O8 271.0459 271.0472 4.5

11 3.9 49b C11H11O8 271.0459 271.0461 0.7

12 4.7 49c C11H11O8 271.0459 271.0455 1.6

13 9.7 49e C11H11O8 271.0459 271.0450 3.5

14 4.7 50b C11H13O10 305.0514 305.0500 4.6

5.3.3 Transesterification of caffeic acid (3) in model roasting and in roasted coffee samples We report ten new esters of the caffeic acid (52-61) identified in roasted coffee samples by LC-TOF-MS and FT-ICR-MS (Table 5.10, Table 5.11 and Figure 5.2), out of which one was identified as a dimer of caffeic acid, three were found to be esters bound to a di-caffeoyl moiety (56, 58 and 59), three di-esters as oxalate (51), acetate (53) and succinate (54) were identified and four were identified as mono-esters of caffeic acid.

Model roasting experiment results lead us to concur that caffeic acid produces the least number of condensation products compared to 5-CQA and QA. We observed mono- and di-

159

acetate of caffeic acid (52, 53) as esters of degraded malonic acid along with compound 55, which also found to be a peculiar product of a model roasting experiment. It appears to be a dimer of caffeic acid possessing two possible positional isomers; both eluting at different retention times hence can be differentiated (Figure 5.7). However, 55 could not be identified in roasted coffee samples in the data acquired from both of the techniques. Interestingly, we observed the esters of the dimers of the caffeic acid with oxalic acid, glutaric acid and malic acid in roasted coffee samples (56, 58 and 59).

Table 5.10 Compounds transesterified with caffeic acid (CA) identified in FT-ICR-MS data of roasted coffee samples

No. Compounds Condensation Mol. Formula Theoretical m/z Experimental m/z Error involved product No. (M-H) (M-H) (ppm)

1 Malonic 52 C11H9O5 221.045547 221.045583 0.2 acid+CA

2 53 C13H11O6 263.056112 263.056078 0.1

3 Succinic 54 C17H15O10 379.067070 379.067254 0.5 acid+CA

4 Glutaric 60 C14H13O7 293.066676 293.066690 0.0 acid+CA

5 Adipic 57 C15H15O7 307.082326 307.082341 0.0 acid+CA

6 Dextrotartaric 61 C13H11O9 311.040856 311.040386 1.5 acid+CA

In addition to the esters of the small organic acids, we observed further molecular formulas in the FT-ICR-MS data, which confirms the presence of the esters between the free fatty acids and quinic acid in roasted coffee. Linoleoylquinic acid (71) (at m/z

453.285958, C25H41O7 with 0.4 ppm error) and palmitoylquinic acid (72) (at m/z

429.285901, C23H41O7 with 0.3 ppm error) were identified. LC-TOF-MS also confirms the presence of these two products in roasted coffee samples. In contradiction to our expectation, no corrosponding lactones such as palmitoylquinide or linoleoylquinide were found either in model roasting or in roasted coffee samples.

160

Intens. 55b EIC 341.0000 - 55a 2000

1500

1000

500

0 34 36 38 40 42 44 46 48 Time [min]

Figure 5.7 EIC at m/z 341

5.4 Conclusions In conclusion, we have identified 67 isomeric transesters between 5-CQA and small organic acids in simulated roasting experiment, out of which 16 compounds were observed to be present in roasted coffee samples. We established the relationship of the appearance of higher number of the esters generated by glutaric acid and succinic acid to the degradation of citric acid in roasting process. In this work we report ten new chlorogenic acids in the form of aliphatic esters of quinic acid. We have found that free quinic acid undergoes further esterification in coffee roasting by identifying the esters of all the small organic acids incorporated in this study except for the tarteroylquinic acid derivatives. Elucidation of the fate of the free caffeic acid in coffee melanoidines was achieved by reporting various products formed such as dimers of the caffeic acid, esters of the dimers of caffeic acid along with formates and acetates of caffeic acid. Confirmation of the presence of the esters of free fatty acids and quinic acid will help in the further investigation of the foam of the espresso coffee.

161

References

1. Jaiswal, R.; Matei, M.F.; Golon, A.; Witt, M.; Kuhnert, N. Understanding the fate of chlorogenic acids in coffee roasting using mass spectrometry based targeted and non-targeted analytical strategies. Food Funct. 2012, 3, 976-984.

2. Clifford, M. What factors determine the intensity of coffee’s sensory attributes. Tea Coffee Trade J 1987, 159, 35-39.

3. Maier, H. The acids of coffee. Proceedings 12th Asic College 1987, 229-237.

4. Mabrouk, A.F.; Deatherage, F.E. Organic acids in brewed coffee. Food Technol. (Chicago, IL, U. S. ) 1956, 10, 194-197.

5. Lentner, C.; Deatherage, F.E. Organic acids in coffee in relation to the degree of roast. Food Res. 1959, 24, 483-492.

6. Nakabayashi, T. Chemical studies on the quality of coffee. VI. Changes in organic acids and pH of roasted coffee. Nippon Shokuhin Kogyo Gakkaishi 1978, 25, 142-146.

7. Engelhardt, U.H.; Maier, H.G. Acids in coffee. XI. The proportion of individual acids in the total titratable acid. Z Lebensm Unters Forsch 1985, 181, 20-23.

8. Woodman, J.S.; Giddey, A.; Egli, R.H. Carboxylic acids of brewed coffee. Int. Colloq. Chem. Coffee, 3rd 1968, 137-143.

9. Hughes, W.J.; Thorpe, T.M. Determination of organic acids and sucrose in roasted coffee by capillary gas chromatography. J. Food Sci. 1987, 52, 1078-1083.

10. Engelhardt, U.H. Nichtflüchtige Säuren im Kaffee. Technische Universität Carolo- Wilhelmina zu Braunschweig: 1984

11. Clarke, R.; Vitzthum, O. Coffee: recent developments. Wiley. com: 2008

12. Clifford, M. Chemical and physical aspects of green coffee and coffee products, In Coffee, Anonymous ; Springer: 1985; pp. 305-374.

13. Scholze, A.; Maier, H. Quantitative Bestimmung von Säuren in Kaffee mittels Kapillar- Isotachophorese. Lebensm Chem Gerichtl Chem 1982, 36, 111-112.

14. Engelhardt, U.H.; Maier, H.G. Säuren des Kaffees. Zeitschrift für Lebensmittel- Untersuchung und Forschung 1985, 181, 206-209.

15. Scholze, A. Quantitative Bestimmung von Säuren im Kaffee durch Kapillar- Isotachophorese. Technische Universität Carolo-Wilhelmina zu Braunschweig: 1983

16. Balzer, H.H. Chemistry I: Non-volatile compounds. Acids in coffee. Coffee 2001, 18-32.

17. Carisano, A.; Gariboldi, L. Gas chromatographic examination of the fatty acids of coffee oil. J. Sci. Food Agric. 1964, 15, 619-622.

162

18. Speer, K.; Kölling-Speer, I. The lipid fraction of the coffee bean. Brazilian Journal of Plant Physiology 2006, 18, 201-216.

19. Wajda, P.; Walczyk, D. Relation between acid value of extracted fatty matter and age of green coffee beans. J. Sci. Food Agric. 1978, 29, 377-380.

20. Kurzrock, T.; Kölling-Speer, I.; Speer, K. Identification of dehydrocafestol fatty acid esters in coffee. 1998, 27

21. Crozier, A.; Jaganath, I.B.; Clifford, M.N. Dietary phenolics: chemistry, bioavailability and effects on health. Nat. Prod. Rep. 2009, 26, 1001-1043.

22. Maeaettae, K.R.; Kamal-Eldin, A.; Toerroenen, A.R. High-Performance Liquid Chromatography (HPLC) Analysis of Phenolic Compounds in Berries with Diode Array and Electrospray Ionization Mass Spectrometric (MS) Detection: Ribes Species. J. Agric. Food Chem. 2003, 51, 6736-6744.

23. Kampmann, B.; Maier, H.G. Säuren des Kaffees. Zeitschrift für Lebensmittel- Untersuchung und Forschung 1982, 175, 333-336.

24. Hucke, J.; Maier, H.G. Quinic acid lactone in coffee. Z Lebensm Unters Forsch 1985, 180, 479-484.

25. Galli, V.; Barbas, C. Capillary electrophoresis for the analysis of short-chain organic acids in coffee. Journal of Chromatography A 2004, 1032, 299-304.

26. Ginz, M.; Balzer, H.H.; Bradbury, A.G.; Maier, H.G. Formation of aliphatic acids by carbohydrate degradation during roasting of coffee. European Food Research and Technology 2000, 211, 404-410.

27. Scholz-Boettcher, B.M.; Ernst, L.; Maier, H.G. New stereoisomers of quinic acid and their lactones. Liebigs Ann. Chem. 1991, 1029-1036.

28. Bähre, F. Neue nichtflüchtige Säuren im Kaffee. Papierflieger: 1997

163

CHAPTER 6: Which spectroscopic technique allows best differentiation of coffee varieties: Comparing principal component analysis using data derived from CD-, NMR-, IR- spectroscopy and LC-MS in the analysis of the chlorogenic acid fraction in green coffee beans

6.1 Introduction In order to investigate parameters like geographic origin, varieties, adulterations, processing conditions, sensory properties, beneficial health effects, shelf-life or any other desirable or undesirable property of a food, a detailed knowledge of its composition and chemistry is required and therefore becomes foremost a problem of analytical chemistry. Once the chemical constituents of food have been elucidated comparison of the chemical profile of different samples allows differentiation between samples and identification of variations that are of interest to both producer and consumer. In the last decade statistical methods aimed at data reduction, have become the method of choice to undertake such a task, with multi-variant statistical methods, in particular principal component analysis (PCA), becoming increasingly popular.3 The main philosophy of PCA is to reduce a large data set obtained from of a large number of samples by using a selected spectroscopic method, in order to extract the most important variations between the samples without any loss of information. These variations are termed principal components of the samples, whereby each principal component is by definition orthogonal to the next. Ideally, variations between sample groups can be identified and through the variation of a spectroscopic parameter linked to a set of unique marker molecules. PCA is mainly employed as an unsupervised pattern recognition technique providing visualization of a multivariate dataset, thereby revealing trends, observations and outliers. This visualization is achieved by the transformation of variables into a covariance based coordinate system with the principal components as axis, thereby creating a two dimensional representation termed score plot, from which a grouping or pattern of sample groups can be extracted. Next to the score plot a so called loadings plot provides information about the origin of the variances at a molecular level. The success story of PCA has started with Nicholson’s work using high resolution NMR data to identify disease related biomarkers from urine or plasma samples.4 Using PCA NMR data of large patient groups was successfully compared and unique biomarkers for certain diseases identified.4 164

PCA using a wide variety of analytical techniques, including NMR-, IR-, Raman spectroscopy, HPLC, GC, or GC-MS, has been employed as an established statistical method in other areas of research including metabolomics, food analysis and medical research. An important question, which to our knowledge has never been answered, is which spectroscopic technique is most suitable for the differentiation of a given set of food samples. Methods used vary in their practicability and information content. For example, IR- or Raman analysis provides rapid measurements, omitting sample preparation and using portable inexpensive instrumentation, within minutes a measurement and hence ideally a reliable result that allows distinction between the samples. However, distinction between the samples is frequently based on overlapping peaks corresponding not to an individual molecular marker but rather to a large group or family of molecules present in the sample. Techniques like MS or NMR avoid this limitation; however, require extensive sample preparation, costly sophisticated equipment resulting in satisfactory information on the structures of individual markers being present in the sample. Due to the particular complexity of food, all of the techniques mentioned above, have severe limitations with respect to type of materials amenable to investigation, resolution, sensitivity and information provided. To our knowledge, this is the first time to use the performance of a wide selection of spectroscopic techniques, to evaluate their individual usefulness and potential in PCA analysis. In this contribution we report of the use of four different analytical techniques in the characterisation of a single food material. A direct comparison of the analytical techniques used is important to show that not all techniques succeed in sample differentiation at the same level. It is for the first time that the comparison of this kind has become possible. The ability of a certain technique to differentiate sample groups is directly linked to the nature of its chemical constituents and the techniques ability to provide analytical useful information on these constituents. As a food material we have chosen green coffee bean samples for the following three reasons: Firstly, we have in our research group acquired an intimate knowledge of the secondary metabolite profile and phytochemistry of this material, having over the last years identified around 100 different secondary metabolites in the green coffee bean, the large majority being chlorogenic acids.7-9 This moderate amount of secondary metabolites ensures additionally that the large majority of signals in any spectroscopic data sets can be reliably assigned to well characterised compounds. Secondly, coffee is an important commercial commodity, indeed after water and black tea the third most consumed beverage on this planet with an annual production of 4.5 Mt and a market value in excess of 5 Billion US$ of the raw material alone. Thirdly, green coffee beans are produced in 165 two varieties Caffea arabica and Caffea canephora (otherwise known as Robusta coffee) whose distinction and adulteration forms an important problem for the coffee industry. It should, however, be noted that distinction of intact green coffee beans by visual inspection is rather straightforward due to significant morphological differences between Robusta and Arabica coffee beans. Only in the case of processed coffee, either roasted or grounded a distinction based on chemical composition is required to which the methods presented here can be applied. Supposedly high quality coffee blends consist typically of 100% Arabica coffee beans. Lower quality, cheaper blends may have some proportion of Robusta beans, or they may consist entirely of Robusta. Arabica beans produce allegedly a superior taste in the cup, being more flavourful and complex than their Robusta counterparts. Robusta beans in contrast tend to produce a bitterer brew, with a musty flavour and stronger body. Obviously, this difference in sensory properties could be related to the individual phytochemical profile of the two coffee varieties and could be characterised by PCA. Metabolomics, phytochemical profiling using PCA based methods have been frequently applied to the problem of distinguishing green Arabica from Robusta coffee beans. Admittedly distinction between green bean Arabica and Robusta samples does not constitute a difficult scientific challenge as mentioned earlier. However, the distinction has been frequently used as a benchmark for analytical chemistry methodology and should be viewed as such. Briandet and Downey have used IR and NIR spectroscopy to study the differences between the two varieties.10,11 NIR has been further used by Esteban-Diez and Lyman to distinguish Arabica from Robusta green coffee beans.12,13 Wang et al. could show that as well Kona coffee could be distinguished from other varieties using FTIR spectroscopy.14 In all of this work, distinction between varieties was possible due to PCA analysis, however, due to the nature of the spectroscopic technique used, only spectroscopic bands corresponding to groups of compounds rather than individual phytochemical constituents could be identified. Rubayizy15 could show using Raman spectroscopy that levels of the terpene Kahweol and lipid content allows distinction between Arabica and Robusta green coffee beans. Materny and co-workers have demonstrated that Raman microscopy can be employed directly on a single green coffee bean to allow distinction between these two varieties, based on signals corresponding to lipids and chlorogenic acids.16 Valdenebro et al. could show that the geographic origin of green coffee beans can be identified using sterol profiles analysed by GC-MS.17 Korhonova et al. found using GC-MS based PCA that differences in volatile fractions exist between Arabica and Robusta beans.18 Mendonca and Alonso Salces were able 166 to distinguish Arabica and Robusta green coffee beans based on PCA data using HPLC analysis of chlorogenic acid profiles. We have recently reported on PCA analysis of green coffee beans using Raman spectroscopy and microscopy19, 20 and LC-MS methods, as well discussing the importance of scaling and normalisation procedures in identifying meaningful variations in PCA.21

6.2 Materials and methods All the chemicals (analytical grade) were purchased from Sigma-Aldrich (Bremen, Germany) and used as is. 28 different types of Arabica green coffee beans from different origins were purchased from the main supplier of coffee (Münchhausen, Bremen and supermarkets in Bremen Germany), and 10 different types of Robusta green coffee beans were obtained as a generous offer from D.R. Wakefield & Co. Ltd., London, England.

6.2.1 Statistical analysis Principal component analysis (PCA) was performed using the robust commercial software package Unscrambler (v 9.7; CAMO A/S). In order to discriminate between extracted CGA from Arabic and Robusta green coffee beans, a qualitative classification was performed by PCA. PCA is one of the most common multivariate analysis methods used to reduce the dimensionality of large data sets by finding combinations of variables that describe the major trends in the data 1. The first PC carries most information or most explained variance and the second PC carries the maximum residual information, which is not taken into account by the first PC. The similarity between the samples can be displayed by the PCA scores’ scatter plot of two PCs, which shows the distribution of the samples in a new frame plot according to their scores. The loading plot describes the relationship between variables and a specified principal component. The loading value of each variable on a specific PC reflects how much the variable contributed to that PC.

6.2.1.1 Methanolic extract of coffee beans 10 g of each sample of different green Robusta and Arabica coffee beans was frozen using liquid nitrogen before grinding. The methanolic extract was prepared by Soxhlet extraction using aqueous methanol (70%) for 5 hr. The extract was treated with Carrez reagent to precipitate colloidal material, and filtered through Machery-Nagel MN-615 folded filter paper. The methanol was removed in a rotary evaporator at reduced pressure. The aqueous residue was kept in a deep freezer at -80°C for 1.5 hr, followed by lyophilisation under 0.94 mbar for 24 hrs. using Christ Alpha 1-4 LSC in order to remove the water from the extracted 167

CGA under most protective conditions. The extracts of CGA were stored at -20°C until required. Before use for LC-MS, these extracts were thawed at room temperature, dissolved in methanol (60mg/10mL), and filtered through a membrane filter.

6.3 Experimental

1H-NMR spectroscopy

All coffee extracts samples were thawed at room temperature, dissolved in DMSO-d6 (10 mg of each coffee extract in 0.6 ml DMSO-d6), sonicated for three minutes. NMRs were measured within one hour after sonication. 1H NMR spectra were acquired on a JEOL ECX- 400 spectrometer operating at 400 MHz at room temperature, using a 5 mm probe.

Circular Dichroism spectroscopy

CD spectra were obtained using Jasco J-810 spectrometer. All coffee extracts samples were dissolved in DMSO (12.5 mg in 2 ml DMSO). Measurement conditions were kept as, Band width-1 nm, Response-0.5 sec, Sensitivity-standard, Wavelength range 280 to 400 nm, Data pitch-1 nm, Scanning speed-1000 nm/min, Accumulation-2. All measurements were done at room temperature.

IR-spectroscopy

IR spectra for all coffee extracts samples were obtained using Bruker vector 33 ATR spectrometer. Transmittance spectra were recorded against the wavelength range of 400 cm-1 to 4000 cm-1. Spectra were recorded directly of the dry coffee extracts samples without any solvent.

LC-TOF-MS

The LC equipment (Agilent 1100 series, Karlsruhe, Germany) comprised a binary pump, an auto sampler with a 100 μL loop, and a DAD detector with a light-pipe flow cell (recording at 320 and 254 nm and scanning from 200 to 600 nm). This was interfaced with a MicrOTOF Focus mass spectrometer (Bruker Daltonics, Bremen, Germany) fitted with an ESI source and internal calibration was achieved with 10 mL of 0.1 M sodium formate solution injected through a six port valve prior to each chromatographic run. Calibration was carried out using the enhanced quadratic mode.

HPLC

Separation was achieved on a 150 x 3 mm i.d. column containing diphenyl 5 μm, with a 5 mm x 3 mm i.d. guard column (Varian, Darmstadt, Germany). Solvent A was water/formic acid 168

(1000:0.005 v/v) and solvent B was methanol. Solvents were delivered at a total flow rate of 500 μL/min. The gradient profile was from 10% B to 70% B linearly in 60 min followed by 10 min isocratic, and a return to 10% B at 90 min and 10 min isocratic to re-equilibrate.

LC-MSn

The LC equipment (Agilent 1100 series, Karlsruhe, Germany) comprised a binary pump, an auto sampler with a 100 µL loop, and a DAD detector with a light-pipe flow cell (recording at 320 and 254 nm and scanning from 200 to 600 nm). This was interfaced with an ion-trap mass spectrometer fitted with an ESI source (Bruker Daltonics HCT Ultra, Bremen, Germany) operating in full scan, auto MSn mode to obtain fragment ion m/z. As necessary, MS2, MS3 and MS4 fragment-targeted experiments were performed to focus only on compounds producing a parent ion at m/z 397, 559, and 573. Tandem mass spectra were acquired in Auto- MSn mode (smart fragmentation) using a ramping of the collision energy. Maximum fragmentation amplitude was set to 1 Volt, starting at 30% and ending at 200%. MS operating conditions (negative mode) had been optimized using 5-caffeoylquinic acid with a capillary temperature of 365 oC, a dry gas flow rate of 10 L/min, and a nebulizer pressure of 10 psi.

Data processing

LC-MS data were processed using Data Analysis 4.0 (Bruker Daltonics, Bremen). Raw calibrated LC-MS data were further processed by Profile Analysis 2.0 (Bruker Daltonics, Bremen) and if required further processed using Origin 7.0 and Matlab. Buckets were created in an m/z value range between 300 and 900, unless stated otherwise, with a bucket size of 60 s and 1 Da. Kernels were defined as 20 s and 0.2 Da.

6.4 Results and discussion The aim of this contribution is to assess the value of PCA analysis in terms of the data obtained from four different spectroscopic techniques (IR CD, NMR and MS) and the ability of the PCA analysis to differentiate between samples, taking green coffee beans as an example. Detailed questions that require addressing are whether it is possible to distinguish green coffee beans according to several parameters including variety of the coffee (Arabica or Robusta), geographical origin of the coffee, growth conditions (e.g. altitude) or processing conditions. How should the PCA parameters and methods be chosen in order to achieve optimal distinction? Should a distinction by PCA be possible? How can such a distinction be rationalised on a molecular level? For each PCA scores plot a PCA analysis produces a so called loading plot, in which the most important data points that are responsible for the distinction are displayed. Ideally, a PCA 169 analysis provides a list of unique molecular markers, unique to each sample group, distinguishing two sample groups. What is the nature of these points in the loading plot? Are they molecular markers identified by PCA unique to a sample or not and if not how should PCA be carried out to identify unique molecular markers? In order to address all of these questions we have analyzed a series of aqueous methanolic extracts of 38 different green coffee bean samples by high resolution LC-ESI-TOF-MS in the negative ion mode, CD (circular dichroism), IR (Infrared) and NMR spectroscopy (Table 6.1). For the extraction process we used an optimized extraction method, if compared to previous work,1 using a mild Soxhlet method followed by protein removal with Carrez reagent and subsequent freeze drying to yield bright yellow to orange powders. A total of 38 commercial green bean coffee samples, 10 Robusta samples and 28 Arabica samples of different geographic origins were extracted. 6.4.1 LC-MS LC-MS conditions used were as described earlier.19 In addition to the high resolution mass measurements we carried out LC-ESI-tandem-MS measurements using an ion trap mass spectrometer to be able to assign individual compounds not only on the basis of retention time and high resolution m/z value, but as well to use fragmentation data for correct structure assignment. Similar to previous work, around 50-100 well resolved chromatographic peaks could be identified in each chromatogram and peaks assigned to individual distinct compounds, in the majority chlorogenic acids. A list of selected compounds identified is given in Table 6.2. A typical chromatogram of a Robusta sample is shown in Figure 6.1.

170

Intens. TANZROBUSTA_111.D: TIC -All MS x107 1.2

1.0 a

0.8

0.6

0.4

0.2

0.0 Intens. TANZROBUSTA_111.D: UV Chromatogram, 318-322 nm [mAU] 1000 b

800

600

400

200

0 10 20 30 40 50 Time [min]

Figure 6.1 Representative chromatogram of green coffee extract of sample No. 33 (Tanzania Robusta), a) TIC in negative ion mode; b) UV-VIS chromatogram monitored at 320 nm

6.4.2 Circular Dichroism spectroscopy The compounds under investigation within the green bean coffee extracts are almost exclusively chlorogenic acids and their individual profile is assumed to be suited for investigation by CD spectroscopy. CGAs are chiral non-racemic compounds of natural origin that carry depending on their structure one to three chromophoric moieties. Mono-acyl quinic acids produce simple CD spectra with a single maximum or minimum, whereas diacyl quinic acids produce more complex CD spectra with a characteristic Cotton effect, due to the interaction of the two distinct cinnamate chromophores. Therefore, PCA analysis using CD spectroscopy should be ideally suited to assess variations in the relative ratio of monoacyl to diacyl quinic acids. Typical CD spectra of 3-CQA and 3,5-diCQA are shown in Figure 6.2. To the best of our knowledge no PCA was ever carried out using CD spectroscopy, despite the fact that most natural products are chiral and many natural products possess diagnostic chromophores. Two characteristic CD spectra of a typical chlorogenic acid extract from green coffee beans are as well shown in Figure 6.2. From the spectra, the individual regions characteristic for single chromophore and dichromophoric systems can be appreciated. 171

50 3-CQA

0 50 100 150 200 250 300 350 400 450 CD (mdeg) CD -10

-20

-30

-40

8 3,5-diCQA 6

0 100 150 200 250 300 350 400 450 -2

CD (mdeg) CD -4

-6

-8

-10

-12

Figure 6.2 CD spectra of 3-CQA and 3,5-diCQA

A non-targeted (unsupervised) PCA analysis was carried out, in which the full data set was processed. Once the principal components were calculated, an inspection of the various PCA score plots allowed the identification of groups of samples. By inspection of the characteristics of each individual data point in the groups in the score plot a conclusion can be drawn with respect to the nature of these groups. Data points can thus be labelled according to the groups identified.

172

Table 6.1 Origins, nature and grouping of green bean coffee samples analysed and included in PCA analysis

Sample No. Origin / Type Arabica / Robusta Group

1 Tanzania Arabica A1

2 Guatemala SHG Arabica A1

3 Peru Bio Arabica A1

4 Nicaragua Maragogype Arabica A1

5 Kenya AA Arabica A1

6 Athiopien Wild Forest Bio Arabica A1

7 Athiopien Yivgachette Arabica A1

8 Athiopien Mokka Sidamo 2 Arabica A1

9 Reizaow Arabica A1

10 Coffeein Free Arabica A1

11 Costarica 2 Arabica A1

12 Brasilien 1 Arabica A1

13 Brasilien 2 Arabica A1

14 Maragogype Arabica A2

15 Malawi Pamwamba Arabica A2

16 Panama Boquete Arabica A2

17 Kenia 1 Arabica A2

18 Honduras Bio Arabica A2

19 Kameruls Arabica A2

20 Nicaragua Mataglpa Arabica A2

21 Costarica 1 Arabica A2

22 Columbia Exulso Arabica A2

23 Papua Neuguinea Arabica A3

24 Athiopien Mokka Sidamo 1 Arabica A3 173

25 Costarica 3 Arabica A3

26 Ethiopien Arabica A3

27 Indian Perl Mountain Arabica A3

28 Brazilien Santos Arabica A3

29 Indian 1 Robusta R

30 India Cherry AB Robusta R

31 Uganda Robusta R

32 India Parchment Robusta R

33 Tanzania Robusta R

34 Indonesia 1 Robusta R

35 Togo 1 Robusta R

36 Cameron Robusta R

37 Indonesia 2 Robusta R

38 India Cherry A Robusta R

Table 6.2 Numbering, nomenclature and high resolution MS data of selected secondary metabolites identified in green bean coffee samples 21, 22 No. Name Mol. Theor. m/z (M- Exp. m/z (M- Error formula H) H) (ppm)

1 3-O-caffeoylquinic acid C16H18O9 353.0878 353.0881 -0.7

2 4-O-caffeoylquinic acid C16H18O9 353.0878 353.0884 -1.6

3 5-O-caffeoylquinic acid C16H18O9 353.0878 353.0892 -3.9

4 3-O-feruloylquinic acid C17H20O9 367.0929 367.1047 -3.4

5 4-O-feruloylquinic acid C17H20O9 367.0929 367.1038 -0.8

6 5-O-feruloylquinic acid C17H20O9 367.0929 367.1045 -2.9

7 3-O-p-coumaroylquinic acid C16H18O8 337.0929 337.0931 -0.5

8 4-O-p-coumaroylquinic acid C16H18O8 337.0929 337.0921 2.4

9 5-O-p-coumaroylquinic acid C16H18O8 337.0929 337.0921 2.4

10 3-O-dimethoxycinnamoylquinic acid C18H22O9 381.1191 381.1202 -2.8 174

11 4-O-dimethoxycinnamoylquinic acid C18H22O9 381.1191 381.1191 -2.5

12 5-O-dimethoxycinnamoylquinic acid C18H22O9 381.1191 381.1202 -2.8

13 3-O-sinapoylquinic acid C18H22O10 397.1140 397.1125 3.8

14 4-O-sinapoylquinic acid C18H22O10 397.1140 397.1150 -2.5

15 5-O-sinapoylquinic acid C18H22O10 397.1140 397.1140 -4.9

16 3,4-di-O-caffeoylquinic acid C25H24O12 515.1195 515.1190 1.0

17 3,5-di-O-caffeoylquinic acid C25H24O12 515.1195 515.1172 4.5

18 4,5-di-O-caffeoylquinic acid C25H24O12 515.1195 515.1170 4.9

19 3,4-di-O-feruloylquinic acid C27H28O12 543.1508 543.1512 -0.8

20 3,5-di-O-feruloylquinic acid C27H28O12 543.1508 543.1514 -1.1

21 4,5-di-O-feruloylquinic acid C27H28O12 543.1508 543.1539 -3.4

25 3-O-feruloyl-4-O-caffeoylquinic acid C26H26O12 529.1351 529.1343 1.7

26 3-O-caffeoyl-4-O-feruloylquinic acid C26H26O12 529.1351 529.1351 -0.1

27 3-O-feruloyl-5-O-caffeoylquinic acid C26H26O12 529.1351 529.1373 -4.0

28 3-O-caffeoyl-5-O-feruloylquinic acid C26H26O12 529.1351 529.1367 -3.0

29 4-O-feruloyl-5-O-caffeoylquinic acid C26H26O12 529.1351 529.1351 0.1

30 4-O-caffeoyl-5-O-feruloylquinic acid C26H26O12 529.1351 529.1349 0.5

31 3-O-dimethoxycinnamoyl-4-O- C27H28O12 543.1508 543.1488 3.6 caffeoylquinic acid

32 3-O-dimethoxycinnamoyl-5-O- C27H28O12 543.1508 543.1491 3.1 caffeoylquinic acid

33 4-O-dimethoxycinnamoyl-5-O- C27H28O12 543.1508 543.1526 -3.4 caffeoylquinic acid

34 3-O-dimethoxycinnamoyl-4-O- C27H28O12 543.1508 543.1508 -4.1 feruloylquinic acid

35 3-O-dimethoxycinnamoyl-5-O- C27H28O12 543.1508 543.1515 -1.4 feruloylquinic acid

36 4-O-dimethoxycinnamoyl-5-O- C27H28O12 543.1508 543.1525 -3.1 feruloylquinic acid

37 3-O-p-coumaroyl-4-O-caffeoylquinic C25H24O11 499.1246 499.1227 3.7 acid

38 3-O-caffeoyl-4-O-p-coumaroylquinic C25H24O11 499.1246 499.1247 -0.2 acid

39 3-O-p-coumaroyl-5-O-caffeoylquinic C25H24O11 499.1246 499.1248 -0.5 175

acid

40 3-O-caffeoyl-5-O-p-coumaroylquinic C25H24O11 499.1246 499.1247 -0.2 acid

41 4-O-caffeoyl-5-O-p-coumaroylquinic C25H24O11 499.1246 499.1246 -4.9 acid

42 4-O-p-coumaroyl-5-O-caffeoylquinic C25H24O11 499.1246 499.1249 -0.6 acid

43 3-O-p-coumaroyl-4-O-feruloylquinic C26H26O11 513.1402 513.1389 2.6 acid

44 3-O-p-coumaroyl-5-O-feruloylquinic C26H26O11 513.1402 513.1141 -2.9 acid

45 4-O-p-coumaroyl-5-O-feruloylquinic C26H26O11 513.1402 513.1406 -0.7 acid

49 3-O-sinapoyl-5-O-caffeoylquinic acid C27H28O13 559.1457 559.1481 -4.2

50 3-O-sinapoyl-4-O-caffeoylquinic acid C27H28O13 559.1457 559.1472 -2.6

51 3-O-(3,5-dihydroxy-4- C27H28O13 559.1457 559.1458 -0.2 methoxy)cinnamoyl-4-O-feruloylquinic acid

52 4-O-sinapoyl-3-O-caffeoylquinic acid C27H28O13 559.1457 559.1457 0.9

53 3-O-sinapoyl-5-O-feruloylquinic acid C28H30O13 573.1614 573.1641 -4.7

54 4-O-sinapoyl-5-O-feruloylquinic acid C28H30O13 573.1614 573.1599 -2.5

55 4-O-sinapoyl-3-O-feruloylquinic acid C28H30O13 573.1614 573.1634 -3.5

56 4-O-trimethoxycinnamoyl-5-O- C28H30O13 573.1614 573.1611 0.4 caffeoylquinic acid

57 3-O-trimethoxycinnamoyl-5-O- C28H30O13 573.1614 573.1623 -1.7 caffeoylquinic acid

58 3-O-trimethoxycinnamoyl-5-O- C29H32O13 587.1770 587.1748 3.8 feruloylquinic acid

59 3-O-trimethoxycinnamoyl-4-O- C29H32O13 587.1770 587.1766 0.7 feruloylquinic acid

60 4-O-trimethoxycinnamoyl-5-O- C29H32O13 587.1770 587.1764 1.0 feruloylquinic acid

61 3-O-dimethoxycinnamoyl-4-O-feruloyl- C37H36O15 719.1981 719.2001 -2.7 5-O-caffeoylquinic acid

62 3,4,5-tri-O-caffeoylquinic acid C34H29O15 677.1512 677.1522 -3.5

63 3,5-di-O-caffeoyl-4-O-feruloylquinic C35H31O15 691.1668 691.1647 3.1 acid 176

64 3-O-feruloyl-4,5-di-O-caffeoylquinic C35H31O15 691.1668 691.1711 -6.2* acid

65 3,4-di-O-caffeoyl-5-O-feruloylquinic C35H31O15 691.1668 691.1647 3.1 acid

66 3-O-caffeoyl-4,5-di-O-feruloylquinic C36H33O15 705.1825 705.1851 -3.8 acid

67 3,4-di-O-feruloyl-5-O-caffeoylquinic C36H33O15 705.1825 705.1833 -1.1 acid

68 3,4-di-O-caffeoyl-5-O-sinapoylquinic C36H33O16 721.1774 721.1795 -2.9 acid

69 3-O-sinapoyl-4,5-di-O-caffeoylquinic C36H33O16 721.1774 721.1766 1.1 acid

The PCA score and loading plots of the obtained CD data is shown in Figure 6.3 using 34 spectra of reasonable quality. While the majority of Robusta coffee samples cluster in one region and the majority of Arabica sample cluster in a second region differentiated on the PC 1 axis, a total of four Arabica samples fall within the plot area of the Robusta samples. Therefore it must be concluded that a reliable distinction, in the absence of further undue data manipulation, is not possible using CD data. 2 1,5 1 0,5 PC1 Arabica 0 -15 -10 -5 0 5 10 15 -0,5 -1 -1,5

-2 Robusta -2,5 PC2 -3

Figure 6.3 The PCA score and loading plots of the obtained CD spectral data (Continued)

177

Figure 6.3 The PCA score and loading plots of the obtained CD spectral data

6.4.3 Infrared spectroscopy As mentioned previously, vibrational spectroscopy has in the past been successfully used to differentiate between Robusta and Arabica green coffee bean samples using both IR and Raman techniques. Loading plots revealed that distinction is mainly based on ester C=O absorptions of CGAs and lipids. A typical ATR-IR spectrum is shown in Figure 6.4. The PCA score and loading plots of 20 samples providing good quality IR spectra is shown in Figure 6.5. As can be seen from the score plot in Figure 6.5, a reliable distinction between Arabica and Robusta samples is possible in the PC 1 dimension with loading plots confirming differences of absorption in the ester carbonyl region of the spectra.

178

Figure 6.4 ATR-IR spectrum of Panama Boguete Arabica extract

1,2

0,8

0,6

Arabica 0,4

0,2 PC1 0 -3 -2 -1 0 1 2 3 4 5 -0,2

-0,4

-0,6 Robusta

-0,8

-1 PC2

Figure 6.5 The PCA score and loading plots of the obtained IR spectral data (Continued)

179

Figure 6.5 The PCA score and loading plots of the obtained IR spectral data

6.4.4 1H NMR spectroscopy NMR based methods have been used previously on several occasions for the distinction between Arabica and Robusta green bean coffee samples.22-25 1H-NMR spectra of all 38 samples have been acquired in DMSO-d6 and a representative spectrum is shown in Figure 6.6. The spectrum is dominated by signals corresponding to chlorogenic acids, saccharides, caffeine and small organic acids. The resulting PCA score plot is shown in Figure 6.7.

180

1 Figure 6.6 H-NMR spectra of Tanzania Robusta in DMSO-d6

2,5 PC2 2 Arabica 1,5 Robusta 1

0,5

0 -1,5 -1 -0,5 0 0,5 1 1,5 2 2,5 3 3,5 4 -0,5 PC1

-1

-1,5

-2

Figure 6.7 The PCA score plot of the obtained NMR spectral data

181

Again the Robusta green bean samples cluster nicely in one region of the plot, whereas the Arabica samples are scattered all over the score plot making reliable distinction not possible. A closer analysis of the spectra reveals that CGA signals suffer from relative large chemical shift variations (> 0.2 ppm) presumably due to the well reported CGA-caffeine interactions. Non-covalent complexes stabilised by π-π interactions are formed between caffeine and CGAs in aqueous solutions resulting in concentration dependant aromatic solvent induced chemical shift changes and hence spectroscopic changes affecting the success of PCA analysis. This problem can be overcome by incorporating different NMR techniques such as 13C NMR spectroscopy as shown by Wei et al.31 and DOSY NMR as a virtual separation technique, which has been proved to discriminate the complex natural mixtures to the regioisomeric level by using matrix assisted approach as shown by Gresley et al.. 32

6.5 Conclusions In conclusion we have shown that PCA analysis using data sets obtained by different spectroscopic methods of identical coffee extracts show distinctly different success in distinguishing groups of samples. In the particular case of chlorogenic acid extracts from Arabica and Robusta coffee the use of LC-MS data and IR spectroscopy allowed straightforward distinction of sample groups. In contrast Circular Dichroism and NMR spectroscopy failed to give a satisfactory level of distinction. From these results it becomes obvious that one size does not fit all and that in order to carry out successful multivariant statistical analysis great care and consideration needs to be taken to choose the correct experimental technique. In this particular case NMR spectroscopy failed to achieve differentiation of samples since the many chlorogenic acid derivatives present in green coffee beans are structurally very similar and hence do not produce NMR signals, in which the individual components are well resolved. Secondly in NMR spectroscopy non-covalent interactions of CGAs with caffeine and an accompanying concentration dependant change of chemical shifts hinders a reliable analysis. From this we propose that prior to carrying out PCA analysis the ability of the spectroscopic method used to differentiate and resolve the compounds of interest must be assessed. Secondly it should be established whether interactions between individual components result in variations of experimental parameters.

182

References 1 J. W. Drynan, M. N. Clifford, J. Obuchowicz and N. Kuhnert, Natural Product Reports, 2010, 27, 417-462. 2 R. Jaiswal, T. Sovdat, F. Vivan and N. Kuhnert, Journal of Agricultural and Food Chemistry, 2010, 58, 5471-5484. 3 S. Wold, K. Esbensen and P. Geladi, Chemometrics and Intelligent Laboratory Systems, 1987, 2, 37-52. 4 J. K. Nicholson, J. C. Lindon and E. Holmes, Xenobiotica, 1999, 29, 1181-1189. 5 D. Krug, G. Zurek, B. Schneider, C. Bassmann and R. Muller, Lc Gc Europe, 2007, 41-42. 6 D. Krug, G. Zurek, B. Schneider, R. Garcia and R. Muller, Analytica Chimica Acta, 2008, 624, 97-106. 7 M. N. Clifford, K. L. Johnston, S. Knight and N. Kuhnert, Journal of Agricultural and Food Chemistry, 2003, 51, 2900-2911. 8 M. N. Clifford, S. Knight, B. Surucu and N. Kuhnert, Journal of Agricultural and Food Chemistry, 2006, 54, 1957-1969. 9 M. N. Clifford, S. Marks, S. Knight and N. Kuhnert, Journal of Agricultural and Food Chemistry, 2006, 54, 4095-4101. 10 R. Briandet, E. K. Kemsley and R. H. Wilson, Journal of Agricultural and Food Chemistry, 1996, 44, 170-174. 11 G. Downey, R. Briandet, R. H. Wilson and E. K. Kemsley, Journal of Agricultural and Food Chemistry, 1997, 45, 4357-4361. 12 D. J. Lyman, R. Benck, S. Dell, S. Merle and J. Murray-Wijelath, Journal of Agricultural and Food Chemistry, 2003, 51, 3268-3272. 13 I. Esteban-Diez, J. M. Gonzalez-Saiz, C. Saenz-Gonzalez and C. Pizarro, Talanta, 2007, 71, 221-229. 14 J. Wang, S. Jun, H. C. Bittenbender, L. Gautz and Q. X. Li, Journal of Food Science, 2009, 74, C385-C391. 15 A. B. Rubayiza and M. Meurens, Journal of Agricultural and Food Chemistry, 2005, 53, 4654-4659. 16 R. M. El-Abassy, P. Donfack and A. Materny, Food chemistry, in press. 17 M. S. Valdenebro, M. Leon-Camacho, F. Pablos, A. G. Gonzalez and M. J. Martin, Analyst, 1999, 124, 999-1002. 18 M. Korhonova, K. Hron, D. Klimcikova, L. Mueller, P. Bednar and P. Bartak, Talanta, 2009, 80, 710-715. 183

19 N. Kuhnert, R. Jaiswal, M. F. Matei, T. Sovdat and S. Deshpande, Rapid Communications in Mass Spectrometry, 2010, 24, 1575-1582. 20 R. A. van den Berg, H. C. J. Hoefsloot, J. A. Westerhuis, A. K. Smilde and M. J. van der Werf, Bmc Genomics, 2006, 7. 21 R. Jaiswal and N. Kuhnert, Rapid Communications in Mass Spectrometry, 2010, 24, 2283- 2294. 22 R. Jaiswal, M. A. Patras, P. J. Eravuchira and N. Kuhnert, Journal of Agricultural and Food Chemistry, 2010, 58, 8722-8737. 23 J. Baggenstoss, L. Poisson, R. Kaegi, R. Perren and F. Eschert, Journal of Agricultural and Food Chemistry, 2008, 56, 5847-5851. 24 S. Gal, P. Windemann and E. Baumgartner, Chimia, 1976, 30, 68-71. 25 I. M. Kamal, V. Sobolik, M. Kristiawan, S. M. Mounir and K. Allaf, Innovative Food Science & Emerging Technologies, 2008, 9, 534-541. 26. M. N. Clifford, W. Zheng and N. Kuhnert, Phytochemical Analysis, 2006, 17, 384- 393. 27 R. A. van den Berg, C. M. Rubingh, J. A. Westerhuis, M. J. van der Werf and A. K. Smilde, Analytica Chimica Acta, 2009, 651, 173-181. 28. D. Perrone, A. Farah, C. M. Donangelo, T. de Paulis and P. R. Martin, Food Chemistry, 2008, 106, 859-867. 29. M. N. Clifford, J. Kirkpatrick, N. Kuhnert, H. Roozendaal and P. R. Salgado, Food Chemistry, 2008, 106, 379-385. 30. M. N. Clifford, W. G. Wu and N. Kuhnert, Food Chemistry, 2006, 95, 574-578. 31. F. Wei, K. Furihata, M. Koda, F. Hu, R. Kato, T. Miyakawa and M. Tanokura, Journal of Agricultural and Food Chemistry 2012, 60 (40), 10118-10125. 32. A. Gresley, J. Kenny, C. Cassar, A. Kelly, A. Sinclair and M. Fielder, Food Chemistry 2012, 135 (4), 2879-2886.

184

Conclusions

In this study it was observed that the acyl migration phenomena occur before dehydration takes place at the quinic acid moiety. Acyl migration is facilitated in presence of the liquid media as compared to the roasting process. Therefore, the lower temperature roasts like the ‘cinnamon roast’ produce a large number of acyl migration products than higher temperature roasts which generate dehydration products like lactones and shikimates high in numbers.

Esters present on C3 position of the quinic acid are prone to hydrolysis of the ester bond than undergoing acyl migration in any experimental condition. The amount of esters present on C3 and C4 positions of quinic acid moiety in a cup of coffee after roasting and brewing processes is highly contributed by C5 positioned esters in case of mono-caffeoylquinic acids content. In contrast to this observation we found that acyl migration to C5 position from C3 and C4 is only possible in base hydrolysis i.e. it is highly pH dependant. 1,3-diCQA and 5-CQA were observed to be more stable than the rest of the reference standards in both roasting and brewing conditions.

In this contribution, muco-quinic acid, scyllo-quinic acid, epi-quinic acid and cis-quinic acid were selectively synthesized. Their behavior in LC-MSn along with commercially available (- )-quinic acid was studied. For the first time it was observed that these diastereoisomers are distinguishable on the basis of their fragmentation behavior as well as their chromatographic elution order. In this study, it was observed that muco-quinic acid, scyllo-quinic acid and epi- quinic acid are present in hydrolyzed Guatemala roasted coffee sample as possible products of roasting. Non selective isomerization of (-)-quinic acid using acetic acid/conc. H2SO4 was performed from which, epi-quinic acid, scyllo-quinic acid and (-)-quinic acid could be identified using newly assigned fragmentation schemes and retention times characteristic to the specific compound.

In this work, the further condensation of the CGAs and their decomposition products with the non-volatile fraction of the total acid content of the roasted coffee samples is reported. Selected small organic acids were heated individually with 5-caffeoyl quinic acid to check if simulated roasting conditions facilitate the formation of the transesterification products. Same experimental conditions were used incorporating caffeic acid and quinic acid as well. Also, 5- caffeoylquinic acid, caffeic acid and quinic acid were heated in presence of the mixture of all the organic acids separately to check, which of the organic acid show greater affinity towards the formation of the condensed esters. All the samples acquired from these experiments were analyzed by high resolution ESI-TOF-MS. Four green coffee samples were also roasted in the 185 conditions described earlier and then analyzed by ESI-FT-ICR-MS to identify the transesterification products in roasted coffee samples. Ten new chlorogenic acid derivatives were identified in the roasted coffee samples.

PCA analysis using data sets obtained by different spectroscopic methods of identical coffee extracts show distinctly different success in distinguishing groups of samples. In the particular case of chlorogenic acid extracts from Arabica and Robusta coffee the use of LC-MS data and IR spectroscopy allowed straightforward distinction of sample groups. In contrast, Circular Dichroism and NMR spectroscopy failed to give a satisfactory level of distinction. From these results it becomes obvious that one size does not fit all and that in order to carry out successful multivariant statistical analysis great care and consideration needs to be taken to choose the correct experimental technique. In this particular case NMR spectroscopy failed to achieve differentiation of samples since the many chlorogenic acid derivatives present in green coffee beans are structurally very similar and hence do not produce NMR signals, in which the individual components are well resolved. Secondly in NMR spectroscopy non- covalent interactions of CGAs with caffeine and an accompanying concentration dependent change of chemical shifts hinders a reliable analysis. It is established from this work that prior to carrying out PCA analysis the ability of the spectroscopic method used to differentiate and resolve the compounds of interest must be assessed. Secondly it should be established whether interactions between individual components result in variations of experimental parameters.

186

List of publications and manuscripts

1. Eravuchira, P.J.; El-Abassy, R.M.; Deshpande, S.; Matei, M.F.; Mishra, S.; Tandon, P.; Kuhnert, N.; Materny, A. Raman spectroscopic characterization of different regioisomers of monoacyl and diacyl chlorogenic acid. Vib. Spectrosc. 2012, 61, 10-16.

2. Jaiswal, R.; Deshpande, S.; Kuhnert, N. Profiling the chlorogenic acids of Rudbeckia hirta, Helianthus tuberosus, Carlina acaulis and Symphyotrichum novae-angliae leaves by LC-MSn. Phytochem. Anal. 2011, 22, 432-441.

3. Kuhnert, N.; Jaiswal, R.; Matei, M.F.; Sovdat, T.; Deshpande, S. How to distinguish between feruloyl quinic acids and isoferuloyl quinic acids by liquid chromatography/tandem mass spectrometry. Rapid Commun. Mass Spectrom. 2010, 24, 1575-1582.

4. Kuhnert, N.; Dairpoosh, F.; Jaiswal, R.; Matei, M.; Deshpande, S.; Golon, A.; Nour, H.; Karakose, H.; Hourani, N. Hill coefficients of dietary polyphenolic enzyme inhibitiors: can beneficial health effects of dietary polyphenols be explained by allosteric enzyme denaturing? J Chem Biol 2011, 4, 109-116.

5. Jaiswal, R.; Deshpande, S.; Kuhnert, N.; An investigation of the Hydroxycinnamate Profile of Six Galium Plants from the Rubiaceae Family. Phytochemical Anal. 2013, (Submitted).

6. Kuhnert, N.; Karakoese, H.; Jaiswal, R.; Deshpande, S. Investigating the Photochemical Changes of Chlorogenic Acids Induced by UV Light in Model Systems and in Agricultural Practice with Stevia rebaudiana Cultivation as an Example. 2013, (Manuscript) 7. Deshpande, S.; El-Abassy, R.M.; Jaiswal, R.; Eravuchira, P.J.; Kammer, B.; Materny, A.; Kuhnert, N. Which spectroscopic technique allows best differentiation of coffee varieties: Comparing principal component analysis using data derived from CD-, NMR-, IR- spectroscopy and LC-MS in the analysis of the chlorogenic acid fraction in green coffee beans. Analytical Meth. 2013, (Submitted)

8. Deshpande, S.; Jaiswal, R.; Matei, M.; Kuhnert, N. Acyl migration in mono- and dicaffeoylquinic acids under basic and aqueous acidic conditions and dry roasting conditions. 2013, (Manuscript)

9. Deshpande, S.; Matei, M.; Jaiswal, R.; Bassem, B.; Kuhnert, N. Synthesis, structure and tandem MS investigation of diastereomers of quinic acid. 2013, (Manuscript) 187

10. Deshpande, S.; Kuhnert, N. Transesterification of chlorogenic acids with small organic acids present in the coffee bean. 2013, (Manuscript)

188

Curriculum Vitae

Dr. Sagar Anil Deshpande

Seefahrtstrasse 7,

Bremen 28759,

Germany

Telephone (Germany): +49-17684146335

Telephone (India): +91-7719995630

Email: [email protected]

Nationality: Indian

Birth Place: Pune, India

Marital Status: Married

Skills  Synthesis, isolation, identification and structure elucidation of natural products.  Analysis of complex mixtures, food materials, beverages, food processing and economically important plants.  Expertise in analytical and preparative HPLC, LC-TOF-MS, FT-ICR- MS, LC-MS/MS, LC-Tandem-MS, MALDI-TOF, NMR, IR, CD and UV- Vis spectrometers.

Education Jacobs University Bremen 24th January 2014

 Doctor of Philosophy (PhD) in Chemistry with Prof. Dr. Nikolai Kuhnert Dissertation: “Mass Spectrometry Based Investigation of Chlorogenic Acid Reactivity and Profile in Model Systems and Coffee processing”, ten publications till date. 189

Jacobs University Bremen September 2010  Master of Science (MSc) in Nanomolecular Science, with the completion of Thesis entitled, “Analysis of Plant Secondary Metabolites with Different Analytical Techniques” with Prof. Dr. Nikolai Kuhnert. N. Wadia College, Pune University October 2007  Master of Science (MSc) in Organic Chemistry, with the completion of Thesis entitled, “Methoxycarbonylation of Amines with Organic Carbonates Catalyzed by Lead Compounds” with Mr. Sunil Joshi (Scientist E II), Homogeneous Catalysis Division, National Chemical Laboratory (NCL) Pune. S. P. College, Pune University May 2005  Bachelor of Science (BSc) in Chemistry.

Research PhD research, Jacobs University Bremen Experience September 2010– January 2014

Synthesis of natural products, especially diastereomers of (-)-quinic acid, chlorogenic acids, chlorogenic acid lactones and hydroxycinnamoyl- shikimates; synthesis of natural products in three to eight steps, using protecting group strategies; synthetic compounds characterization by melting point, IR, NMR, HRMS, Tandem-MS, XRD. Monitoring chemical transformations in chlorogenic acid profile in green coffee during coffee processing incorporating analysis of green and roasted coffee by LC-TOF-MS, FT-ICR-MS and LC-Tandem-MS. Structural elucidation using LC-tandem MS. Analysis of stevia, arnica, burdock and gardenia LC-TOF-MS and LC-Tandem- MS for their polyphenol contents.

MSc laboratory rotations and thesis research, Jacobs University Bremen August 2008– September 2010

Organic and organometallic synthesis of chlorogenic acids. NMR, IR, fluorescence, UV-visible and CD spectroscopy, Preparative HPLC, tandem and high resolution mass spectrometry coupled with LC/GC. Principal Component Analysis of green coffee bean extracts.

MSc thesis research, National Chemical Laboratory, Pune May 2006– January 2007

Catalytic synthesis of carbamates, optimization of the process through screening of substrates and catalysts.

Other Experience Teaching assistant, Advanced integrated organic and analytical chemistry laboratory, Jacobs University Bremen September 2013– December 2013 190

Supervising and grading BSc students for laboratory performance, delivering

introductory talks for experiments including the directions for instrument

handling.

Supervision of BSc research projects, Jacobs University Bremen

May 2012- July 2012

Training, directly supervising and partially grading students for BSc laboratory rotations and theses.

Resident associate for Bremische and Jacobs University Bremen May 2012 – Present

Representative of Jacobs University and member of Bremische staff. Ensuring harmony between students living off campus and locals. Providing assistance and solutions for tenancy related issues.

Awards Full merit based Jacobs University stipend for MSc studies

August 2008 – September 2010

PhD stipend for the third party funded project from Kraft Foods

September 2010 – August 2013

Language skills Marathi: Mother tongue English: Fluent

Hindi: Fluent German: Basic knowledge (A2)

Computer skills ChemDraw and ChemSketch.

SciFinder, Reaxys (Beilstein), Web of Knowledge, RefWorks and EndNote.

Data Analysis (4.0) and Profile Analysis (Bruker Daltonics).

Microsoft Office and Microsoft Operating Systems (Windows all versions).

References

Prof. Dr. Nikolai Kuhnert, FRSC Prof. Dr. Michael N. Clifford (Emeritus)

Room 117, Research III Centre for Nutrition and Food Safety

Chemistry, School of Engineering and Science School of Biomedical and Molecular Sciences

Campus Ring 8 University of Surrey

Jacobs University Bremen Guildford, Surry GU2 7XH, UK

Bremen 28759, Germany Email: [email protected]

Email: [email protected] (To be contacted by email only)

Phone: 0049-4212003120

191

List of publications and manuscripts

5. Jaiswal, R.; Deshpande, S.; Kuhnert, N.; An investigation of the Hydroxycinnamate Profile of Six Galium Plants from the Rubiaceae Family. Phytochemical Anal. 2013, (Submitted).

7. Deshpande, S.; El-Abassy, R.M.; Jaiswal, R.; Eravuchira, P.J.; Kammer, B.; Materny, A.; Kuhnert, N. Which spectroscopic technique allows best differentiation of coffee varieties: Comparing principal component analysis using data derived from CD-, NMR-, IR- spectroscopy and LC-MS in the analysis of the chlorogenic acid fraction in green coffee beans. Analytical Meth. 2013, (Submitted)

8. Deshpande, S.; Jaiswal, R.; Matei, M.; Kuhnert, N. Acyl migration in mono- and di-caffeoylquinic acids under basic and aqueous acidic conditions and dry roasting conditions. 2013, (Kraft Foods approval pending)

9. Deshpande, S.; Matei, M.; Jaiswal, R.; Bassem, B.; Kuhnert, N. Synthesis, structure and tandem MS investigation of diastereomers of quinic acid. 2013. (Kraft Foods approval pending)

10. Deshpande, S.; Kuhnert, N. Transesterification of chlorogenic acids with small organic acids present in the coffee bean. 2013. (Kraft Foods approval pending)

Conferences and Workshops

 Presented a poster in 5th International Conference on Polyphenols and Health (ICPH) in Sitges, Barcelona in 2011.  Presented a poster in GDCh-Wissenschaftsforum Chemie 2011 in Bremen, 04 - 07 September 2011  Presented a poster at the National Chemical Laboratory, Pune, India. This was a part of 6TH International Annual Symposium on Catalysis (CAMURE 6) January 2007  Mass Spectrometry Workshop, Jacobs University Bremen, Bremen, Germany, July 24 – 27, 2013