Phytochemical Characterization of Stevia rebaudiana by Hande Karaköse

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 (supervisor) Professor of Organic and Analytical Chemistry, Jacobs University Bremen Prof. Dr. Gerd-Volker Röschenthaler Professor of Chemistry, Jacobs University Bremen Dr. Adam Le Gresley Doctor of Organic Chemistry, Kingston University London

Date of Defense: December 21, 2012

School of Engineering and Science

Declaration of Authorship

I, Hande Karaköse, hereby declare that the thesis I am submitting is entirely my own original work unless where clearly indicated otherwise. I have used only the sources, the data and the support that I have clearly mentioned. This PhD thesis has not been submitted for conferral of degree elsewhere.

Bremen, November 30, 2012

Signature

Table of Contents

Acknowledgments...... i

Abbreviations ...... ii

List of Figures ...... iv

List of Tables ...... vii

Abstract ...... viii

Chapter Page

I. INTRODUCTION ...... 1

II. REVIEW OF LITERATURE...... 4

2.1. Steviol Glycosides in Stevia rebaudiana ...... 4 2.2. Pharmacology, Toxicology and Regulations ...... 5 2.3. Pharmacokinetics of Stevioside: Absorption, Metabolism, Excretion ...... 9 2.4. Biosynthesis of the Steviol Glycosides ...... 10 2.5. Analysis of Steviol Glycosides of Stevia rebaudiana...... 14 2.6. Phenolic Acids ...... 16 2.7. Proteomics of Stevia rebaudiana ...... 21 2.8. Lipid Analysis ...... 25

III. RESEARCH OBJECTIVE ...... 30

Chapter Page

IV. Steviol Glycosides Analysis by LC-MS ...... 31

4.1. Overview ...... 31 4.2. Materials & Methods ...... 31 4.2.1. Extraction method ...... 31 4.2.2. LC-MS analysis of steviol glycosides ...... 31 4.2.3. HPLC conditions ...... 32 4.2.4. Calibration curve of steviol glycoside standards ...... 33 4.2.5. Method Validation ...... 33 4.2.6. Solid phase extraction (SPE) of steviol glycosides ...... 33 4.3. Results & Discussion ...... 34 4.3.1. Identification of steviol glycosides ...... 36 4.3.2. Method Validation ...... 39 4.3.3. Comparison to SPE sample clean up ...... 40 4.3.4. Quantification of steviol glycosides ...... 42 4.4. Conclusion ...... 44

V. Polyphenols in Stevia rebaudiana ...... 45 5.1. Overview ...... 45 5.2. Materials & Methods ...... 45 5.2.1. Sample preparation ...... 45 5.2.2. LC-MS analysis of polyphenols ...... 45 5.2.3. Calibration curve of standard compounds ...... 46 5.2.4. Hydrolysis of flavonoid glycosides ...... 46 5.2.5. Statistical analysis ...... 46 5.3. Results & Discussion ...... 47 5.3.1. Characterization of chlorogenic acids ...... 50 5.3.2. Characterization of flavonoid glycosides ...... 54 5.3.3. Quantification of chlorogenic acids and flavonoid glycosides ...... 56 3.3.1. Sample variation ...... 57 3.3.2. Flavonoid quantification ...... 65 3.3.3. Principal component analysis (PCA) ...... 67 5.3.4. Statistical evaluation of quantification data of polyphenols in stevia ...69 3.4.1. Statistical spread of data ...... 69 3.4.2. Correlations ...... 70 3.4.3. Analysis of variance (ANOVA) ...... 76 5.4. Conclusion...... 80

Chapter Page

VI. Lipid Analysis of Stevia ...... 81

6.1. Overview ...... 81 6.2. Materials & Methods ...... 81 6.2.1. Extraction method ...... 81 6.2.2. Methyl ester formation ...... 81 6.2.3. GC-FID conditions...... 81 6.2.4. GC-MS conditions ...... 82 6.2.5. Calibration curve of FAME ...... 82 6.2.6. MALDI-TOF MS ...... 82 6.3. Results & Discussion ...... 83 6.4. Conclusion ...... 93

VII. Proteomics of Stevia ...... 94

7.1. Overview ...... 94 7.2. Materials & Methods ...... 94 7.2.1. Extraction of proteins ...... 94 7.2.2. Protein analysis ...... 95 7.2.3. MALDI-TOF MS conditions ...... 97 7.3. Results & Discussion ...... 98 7.3.1. SDS results ...... 98 7.3.2. 2D-SDS ...... 99 7.3.3. MALDI-TOF MS results ...... 100 7.4. Conclusion ...... 105

VIII. Summary ...... 106

IX. References ...... 107

APPENDIX ...... 113 Publications ...... 149 Curriculum Vitae

ACKNOWLEDGMENTS

This research has been completed with the support of a large number of people. I would like to express my gratitude to them. First of all, sincere thanks to my supervisor Prof.Nikolai Kuhnert for his guidance and expert advices during my study. I wish to acknowledge the support of European Union (Project DIVAS) and Jacobs University Bremen for the full scholarship and funding of the research. I am grateful also to Dr. Kienle at the University of Hohenheim for the valuable discussions. A special thanks to my friends, Agnieszka Golon and Rohan Shah for their assistance during the experimental work of the study. Also, thanks to my colleagues at Jacobs University for providing a pleasant working atmosphere and Anja Müller for the technical assistance. Finally, my warmest thanks belong to my parents Dilek and Nejdet and my brother Çağatay, for their confidence in me and for being always supportive and interested in my work.

i

Phytochemical Characterization of Stevia rebaudiana

ABBREVIATIONS

HPLC high performance liquid chromatography MS mass spectrometry LC-MS liquid chromatography coupled with mass spectrometry GC-MS gas chromatography coupled with mass spectrometry GC-FID gas chromatography coupled with flame ionization detector EI-MS electron impact ionization mass spectrometry TIC total ion chromatogram EIC extracted ion chromatogram BPC base peak chromatogram MS2/MS3 tandem mass ESI-MS electrospray ionization mass spectrometry m/z mass-charge ratio UV ultra-violet MALDI matrix assisted laser ionization TOF-MS time of flight mass spectrometry HR-MS high resolution mass spectrometry HILIC hydrophilic interaction chromatography CGAs chlorogenic acids CQA caffeoylquinic acid FQA feruloylquinic acid MeOH methanol ACN acetonitrile DXS deoxyxylose-5-phosphate synthase CDPS copalyl diphosphate synthase KS kaurene synthase KO kaurene oxidase KAH kaurenoic acid hydroxylase UGTs UDP-glycosyltransferases RT Retention time RebA rebaudioside A

Hande Karaköse Jacobs University Bremen ii

Phytochemical Characterization of Stevia rebaudiana

RebC rebaudioside C PCA principal component analysis ANOVA analysis of variance TCA trichloroacetic acid IEF isoelectric focusing HCCA α-cyano-4-hydroxycinnamic acid FAME fatty acid methyl esters FTICR Fourier transform ion cyclotron resonance APCI atomic pressure chemical ionization SPE solid phase extraction S/N signal to noise ratio LOD limit of detection LOQ limit of quantification RSD % relative standard deviation % K7g kaempferol-7-O-glycoside Q3g quercetin-3-O-glycoside KS kolmogorov-smirnov test DTT dithiothreitol TFA trifluoroacetic acid MVA mevalonic acid pathway MEP 2-C-methyl-D-erythritol 4-phosphate/1-deoxy-D-xylulose 5-phosphate pathway APS ammonium persulfate

Hande Karaköse Jacobs University Bremen iii

Phytochemical Characterization of Stevia rebaudiana

LIST OF FIGURES

Figure 1.General structure of steviol glycosides and representative structure of rebaudioside A Figure 2.Structures of few artificial sweeteners Figure 3.Structures of steviol metabolites Figure 4.Hypothetical excretion route of stevioside Figure 5.Steviol glycoside biosynthesis via the MEP pathway Figure 6.Alternative MVA pathway Figure 7.Examples for hydroxybenzoic and hydroxycinnamic acids Figure 8.General structure of quinic acid and one of chlorogenic acids as an example Figure 9.Generic structure of major classes of flavonoids Figure 10.Strategies for MS based protein identification Figure 11.Peptide fragmentation nomenclature Figure 12.Examples of lipid categories Figure 13.Total ion chromatogram in negative ion mode using C18 column of methanolic Stevia rebaudiana extract showing phenolics (chlorogenic acids, flavonoids) and steviol glycosides Figure 14.Base peak chromatogram of steviol glycosides obtained using HILIC column Figure 15.Mechanism of fragmentation in tandem MS spectra of rebaudioside A and rebaudioside E illustrating how isomeric compounds can be distinguished by tandem MS Figure 16.Tandem MS spectra of rebaudioside A (above) and rebaudioside E (below) in negative ion mode Figure 17.Tandem MS spectra of rebaudioside D in negative ion mode Figure 18.Total ion chromatograms for comparison of different amounts of material I in SPE cleanup procedure Figure 19.Total ion chromatograms for comparison of SPE cleanup of the stevia extract with materials I and II cartridges Figure 20.Radar plot of steviol glycoside concentrations varying between seven varieties (average values taken within +/- 3σ) and in comparison to non-EU samples Figure 21.Radar plot of steviol glycoside concentrations varying between all origins (average values taken within +/- 3σ) and in comparison to non-EU samples Figure 22.Base peak chromatogram in negative ion mode using C18 column of methanolic Stevia rebaudiana extract showing chlorogonic acids, flavonoids and steviol glycosides

Hande Karaköse Jacobs University Bremen iv

Phytochemical Characterization of Stevia rebaudiana

Figure 23.Structures of caffeoylquinic acids and flavonoid glycosides Figure 24.Extracted ion chromatogram of m/z 353 of three mono-caffeoylquinic acids 3- caffeoylquinic acid, 5-caffeoylquinic acid, and 4-caffeoylquinic acid (from left to right) in negative ion mode Figure 25.Consecutively MS, MS2 and MS3 spectra of 3-caffeoylquinic acid in negative ion mode Figure 26.Consecutively MS, MS2 and MS3 spectra of 4-caffeoylquinic acid in negative ion mode Figure 27.Consecutively MS, MS2, MS3 and MS4spectra of 3,5-dicaffeoylquinic acid in negative ion mode Figure 28.Consecutively MS, MS2, MS3 and MS4spectra of 4,5-dicaffeoylquinic acid in negative ion mode Figure 29.Chemical structure of four flavonoid aglycones identified in Stevia rebaudiana leaves Figure 30.Extracted ion chromatogram of m/z 447.0 in negative ion mode Figure 31.An example of tandem MS spectra for compound 1, revealing its identity as kaempferol glucopyranoside Figure 32.Fragmentation illustration on luteolin-7-glucoside Figure 33.Radar plot of individual chlorogenic acid concentrations varying between seven varieties (average values taken within +/- 3σ) and in comparison to non-EU samples Figure 34.Radar plot of mono- and di-acyl quinic acids concentrations varying between seven varieties (average values taken within +/- 3σ) and in comparison to non-EU samples Figure 35.Bar plot of total mono- and di-acyl quinic acids concentrations varying between seven varieties (average values taken within +/- 3σ) and in comparison to non-EU samples Figure 36.Map showing the origins of stevia cultivation within the project Figure 37.Radar plot of mono- and di-acyl quinic acids concentrations varying between all origins (average values taken within +/- 3σ) and in comparison to non-EU samples Figure 38.Bar plot of total mono- and di-acyl quinic acids concentrations varying between all origins (average values taken within +/- 3σ) and in comparison to non-EU samples Figure 39a.PCA analysis of phenol profile of 35 stevia leaf LC-MS datasets Figure 39b.PCA analysis of phenol profile of 40 stevia leaf LC-MS datasets Figure 40.Histogram of 5-CQA, showing the normal distribution of the dataset

Hande Karaköse Jacobs University Bremen v

Phytochemical Characterization of Stevia rebaudiana

Figure 41.Graph showing the correlation between a) 3,5-diCQA/4,5-diCQA b) 3-CQA/5-CQA c) 5-CQA/4,5-diCQA and d) 4-CQA/3,5-diCQA Figure 42.Linear dependency of cis-5-CQA with 5-CQA and two isomers of cis-4,5-diCQAs Figure 43.Amount of 5-CQA in mg/100g dry leaves from three harvests from location A (TCV) and location B (Amfilikeia) during 2011 Figure 44.log of trans/cis-5-CQA concentrations against the number of sunshine hours in the month for a total of ten harvests from six locations Figure 45.GC-MS chromatogram of total lipid extracts from Stevia rebaudiana leaves from sample (Uconor, Var.4) Figure 46.GC-MS chromatogram of FAME standard mixture Figure 47.Representative EI-MS spectra obtained from GC-MS measurement of stevia extract Figure 48.Structures of fatty acids in Stevia rebaudiana extract Figure 49.Fatty acid profile of average stevia leaf in % X:Y denominates the number of carbon atoms in the fatty acid (X) and the number of double bonds in the fatty acid (Y) Figure 50.MALDI-MS spectrum of total lipid extract in positive ion mode using 2,5-DHB as a matrix Figure 51.Chemical structures of terpenes identified in Stevia rebaudiana leaves Figure 52.GC chromatogram of methylesterified steviol and stevia extract Figure 53.Extraction procedure of proteins Figure 54.SDS gel for sample number 8 TCV harvest I, loaded on gel at different concentrations 1 mg/mL and 0.5 mg/mL Figure 55.2D-SDS separation of stevia total protein extract. 7cm strip of pH 4-7, where spot 1 and 2 are at 55 kDa, and spot 5 at 15kDa Figure 56.MALDI-TOF MS spectra and mass list of trypsin digested 2D-SDS spot and Mascot search result showing the sequence information for RuBisCO enzyme with the score of 45% Figure 57.Mascot search result of MALDI spectra Figure 58.MS/MS de novo sequencing of the m/z 1230. Series of y and b fragments are labeled Figure 59.Structure and fragmentation of m/z 1230 based on denovo sequencing

Hande Karaköse Jacobs University Bremen vi

Phytochemical Characterization of Stevia rebaudiana

LIST of TABLES

Table 1.High resolution mass spectrometrical data of stevia terpene glycosides in negative ion mode from LC-TOF MS analysis Table 2.Enzymes involved in biosynthesis of steviol glycosides Table 3.Steviol glycosides values from 166 samples Table 4.Chromatographic and MS data on flavonoid glycosides and CGAs present in stevia Table 5.Average values (taken within +/- 3σ) for chlorogenic acids in seven different varieties Table 6.Average values (taken within +/- 3σ) for chlorogenic acids between origins Table 7.Average values (taken within +/- 3σ) for chlorogenic acids between harvests Table 8.Comparison of average values (taken within +/- 3 σ) for chlorogenic acids between three harvests of same variety and origin Table 9.Flavonoid glycosides average values for two major flavonoids in samples between origins determined by LC-MS directly from extracts without hydrolysis Table 10.Flavonoid glycosides average values between varieties Table 11.Values for flavonoids quercetin, kaempferol, luteolin and apigenin determined after hydrolysis of total polyphenol fractions using HCl/MeOH, determined by LC-MS Table 12.Descriptive statistics of caffeoylquinic acids Table 13.Correlation coefficients of mono and di-CQAs Table 14.Correlation coefficients of cis isomers according to Spearman’s rule Table 15.Results of test of homogeinity of variances Table 16.ANOVA results for effect of origin on stevia CGA content Table 17.Test of homogeneity of variances Table 18.ANOVA results for effect of variety on stevia CGA content Table 19.Total lipid values in weight % from 46 samples Table 20.Quantities of polyunsaturated fatty acids Table 21.Retention time and NIST scores of some terpenes identified in stevia extract Table 22.Amount and properties of chosen stevia leaves for protein extraction Table 23.Preparation of separation and stacking gel for 2D SDS PAGE

Hande Karaköse Jacobs University Bremen vii

Phytochemical Characterization of Stevia rebaudiana

ABSTRACT

Stevia rebaudiana (Bertoni) is from the Asteraceae family of plants with significant economic value due to its high content of natural zero calorie steviol glycoside sweeteners in its leaves. The leaves contain ent-kaurene glycosides, comprising stevioside, rebaudioside A, B, C, D, E, F and dulcoside A. Rebaudioside A and stevioside are the most abundant diterpene glycosides (steviol glycosides) in the leaves. The phytochemical characterization of stevia leaves is playing an important role in a future EU consumption of stevia as a novel food. For this purpose, the chemical composition of stevia (phenols and steviol glycosides detailed, with lipids and proteins in representative cases) was studied and methods have been developed for quantitative and qualitative analysis. Stevia leaves cultivated in more than ten locations inside and outside of Europe with seven different varieties, corresponding to total of 166 stevia samples, were extracted and their chemical composition was profiled and quantified by LC-MS for steviol glycosides and polyphenols (chlorogenic acids and flavonoids). Profiling, identification and quantification of terpenoids and lipids were achieved by using GC, GC-MS and MALDI-TOF techniques. In addition, protein extraction and analysis was carried out to identify potentially allergenic proteins in stevia leaves. Protein separation and isolation was achieved with 2-dimensional electrophoresis (2DE) and MALDI-TOF MS analysis was performed for the identification of individual proteins. Furthermore, as stevia may cultivated within various regions of the EU with different soil and climatic conditions it is important to know whether an EU-common specification will be achieved and how stevia leaves from regions outside EU can be distinguished on a scientific basis. For this purpose, principal component analysis (PCA) was performed based on the LC-MS dataset of stevia phenols. In addition, effect of growth origin and variety on stevia secondary metabolite profile was analyzed statistically by ANOVA (analysis of variance).

Hande Karaköse Jacobs University Bremen viii

1. INTRODUCTION

Stevia rebaudiana (stevia or S.rebaudiana) is native to Paraguay and belongs to the Asteraceae family of plants. Stevia and its sweet taste was botanically described by M.S. Bertoni in 18991. The high content of natural, zero-calorie sweeteners in its leaves makes stevia of a significant economic value in the food industry in many applications as a sweetener. Interest in stevia products has dramatically increased recently due to its approval by European and US legislating authorities. Stevia is likely to become a major source of high-potency sweetener for the growing natural food market in the future.

The majority of the annual stevia production of an estimated 4000 t is produced in China and South America. The stevia crop has been shown to be highly adaptable to cultivation in many other parts of the world. S. rebaudiana occurs naturally on acid soils of pH 4 – 5, but will also grow on soils with pH levels of 6.5 – 7.5 making it an interesting alternative to plants cultivated on poor soils such as tobacco2.

Stevia contains ent-kaurene glycosides, comprising stevioside, rebaudioside A, B, C, D, E, F and dulcoside A (Figure 1, Table 1), which give the leaves its characteristic taste of 200-400 times sweeter than sucrose. Stevioside has a sweetening power 300 times that of sucrose, and rebaudioside A is 400 times sweeter than sucrose3. Rebaudioside A and stevioside are the most abundant compounds; steviolbioside and rebaudioside B are believed to be formed by partial hydrolysis during the extraction process4. The rest of the steviol glycosides (e.g. dulcoside A, rebaudioside C) are at trace levels. In addition to being a natural sweetener, steviol glycosides have functional and sensory properties superior to those of many other high-potency sweeteners. Stevia leaves can be used in their natural state (fresh or dried form), due to its high sweetening intensity. Only small quantities are needed for comparison with white sugar. It does not increase the blood sugar level therefore; it can be used by diabetics without adverse glycemic responses. The human fecal microflora hydrolyzes stevioside and rebaudioside A to their common aglycon steviol in 10 and 24 h, respectively but steviol is not degraded by the human body 5.

In addition to diterpene glycosides, a number of secondary plant metabolites have been identified from S. rebaudiana including labdane-type diterpenes, triterpenoids and steroids, phenolic acids (flavonoid glycosides and chlorogenic acids), and oil components. From S. rebaudiana, ten

1

Phytochemical Characterization of Stevia rebaudiana labdane-type diterpenoids were identified, including austroinulin,isoaustroinulin6, sterebins (A - H)7, 8. A triterpenoid, lupeol 3-palmitate, was also separated from stevia9. As plant sterols, β- sitosterol, stigmasterol and campesterol were identified from S. rebaudiana10.

The presence of chlorogenic acids (CGAs) and flavonoid glycosides in stevia leaves gives the plant additional health benefits, and it could as well affect its organoleptic properties. CGAs are a large family of esters formed between quinic acid and certain trans - hydroxycinnamic acids, most commonly caffeic, p-coumaric, and ferulic acid. Similar to chlorogenic acids the presence of flavonoid compounds adds a health benefit to the usage of stevia leaves in food products. Flavonoids are a class of secondary metabolites that are produced ubiquitously in fruits and vegetables. By definition flavonoids are compounds with a C6-C3-C6 structure comprising two aromatic ring, one fused as a benzopyran. The secondary metabolites of interest in the present study were; steviol glycosides, chlorogenic acids, flavonoid glycosides, lipids, volatile terpenes and proteins. The main objective of this project was to provide a scientific basis for a future EU specification for stevia. The steviol glycoside and polyphenol profile and quantities of stevia samples cultivated in different European and non-European countries with seven different botanical varieties, harvested at three different times were obtained by analyzing stevia leaf extracts using a HPLC-TOF MS system. The identification of the compounds was achieved by analyzing tandem mass spectra and high resolution mass spectrometry (HR-MS) and for selected unknown phenolic compounds spectroscopic MS rules previously developed in our laboratory was used to elucidate structures. Stevia leave proteins were purified separated and sequenced with an aim to identify potentially allergenic proteins using 2D gel electrophoresis and MALDI-TOF MS technique. Lipids and volatile terpenes were determined by subjecting non-polar solvent extracts of stevia leaves to GC-MS and MALDI-TOF MS. Identification of the compounds was achieved using NIST library and comparison of retention times and GC-MS data of fatty acid methyl ester standard mixtures. Statistical analysis (PCA, correlation studies and ANOVA) served for differentiating the non-EU and EU cultivated stevia samples and for studying the relations between each components and growth conditions of stevia.

Hande Karaköse Jacobs University Bremen 2

Phytochemical Characterization of Stevia rebaudiana

OH OR1 OH OH OH OH CH OH CH3 2 HO O HO

O OH O HO O HO OH OH O OH O OH H3C COOR General structure of steviol glycosides O O O O OH OH CH3 CH2 CH3 CH2 HO HO

HO HO O O CH3 Rebaudioside A CH3 Stevioside HO O O HO O O OH OH

Figure 1.General structure of steviol glycoside and representative structure of rebaudioside A.

Table 1.High resolution mass spectrometrical data of stevia terpene glycosides in negative ion mode from LC-TOF MS analysis

Compound R R1 Molecular Experimental Theoretical Relative Error Formula m/z (M-H+)- m/z (M-H+)- (ppm)

Steviol H H C20H30O3 317.0819 317.0717 9.0

2 1 Steviolbioside H glc - glc C32H50O13 641.3181 641.3179 0.4

Rubusoside Glc glc C32H50O13 641.3166 641.3179 2.0

2 1 Stevioside Glc glc - glc C38H60O18 803.3751 803.3707 5.5

2 1 Rebaudioside A Glc glc3 - glc C44H70O23 965.425 965.4235 1.6

1glc 2 1 Rebaudioside B H glc3 - glc C38H60O18 803.368 803.3707 2.8

1glc 2 1 Rebaudioside C Glc glc3 - rham C44H70O22 949.427 949.4286 1.7 (Dulcoside B) 1glc 2 1 2 1 Rebaudioside D glc - glc glc3 - rham C50H80O28 1127.4726 1127.4763 3.3

1glc 2 1 2 1 Rebaudioside E glc - glc glc - glc C44H70O23 965.4199 965.4235 3.7

2 1 Rebaudioside F Glc glc3 - xyl C43H68O22 935.4097 935.4129 3.5

1glc 2 1 Dulcoside A Glc glc - rham C38H60O17 787.3732 787.3758 3.3

Hande Karaköse Jacobs University Bremen 3

Phytochemical Characterization of Stevia rebaudiana

2. REVIEW OF LITERATURE 2.1. Steviol Glycosides in Stevia rebaudiana The most commercially important compounds from leaves of Stevia rebaudiana Bertoni (S.rebaudiana/stevia) are the sweet tasting ent-kaurene diterpenoid glycosides. Two main glycosides are stevioside and rebaudioside A. There are other related compounds including Rebaudioside B-E, Dulcoside A and C which occur as minor components. Summaries of the compounds from stevia are shown in Table 1. Diterpene glycosides from S. rebaudiana contain a common aglycone called steviol (13- hydroxy-ent-kaur-16-en-19-oic acid), and differ only in the glycosidic constituents attached at C-13 and/or C-19. Stevioside is the main sweet tasting glycoside in stevia (5-10 %) and was reported to be 250-300 times sweeter than sucrose. Rebaudioside A (2-4%) is the second most abundant ent-kaurene and sweetest compound in stevia, its sweetness is 400 times more than sucrose. It was reported to have a more pleasant taste and it is more water soluble than stevioside. Rebaudioside B, D, and E may be also present in minor quantities; however, it is suspected that rebaudioside B is a byproduct of the isolation technique11. The two main compounds stevioside and rebaudioside, primarily responsible for the sweet taste of stevia leaves, were first isolated by two French chemists, Bridel and Lavielle (1931)12. The diterpene, steviol (Table 1) is the aglycone of stevia glycosides. Diterpene glycosides form with the formation of ester bond between glucose molecule and carboxyl group of steviol and replacing of hydroxyl hydrogen with combinations of glucose, rhamnose and xylose. Stevioside has two linked glucose molecules at the hydroxyl site, whereas rebaudioside A has three glucoses, with the central glucose of the saccharate connected to the central steviol structure. Rebaudioside C and Dulcoside A possess a rhamnose sugar, whereas Rebaudioside F possesses one xylose unit in its structure. After sensory panel testing, Rebaudioside A was reported to have the least bitterness of all the steviol glycosides in the stevia plant. Glycosides are molecules that contain glucose and other non-sugar substances called aglycones. The taste receptor of tongue reacts to the glucose in the glycosides, thus steviol glycosides with more glucose (e.g. rebaudioside A) taste sweeter than those with less glucose (e.g.stevioside)13. The bitter receptors of the tongue react to the aglycones, or to polyphenols in the case of stevia leaves usage.

Hande Karaköse Jacobs University Bremen 4

Phytochemical Characterization of Stevia rebaudiana

2.2. Pharmacology, Toxicology and Regulations A sugar substitute is a food additive that enhances the effect of sugar in taste. There are natural and synthetic sugar substitutes. Those that are not natural are, in general, called artificial sweeteners. Food additives must be approved by the FDA, which publishes a Generally Recognized as Safe (GRAS) list of additives. The majority of sugar substitutes approved for use are artificially-synthesized compounds. Sugar substitutes are used for reasons, including weight loss, dental care, diabetes and hypoglycemia. Sugar substitutes which are commonly used in foods are; aspartame, cyclamate, saccharin and sucralose. Starting with aspartame, it was produced from two amino acids: aspartic acid and phenylalanine. It is about 200 times sweeter than sucrose. The safety of aspartame has been studied extensively including animal studies, clinical and epidemiological research14. Hypotheses of adverse health effects have focused on the three metabolites of aspartame, which are aspartic acid, methanol and phenylalanine and further breakdown products including formic acid and formaldehyde15. Aspartame is rapidly hydrolyzed in the small intestines. Even with ingestion of very high doses of aspartame (over 200 mg/kg), no aspartame is found in the blood due to the rapid breakdown16. Furthermore, people with the genetic disorder phenylketonuria should avoid aspartame since they have a decreased ability to metabolize naturally occurring essential amino acid phenylalanine. The acceptable daily intake (ADI) value for aspartame is determined as 40 mg/kg of body weight 17.

Sucralose is a chlorinated sugar which is 600 times sweeter than sucrose. FDA approved usage of sucralose after reviewing 110 studies in humans and animals18. However, some adverse effects were observed at doses that significantly exceeded the estimated daily intake which is 1.1 mg/kg/day 19.

Saccharin was produced first in 1878 by a chemist working on coal tar derivatives. Studies in laboratory rats during the early 1970s linked saccharin with the development of bladder cancer in rodents. As a consequence, all food containing saccharin was labeled with a warning20. However, in 2000, the warning labels were removed because rodents, unlike humans, have a unique combination of high pH, high calcium phosphate, and high protein levels in their urine, which leads to formation of microcrystals that damages the bladder and over-produced cells to repair the damage leads to tumor formation. Since this does not occur in humans, the conclusion was there is no cancer risk21, 22. In the European Union, saccharin is also known by the E number (additive code) E954.

Hande Karaköse Jacobs University Bremen 5

Phytochemical Characterization of Stevia rebaudiana

Beginning in the 1970s with saccharin until the present day, artificial sweeteners have generated a lot of controversy. The chemical nature of many artificial sweeteners (Figure 2) do not present a good public image; e.g. saccharin was first produced from a coal-tar derivative23, aspartame breaks down into formaldehyde upon digestion15, it also presents a health hazard to people born with phenylketonuria, and sucralose is manufactured by the selective chlorination of sucrose24. With the scientific evidence on a particular sweetener, which is often inconclusive, and with many interests at stake, including the food additive approval process, and potential political and economic consequences, the results of these disagreements have not been entirely consistent or logical. Aspartame, for example, gained FDA approval over vocal opposition from certain public health advocates, while stevia extract, a substance which arguably presents health risks, cannot have FDA approval and avoids a complete ban only by classification as a “dietary supplement” rather than as a food additive25.

HO OH O Cl Cl HO O O NH O OCH3 O N HO O S H O OH NH2 O OH Cl O

aspartame sucralose saccharin

Figure 2. Structures of few artificial sweeteners.

Stevia was used extensively by the Guarani Indians for more than 1,500 years in Paraguay and Brazil26. Stevia was first used as a sweetener in Japan in the 1970s, and it was a natural substance that had been in use before 1958s with no apparent ill effects. However, the FDA banned stevia as unsafe food additive in 1991 after receiving an anonymous industry complaint, and restricted its import27. The stated reason of FDA was that toxicological information on stevia was inadequate to demonstrate its safety. Health controversies about stevia started with the study of Pezzuto in 198528, which reported that steviol, a breakdown product of stevioside and rebaudioside A is a mutagen in the presence of a liver extract of pre-treated rats. But this finding was criticized and stated that it might be worth exploring the possibility that the mutagenicity of steviol (as in the experiments of Pezzuto et

Hande Karaköse Jacobs University Bremen 6

Phytochemical Characterization of Stevia rebaudiana al.,1986)29 is due to an impurity and the very high dose used in the experiments30. Metabolism of steviol in rat liver is complex and some metabolites detected are shown in Figure 3. The major metabolite is 15-α-hydroxysteviol, which is non-mutagenic both in the presence and absence of a metabolic activating system. Other metabolites are 7-β-hydroxysteviol, 17-hydroxyisosteviol and ent-16-oxo-17-hydroxybeyeran-19-oic acid29, 31. The mutagenic substance was proposed to be 15-oxosteviol. But, this compound was not detected as a metabolite of steviol and it was reported to be bactericidal and weakly mutagenic30. Nevertheless, other bacterial mutagenic assays failed to demonstrate steviol mutagenic activity32. The nature of mutagenic metabolite thus remained in doubt. Stevia remained banned until 1994, when forced under the Dietary Supplement Health and Education Act, the FDA revised the decision on stevia and permitted it to be used as a dieatary supplement. Over the following years studies on the toxicology and adverse effects of stevia showed contradictory results.

OH OH CH2 CH2 H3C H3C

R1 O R2 HOOC CH3 HOOC CH3 R1=OH R2=H; 15α-hydroxysteviol 15-oxosteviol R1=H R2=OH; 7β-hydroxysteviol

OH O CH2OH CH2 O

H3C H3C

HOOC CH3 HOOC CH3

Steviol-16,17-oxide 17-hydroxyisosteviol

Figure 3.Structures of steviol metabolites.

Toskulkao et. al reported stevioside and steviol to have very low acute oral toxicity in the mouse, rat and hamster 33. Xili et al.34 have performed a combined chronic and carcinogenicity study, in Wistar rats and in this study however stevioside administration in the diet showed no carcinogenic effects in the rat. Through the review of many other toxicological studies 35-38 on

Hande Karaköse Jacobs University Bremen 7

Phytochemical Characterization of Stevia rebaudiana stevia and its compounds, the EFSA committee concluded in 1999 that steviol, one metabolite of stevioside, that is produced by the human microflora is genotoxic and induces developmental toxicity. Therefore, the European Commission in 1999 banned stevia and its products in foods in the European Union pending further research. The European Food Safety Authority reevaluated the safety of steviol glycosides and expressed its opinion on 10 March 2010. The Authority established an Acceptable Daily Intake (ADI) for steviol glycosides, expressed as steviol equivalents, of 4 mg/kg (BW/day). The European Commission allowed the usage of steviol glycosides as a food additive, establishing maximum content levels for different types of foods and beverages on 11 November 201139.

Regarding the effect of stevia in diabetes, a 2011 study by Misra et. al. on diabetes induced to rats by injection of alloxan, have shown that leaf extract of S. rebaudiana (200 and 400 mg/kg) produced a significant decrease in the blood glucose level, without producing condition of hypoglycemia after treatment 26. In addition, a 2009 review indicated that stevioside and related compounds have anti-hyperglycemic, anti-hypertensive, anti-inflammatory, anti-tumor, anti- diarrheal, diuretic, and immunomodulatory actions40. The effect of stevioside and steviol on glucose absorption was investigated by Toskulkao et al. and it was reported that 1mM steviol inhibits glucose absorbtion, whereas 5 mM has no inhibitory effect. The inhibition of glucose absorption by steviol was related to steviol concentration and incubation time 41. However, the announced acceptable daily intake of steviol glycosides (4 mg/kg BW/day) would yield a maximum plasma concentration of steviol of approx. 20 μM if stevioside is completely converted to steviol. This concentration is far below the reported value to inhibit intestinal glucose absorption. Therefore, more studies should be conducted using ADI amount to reevaluate the effect of steviol on glucose absorption. However, it is worth pointing out that stevioside does not interfere with glucose absorption40.

Stevioside, other related steviol glycosides, or stevia leaves themselves have been used commercially in many countries, especially in Asia, as food additives for sweetening a variety of products without any side effects. Moreover, phytochemicals (especially polyphenolics and steviol glycosides) of stevia were reported to exhibit significant pharmacological activities.

Hande Karaköse Jacobs University Bremen 8

Phytochemical Characterization of Stevia rebaudiana

2.3. Pharmacokinetics of Stevioside: Absorbtion, Metabolism, Excretion

Absorption and metabolism studies on steviol glycosides showed that the uptake of stevioside by the intestinal tract is extremely low due to its high molecular size and hydrophlicity40, 42, 43. Stevioside is not degraded by the enzymes of the intestinal tract, gastric juice or digestive enzymes from animals and humans 5, 42, 44. However, bacterial intestinal flora of humans is able to convert stevioside to steviol and Bacteroides sp. is responsible for this conversion in the lower gastrointestinal tract of both rat and human 5. Koyoma et. al44. investigated the metabolism of stevia by incubating stevioside, rebaudioside A and steviol with pooled human faecal homogenates obtained from healthy volunteers for 0.8 and 24 h under anaerobic conditions. Stevioside, rebaudioside A were completely hydrolysed in 24 h, and no degradation of steviol was observed. The author proposed a metabolic pathway for rebaudioside A, which suggests that majority of rebaudioside A is hydrolyzed via stevioside to steviol and minority via rebaudioside B to steviol. Steviol was not further metabolized in human intestinal microflora being inconsistent with the study of Pezzuto et.al28 reporting the oxidation of steviol to hydroxysteviol, or to 15-oxo-steviol (Figure 3). Another study in 10 healthy volunteers showed that after 3 days of consumption of stevioside (every day 3 times 250 mg capsules with 8 h intervals), steviol glucoronide is the only excretion product of stevioside in urine. There was no detection of free steviol in urine. Moreover, after enzymatic hydrolysis of urine by β-glucuronidase/sulfatase, steviol was the only aglycone and there was no indication of steviol sulfates 42. The excretion route proposed by Geuns et al. is presented in Figure 4.

OH OH O-Glc-Glc H3C H3C H3C liver H bacteria H H colon H H H H H3C C O H H C H C 3 CO2H HO OH 3 CO2Glc O O CO2H H H stevioside steviol steviol glucuronide

Figure 4.Hypothetical excretion route of stevioside42.

Hande Karaköse Jacobs University Bremen 9

Phytochemical Characterization of Stevia rebaudiana

Another study in humans reported that 72 h after oral stevioside ingestion, steviol glucuronide excretion in urine and free steviol in feces accounting for 62% and 5.2% of the total dose of stevioside administered respectively40, 45. As conclusion from the reviewed literature, steviol glucorunonide is the main metabolite of stevioside consumption and urinary excretion is responsible for the disposal from the body.

2.4. Biosynthesis of the Steviol Glycosides Biosynthesis of steviol glycosides are still subject of discussion. There are two main proposed pathways for steviol glycosides; 2-C-methyl-D-erythritol 4-phosphate/1-deoxy-D-xylulose 5- phosphate pathway (MEP/DOXP pathway) (Figure 5) and mevalonic acid pathway (MVA) (Figure 6).

MEP pathway of isoprenoid (terpenoids) biosynthesis is a metabolic pathway which leads to the formation of isopentenyl pyrophosphate (IPP) (9) and dimethylallyl pyrophosphate (DMAPP) (10) in the plastids of the plants. MEP pathway is an additional alternative pathway to mevalonic acid pathway (MVA) for formation of isoprenoids (terpenoids). MVA reactions take place in cytosol whereas MEP reactions occur in plastids. Pyruvate (1) and glyceraldehyde-3-phosphate (2) are converted by DOXP synthase to 1-deoxy-D-xylulose-5-phosphate (3) and by DOXP reductase to 2-C-methyl-D-erythritol 4-phosphate (4) (MEP). After subsequent reaction steps, the end products IPP (9) and DMAPP (10), which are precursors of terpenoids, are formed. Synthesis of all higher terpenoids occurs via formation of geranyl pyrophosphate (GPP) and geranylgeranyl pyrophosphate (11) (GGPP).

In the proposed MEP pathway, steviol was synthesized from kaurene (13) 46. The plant gene for the first step in the MEP pathway is deoxyxyulose-5-phosphate (DXP) synthase (DXS), which leads to the synthesis of DXP from pyruvate (1) and glyceraldehyde 3-phosphate (2). Once synthesized, DXP can either be used for the production of vitamins like thiamin or in the MEP pathway for isoprenoid synthesis.

The DXS amino acid sequence is highly conserved among plant species, which enabled Totte´ et al. (2003)47 to design primers for RT-PCR and clone the DXS gene from Stevia. The author suggested that steviol was synthesized via mevalonic acid pathway (MVA), involving mevalonic acid in the biosynthesis of steviol, but no direct proof was given to support it. Brandle et. al (2002)48 sequenced 5548 expressed sequence tags (ESTs) from stevia leaf cDNA library. The

Hande Karaköse Jacobs University Bremen 10

Phytochemical Characterization of Stevia rebaudiana

ESTs were classified according to their function in primary or secondary metabolism and many genes specific to MEP pathway but not the MVA pathway were identified, which concludes that the source of IPP for diterpenes is through the MEP pathway.

Steviol glycosides share four common steps in biosynthetic pathway with gibberellic acid formation. After oxidation of ent-kaurene at the C-19 position to ent-kaurenoic acid, the pathways to the steviol glycosides and the gibberellins diverge. Steviol is produced by hydroxylation of ent-kaurenoic acid at the C-13 position. Steviol is then glycosylated by series of UDP-glucosyltransferases (UGTs). UGTs are highly regiospecific and recognize particular substructure of the acceptor molecule rather than the molecule in its entirety49. The MEP pathway of steviol glycosides and the enzymes involved in this pathway is presented in Table 2 and Figure 5.

Table 2.Enzymes involved in biosynthesis of steviol glycosides

Enzyme abbreviation Enzyme

DXS deoxyxyulose-5-phosphate synthase

DXR deoxyxyulose-5-phosphate reductoisomerase

CMS 4-diphosphocytidyl-2-C-methyl-D-erythritol synthase

CMK 4-diphosphocytidyl-2-C-methyl-D-erythritol kinase

MCS 4-diphosphocytidyl-2-C-methyl-D-erythritol 2,4-cyclodiphosphate synthase

HDS 1-hydroxy-2-methyl-2(E)-butenyl 4-diphosphate synthase

HDR 1-hydroxy-2-methyl-2(E)-butenyl 4-diphosphate reductase

GGDPS geranylgeranyl diphosphate synthase

CPS copalyl diphosphate synthase

KS kaurene synthase

KO kaurene oxidase

KAH kaurenoic acid 13-hydroxylase

Hande Karaköse Jacobs University Bremen 11

Phytochemical Characterization of Stevia rebaudiana

O OH - DXS COO 2- O 2- + H OPO3 OPO3 O OH OH 1-deoxyxylulose-5-phosphate Pyruvate glyceraldehyde-3-phosphate NH 2 DXR N CMS OH OH O N O 2- OPO 2- O P2O5 O 3 OH OH OH OH 2-C OH OH 4-diphosphocytidyl-2-C-methyl-D-erythritol - PO2 O NH 2 O CMK - 2- N MCS PO2 PO3 O O OH OH O N O 2- 2-C-methyl-D-erythritol-2,4- O P2O5 O cyclodiphosphate OH OH OH OH HDS 4-diphosphocytidyl-2-C-methyl-D-erythritol-2-phosphate HDR 2- O P2O6 3- 3- OH OP2O6 + OP2O6 1-hydroxy-2methyl-(E)butenyl-4- isopentenyl diphosphate dimethylallyldiphosphate diphosphate GGDPS 3- OP2O6 CDPS 3- OP2O6 geranylgeranyl diphosphate (-)-copalyl diphosphate OH KS KAH KO H H H H COOH COOH H steviol (-)-kaurenoic acid H UGT85C2 (-)-kaurene O glc glc O glc O glc glc O glc glc glc

H H H H UGT H H H H COOH COOH COO glc COO glc steviolmonoside steviolbioside stevioside rebaudioside A Figure 5.Steviol glycoside biosynthesis via the MEP pathway50.

Hande Karaköse Jacobs University Bremen 12

Phytochemical Characterization of Stevia rebaudiana

HO HO O O O O O O AACT HMGS HMGR - SCoA SCoA O- SCoA O SCoA 4 1 2 3 MK

O O HO O HO O P P OH O O O OH P O O-O O - P P PMK O O - OH PPMD - O O O O O O- 7 5 IDI 6

O O O P P O OH O-O 8

Figure 6.Alternative MVA pathway. Enzymes of the MVA pathway are as follows: AACT, AcAc-CoA thiolase; HMGS, HMG-CoA synthase; HMGR, HMG-CoA reductase; MK, mevalonate kinase; PMK, phosphomevalonate kinase; PPMD, diphospho-mevalonate decarboxylase. 1, Ac-CoA; 2, AcAc-CoA; 3, HMG-CoA; 4, MVA; 5, mevalonate 5-phosphate; 6, mevalonate 5-diphosphate. Both MPE and MVA pathways lead to the formation of compound 8, dimethylallyl diphosphate; 7, isopentenyl diphosphate. The interconversion of IPP into DMAPP is catalyzed by IDI, isopentenyl diphosphate isomerase51.

Hande Karaköse Jacobs University Bremen 13

Phytochemical Characterization of Stevia rebaudiana

2.5. Analysis of Steviol Glycosides of Stevia rebaudiana

A wide range of analytical techniques have been used to determine the diterpenoid glycosides in stevia. These techniques include thin layer chromatography (TLC) 52, capillary electrophoresis (CE) 53, near infrared reflectance 54, and enzymatic methods 55. But the most used and most efficient method is high performance liquid chromatography (HPLC or LC). The use of the hyphenated technique coupled with mass spectrometry (LC-MS) in the analysis of plant extracts provides important advantages because of the combination of the separation capabilities of LC and the power of MS as an identification and confirmation method. In many modern HPLC separations, prepacked columns are used and many types are available from the manufacturers. However, it is possible to carry out most separations using silica column for non-polar compounds or reversed phase C18 bonded phase column for polar compounds. The solvent systems used in the analytical HPLC usually include gradient elutions using solvents of aqueous acetic, formic or phosphoric acids with methanol or acetonitrile as an organic modifier. The pH and ionic strength of the mobile phase are known to influence the retention of phenolics in the column depending on protonation, dissociation, or a partial dissociation. A change in pH which increases the ionization of a sample could reduce the retention in a reversed phase separation. Thus, small amounts of acetic (2– 5%), formic, phosphoric or trifluoroacetic acid (0.1%) are included in the solvent system to enhance ionization of phenolic and carboxylic groups and hence to improve peak shapes, resolution and reproducubility of chromatographic runs. However, for steviol glycosides chromatographic separations in HPLC are not so straight forward due to the structural similarity of steviol glycosides. Especially for isomer pairs of stevioside/rebaudioside B and rebeaudioside A/rebaudioside E with the same molecular formula, resulting in very close retention times in LC thus, resulting in selectivity problems due to peak overlap and irreproducibility. Therefore, it is still a challenge to achieve efficient separation and identification for steviol glycoside extracts.

Detection and separation of steviol glycosides on liquid chromatography (LC) were performed 3, 56-60 61, 62 63 employing amino (NH2) , C18 , and hydrophilic interaction chromatography (HILIC) , 64 columns in combination with mostly UV or MS detection.

Hande Karaköse Jacobs University Bremen 14

Phytochemical Characterization of Stevia rebaudiana

Among those, two-dimensional LC 65, 66 and ultra high pressure liquid chromatography (UHPLC) 64, 67 systems were used as well.

Liquid chromatography with amino columns provides good separation of steviol glycosides and have selectivity for the isomers of stevioside/rebaudioside B and rebaudioside A/rebaudioside E but they are having disadvantage of reproducibility and long equilibration times 57. C18 column exhibits longer retention time and more robustness if compared to amino columns but also poor selectivity for the separation of stevioside and rebaudioside A. Gradient elution is not enough to overcome this problem, thus two dimensional systems either with connection of two C18 columns 68 or C18 with amino column 66 were used.

Hydrophilic interaction chromatography is still new and a useful technique for the retention of more polar analytes with increased selectivity if compared to reversed phase chromatography. The interaction of the analytes is believed to be with the water rich layer forming on the surface of the polar stationary phase against the water poor mobile phase. HILIC can offer a tenfold increase in sensitivity over reversed-phase chromatography for detection of polar compounds with mass spectrometry, due to more volatile organic solvent 69. Some papers 63, 64 describe the use of HILIC column for stevia extract. In those studies, steviol glycosides were separated with isocratic elution using 5–20% water in acetonitrile with buffer or formic acid and the robustness of the separation against changes of buffer concentration and percentage of water differ 64. Methods that use UV detection for steviol glycoside quantification are most popular; however suffer from a series of disadvantages. Detection is typically carried out at 200-210 nm using the carboxylic acid and olefinic chromophores. These wavelengths are very close to the UV cutoff of acetonitrile as a solvent and particularly problematic if gradient elution is used. Additionally many further stevia constituents from the matrix and other impurities absorb at these wavelenghths. On no occasion was the absence or presence of co-eluting impurities established in any published steviol glycoside UV method.

Despite some efforts in the development of methods aimed at the identification and quantification of steviol glycosides until today no validated and certified method exists. Furthermore no interlaboratory trials were ever conducted on steviol glycoside quantification allowing a reliable assessment of method validity. Accordingly despite many contributions

Hande Karaköse Jacobs University Bremen 15

Phytochemical Characterization of Stevia rebaudiana published in the field of steviol glycoside analysis, there is still urgent need for critical assessment of published methods and the development of a generally accepted standard method.

2.6. Phenolic Acids

Polyphenols are secondary metabolites that constitute one of the most widespread groups of compounds in plants. They are derivatives of the pentose phosphate, shikimate and phenylpropanoid pathways in plants70. Polyphenolic compounds contribute to pigmentation of flowers, fruits, leaves or seeds and play important role in the growth, reproduction and adaptative strageties of plants71. In food, phenolics contribute to the bitterness, astringency, color, flavor, odor, and oxidative stability of products. Moreover, health-protecting capacity and antinutritional properties of plant phenolics are of great importance to producers, processors and consumers72. The antioxidant activity of the dietary polyphenolics is considered to be much greater than that of the essential vitamins, therefore contributing significantly to the health benefits of fruits73 Phenolic compounds are present in almost all foods of plant origin. Fruits, vegetables, and beverages are the main sources for these compounds in the human diet. The level of phenolics in plant sources depend on cultivation techniques, cultivar, growing conditions, ripening process, as well as processing and storage conditions. In addition, the content of some phenolics may increase under stress conditions such as UV radiation, infection by pathogens and parasites, wounding, air polution and exposure to extreme temperatures74. Fruit and beverages such as coffee, tea and red wine constitute the main sources of polyphenols. Certain polyphenols such as quercetin are found in all plant products (fruit, vegetables, cereals, leguminous plants, fruit juices, tea, wine, infusions, etc), whereas others are specific to particular foods (flavanones in citrus fruit, isoflavones in soya, phloridzin in apples). In most cases, foods contain complex mixtures of polyphenols, which are often poorly characterized 75. Onions are rich sources of flavonoids76. Flavonols, the predominant phenolics, are located mostly in the tomato skin. Cherry tomatoes contained a much higher level of flavonols than larger size tomato cultivars76, 77. Anthocyanins are located in the red onion skin and the outer fleshy layer78. The main group of polyphenols includes simple phenols, phenolic acids (benzoic and cinnamic acid derivatives), coumarins, flavonoids, stilbenes, hydrolyzable and condensed tannins, lignans, and lignins.

Hande Karaköse Jacobs University Bremen 16

Phytochemical Characterization of Stevia rebaudiana

Phenolic acids consist of two subgroups; the hydroxybenzoic and hydroxycinnamic acids. Hydroxybenzoic acids include gallic, p-hydroxybenzoic, protocatechuic, vanillic and syringic acids, which in common have the C6–C1 structure. Hydroxycinnamic acids, on the other hand, are aromatic compounds with a three-carbon side chain (C6–C3), with caffeic, ferulic, p-coumaric and sinapic acids (Figure 7) being the most common 71, 79.

OH COOH

R1 R2 HO OH COOH COOH OH R1

hydroxybenzoic acid protocatechuic acid hydroxycinnamic acid

HO COOH COOH H3CO COOH

HO HO HO

caffeic acid p-coumaric acid ferulic acid

Figure 7.Examples for hydroxybenzoic and hydorxycinnamic acids.

Caffeic acid is the major representative of hydroxycinnamic acids and occurs in foods mainly as chlorogenic acid (5-caffeoylquinic acid). Chlorogenic acids (CGAs) are a family of esters formed between one or more residues of certain trans-cinnamic acids and quinic acid (1L-1 (OH),3,4/5-tetrahydroxycyclohexane carboxylic acid) which have axial hydroxyls on carbons 1 and 3 and equatorial hydroxyls on carbons 4 and 5. During processing, trans isomers may be partially converted to cis isomers 80, 81(Figure 8). The main classes of CGAs found in nature are the caffeoylquinic acids (CQA), dicaffeoylquinic acids (diCQA), and, less commonly, feruloylquinic acids (FQAs), each group with at least three isomers82. CGAs are antioxidant components produced by plants in response to environmental stress conditions such as infections by microbial pathogens, mechanical wounding, and excessive UV or visible light levels83 Chlorogenic acids make up 5-10% of the weight of coffee beans and plays a significant role in coffee color and aroma formation84.

Hande Karaköse Jacobs University Bremen 17

Phytochemical Characterization of Stevia rebaudiana

OH OH O O O 6 OR5 OH C 1 C OR4 O HO HO O O OH OR1 OR3 OH OH OH C O OH HO OH OH quinic acid trans -4-caffeoylquinic acid cis-4-caffeoylquinic acid

Figure 8.General structure of quinic acid and one of chlorogenic acids as an example.

In addition to being found in coffee, these compounds are also found at significant levels in plant foods such as apples, pears, tomato, potato, and eggplant85. Coffee is a major source of chlorogenic acid in the human diet; daily intake in coffee drinkers is 0.5–1 g; coffee abstainers will usually ingest < 100 mg/d 86. In the last few years, CGAs has been the subject of several investigations in their potentially beneficial effects in humans involving their antioxidant activity, among other beneficial effects. Several pharmacological activities of CGAs including antioxidant activity, the ability to increase hepatic glucose utilization,87-94 inhibition of the HIV-1 integrase,95-97 antispasmodic activity,98 and inhibition of the mutagenicity of carcinogenic compounds99 have been revealed by in vitro, in vivo, and human intervention studies so far. CGAs and their metabolites display additional highly favorable pharmacokinetic properties.100-102

Flavonoids are low molecular weight compounds, consisting of fifteen carbon atoms, arranged in a C6–C3–C6 configuration. Essentially the structure consists of two aromatic rings, A and B, joined by a 3-carbon bridge, usually in the form of a heterocyclic ring, C. The aromatic ring A is formed via glucose metabolism with condensation of malonyl-coenyme A (CoA) catalyzed by chalcone synthetase, while ring B and C is derived from phenylalanine through the shikimate pathway, which is converted to cinnamic acid and to coumaric acid. Coumaric acid CoA and three malonyl CoAs are condensed in a single enzymatic step to form naringenin chalcone. The C-ring closes and becomes hydrated to form 3-hydroxyflavonoids (e.g., catechins), 3,4-diol flavonoids (e.g., quercetin), and procyanidins103. Variations in the heterocyclic ring C give rise to the major flavonoid classes, i.e., flavonols, flavones, flavanones, flavanols (or catechins), isoflavones, flavanonols, and anthocyanidins

Hande Karaköse Jacobs University Bremen 18

Phytochemical Characterization of Stevia rebaudiana

(Figure 9), while individual compounds within a class differ in the pattern of substitution of the A and B rings. These substitutions may include oxygenation, alkylation, glycosylation, acylation, and sulfation 79, 104.

3' 2' 4' 8 B O 5' 7 2 1' A C 6' 6 3 5 4

Generic structure

O O O

OH O O O

Flavone Flavonol Flavanone

O O+

OH OH

Flavanol Antocyanidin

Figure 9.Generic structure of major classes of flavonoids.

Within different subclasses of flavonoids, further differentiation is based on the number, position and nature of substituent groups attached on the rings. Mostly they are sugars, such as glucose, galactose, rhamnose, arabinose, xylose and rutinose. Flavonoid glycosides have many isomers with the same molecular weight but different aglycone and sugar component at different positions attaching on the aglycone ring 72, 105, 106. Flavonoid glycosides as well are commonly encountered in plant material and following ingestions these glycosides are hydrolysed by the human microbial gut flora into their aglycones, which are subsequently absorbed and show significant bioavailability.

Hande Karaköse Jacobs University Bremen 19

Phytochemical Characterization of Stevia rebaudiana

Flavonoids play different roles in the ecology of plants. Due to their attractive colors, flavones, flavonols, and anthocyanidins may act as visual signals for pollinating insects. Because of their astringency, catechins and other flavanols can represent a defense system against harm of insects to the plant. Furthermore flavonoids protect plants from UV radiation of sun with their favorable UV-absorbing properties104.

Apart from their roles in plants, flavonoids play important role in human diet. Flavonoids are important antioxidants (hydrogen-donating radical scavengers) due to their high redox potential, which allows them to act as reducing agents, hydrogen donors, and singlet oxygen quenchers. The antioxidant property of flavonoids may protect tissues against oxygen free radicals and lipid peroxidation. Thus, flavonoids might contribute to the prevention of atherosclerosis, cancer and chronic inflammation107. In addition, they have a metal chelating potential, which play an important role in oxygen metabolism and are essential for many physiological functions 108. The proposed binding sites for trace metals to flavonoids are the catechol moiety in ring B, the 3- hydroxyl, 4-oxo groups in the heterocyclic ring, and the 4-oxo, 5-hydroxyl groups between the heterocyclic and the A rings. However, the major contribution to metal chelation is due to the catechol moiety, as exemplified by the more pronounced bathochromic shift produced by chelation of copper to quercetin compared to that of kaempferol (similar in structure to quercetin except that it lacks the catechol group in the B ring)104. Flavonoids and phenolic acids have protective role in carcinogenesis, inflammation, atherosclerosis, thrombosis and have high antioxidant capacity. Furthermore, flavonoids have been reported as aldose reductase inhibitors blocking the sorbitol pathway that is linked to many problems associated with diabetes106. Phenolic acids in stevia were analyzed by HPLC on a C18 column by Kim et. al 109 and the main phenolic compounds found were pyrogallol, 4-methoxybenzoic acid, p-coumaric acid, 4- methylcatechol, sinapic and cinnamic acid. The flavonoids detected in stevia leaves belong to the subgroups of flavonols and flavones. They were identified using two-dimensional UHPLC-DAD and LC-MS/MS and spectroscopic methods (1H and 13C NMR, IR, and 2D NMR) 110, 111.

There is currently great interest in phenolic acids research due to the possibility of improved public health through diet, where preventative health care can be promoted through the consumption of fruit and vegetables. Therefore, the presence of such compounds with proven

Hande Karaköse Jacobs University Bremen 20

Phytochemical Characterization of Stevia rebaudiana health benefits in stevia extracts would affect the the organoleptic properties of stevia based products and would add health aspects to the use of stevia as sweenetening agent.

2.7. Proteomics of Stevia rebaudiana Determination of amino acid content and potential allergenic proteins in food is very significant and necessary research in the food formulation processes. Currently, there has been no published data related to stevia allergens and there is only one published paper for the protein analysis in stevia112. Therefore, screening and quantitative analysis of stevia proteins and any potential allergen is crucial. Detailed and comprehensive characterization of plant-derived food allergens can be carried out using proteomics. In proteomics, after the separation and purification of the protein, proteins are identified by mass spectrometry. Proteomic technologies using 2D-PAGE and immunoblotting are then applied in the identification of new allergens113. In the past, protein determination was carried out by mRNA analysis, but later it was found that there was no correlation with protein content as gene expression is regulated post- transcriptionally and translation from mRNA cause differences 114, 115. Most proteins are chemically modified through post-translational modifications, mainly through the addition of carbohydrate and phosphate groups. Such modifications play an important role in modulating the function of many proteins. The most common post-translational modifications include glycosylation, phosphorylation, ubiquitination, methylation, acetylation, and lipidation115. The major methods to study proteins include, high quality separation of proteins in two dimensions (2D-SDS PAGE), characterization of separated proteins by mass spectrometry and information collection using bioinformatic tools and databases115. Matrix-assisted laser desorption/ionization (MALDI) and electrospray ionization mass spectrometry (ESI) are widely used techniques for proteomic studies. Sample preparation is the most important step in the analysis of proteins from plants due to the low protein content relative to other systems and the large quantities of polysaccharides, lipids, phenolics and other secondary metabolites. Pretreatment of samples for 2D electrophoresis involves solubilization, denaturation and reduction to completely break up the interactions between the proteins and removal of all interfering compounds (phenolic compounds, nucleic acids) to ensure efficient separation115. The most common protein extraction protocol is based on precipitating proteins from homogenized tissue or cells with trichloracetic acid (TCA) in acetone. An alternative protocol is

Hande Karaköse Jacobs University Bremen 21

Phytochemical Characterization of Stevia rebaudiana based on the solubilization of proteins in phenol, followed by their precipitation with ammonium acetate in methanol. No one protocol is necessarily more appropriate than the other; studies have suggested that they are complementary116. In order to evaluate the effectiveness of a given extraction protocol, protein quantification is needed. Bradford, Lowry and BCA assays are the most common colorimetric methods. Identification of proteins proceeds with separation of proteins from the extract with 2D-SDS PAGE and subsequent digestion of the individual separated proteins or digestion of the entire protein mixture followed by separation of the resulted peptides117. 2D SDS PAGE separates proteins in two dimensions; the first dimension in a pH gradient according to their isoelectric point (pI), and in the second dimension, the proteins is separated according to their molecular weight. The first dimension of electrophoresis involves denaturing isoelectric focusing using immobilized pH gradient gels (IPG). IPG strips have a gradient of charge imbedded in acrylamide. IPG strips improve the reproducibility and reliability and overcome pH gradient instability. Strips come in a variety of pH ranges and lengths (from 7 to 24 cm). Samples can be applied on the strips by cup loading or by in-gel rehydration. In cup loading method, the strips are pre-rehydrated with rehydration buffer and the samples are applied into the loading cup at either acidic or basic end. In in-gel rehydration, the sample in lysis buffer is diluted with the rehydration buffer. The IPG matrix absorbs the proteins. Isoelectric focusing (IEF) is carried out on a first dimension electrophoresis unit consisting of five phases of stepped voltage from 500 to 3500 V (Multiphor) or 500 to 8000 V (IPGPhor)115. After completion of the first dimension the proteins are separated according to their mass in second dimension using SDS-PAGE. Separation of proteins in second dimension is based on differences in their electrophoretic mobility due to differences in their size. SDS is a very effective solubilising agent for a wide range of proteins. The majority of proteins bind SDS at a ratio of 1.4 g SDS / 1 g protein to form negatively charged complexes118. Proteins are transferred electrophoretically from the IEF strip into a narrow starting zone prior to entering the main separating gel. This concentrates the proteins and results in very sharp bands or spots. Once the protein samples have entered the separating gel, the negatively charged protein-SDS complexes continue to move towards the anode. As they pass through the separating gel the proteins are resolved on the basis of their size because of the molecular sieving properties of the gel. After 2D electrophoresis, the protein spots can be

Hande Karaköse Jacobs University Bremen 22

Phytochemical Characterization of Stevia rebaudiana visualized by staining with either coomassie brilliant blue stain, which will detect proteins present in amount greater than 100 ng or with silver staining for amounts in ng range 119. The end point of any proteomics expression is to identify and characterize the proteins. Edman degradation was the standard method for protein sequencing for the last 25 years120. Other traditional approaches for protein identification include the use of antibodies to perform Western blots. However, this method has restricted use due to antibodies non-specific binding and the availability of antibodies to all proteins121. Development in mass analysis techniques for mass spectrometry (MS) and the ability to correlate MS data of proteins to sequences in databases have opened up new possibilities in protein sequencing. Protein identification via MS can be carried out in the form of whole-protein analysis (‘top- down’ approach) or analysis of enzymatically or chemically produced peptides (‘bottom-up’ approach). To date, one of the most common methods of identifying proteins is through peptide- mass fingerprinting (PMF). The proteins are digested with a proteolytic enzyme such as trypsin, to produce a set of tryptic fragments unique to each protein115. Summary of other MS-based proteomic strategies is presented in Figure 10.

Figure 10.Strategies for MS-based protein identification122.

Hande Karaköse Jacobs University Bremen 23

Phytochemical Characterization of Stevia rebaudiana

Mass spectrometry analysis of peptides and proteins relies exclusively on soft ionization techniques that create intact gas-phase ions from biomolecules. The creation of intact molecular ions enables accurate measurement of molecular weight. The electrospray ionization (ESI) and matrix-assisted laser desorption/ionization (MALDI) techniques are widely used in proteomic studies. The most common mass analyzers used with MALDI are time-of-flight (TOF) mass spectrometers. MALDI-TOF allows the analysis of high molecular weight compounds with high sensitivity and, soft ionization with little or no fragmentation. MALDI uses a solid matrix to co-crystallize with peptides/proteins on a sample plate and a laser light as its ionizing beam. Ionization occurs when these matrix molecules absorb the energy provided by a laser (usually 337 nm). Release of the energy causes a rapid thermal expansion of matrix and analyte into the gas phase. Proton transfer from analyte to matrix may result in charge reduction to the singly charged ion observed in the gas-phase123 The matrix is typically a small energy absorbing molecule such as α-cyano-4- hydroxycinnamic acid (HCCA) or 2,5,-dihydroxybenzoic acid (2,5-DHB). The molecular weight values of the trypsinized peptides or intact proteins obtained by MALDI-TOF are then used to identify the predicated proteins using web-based search engines such as MASCOT.

In cases where the protein is not present in the database, the proteins may be analyzed by tandem mass spectrometry (ESI-MS). The fragment ions observed in MS/MS spectrum is analyzed to derive the order of the amino acids in the tryptic peptides or in the intact protein. This method is known as de novo sequencing. If the fragment ion carries its charge on the N-terminal, the ion is categorized as a, b or c. If the charge is on the C-terminal of the fragment the type of the ion can be x, y or z (Figure 11). The difference in the mass between adjacent y- or b-ions corresponds to that of an amino acid. This can be used to identify the amino acid and, hence the peptide 115 sequence .

Hande Karaköse Jacobs University Bremen 24

Phytochemical Characterization of Stevia rebaudiana

+ H Y3 Y2 Y H R1 O R2 O R3 O 1 R4 O H N C C N C C N C C N C C OH

b H H b1 H H 2 H H b3 H

a1 a a 2 3

x3 z 3 x1 x2 z2 z1

a a3 1 a2 c1 c2 c3

Figure 11.Peptide fragmentation momenclature.

2.8. Lipid Analysis

Lipids play an important role in physiology and pathophysiology of living systems. All plant cells produce fatty acids from acetyl-CoA by a common pathway localized in plastids124. Fatty acyls (FAs) are group of molecules synthesized by chain-elongation of an acetyl-CoA primer with malonyl-CoA or methylmalonyl-CoA groups. Structures with a glycerol group are represented by two categories: the glycerolipids (GLs), composed mainly of mono-, di- and tri- substituted glycerols, and the glycerophospholipids (GPs), which are defined by the presence of a phosphate (or phosphonate) group esterified to one of the glycerol hydroxyl groups. Other compounds including fatty chains [e.g., fatty acids, fatty alcohols and inter-fatty esters (waxes)] are also considered in this category. The sterol lipids (STs) and prenol lipids (PRs) share a common biosynthetic pathway via the polymerization of dimethylallyl pyrophosphate/isopentenyl pyrophosphate but have obvious differences in terms of their eventual

Hande Karaköse Jacobs University Bremen 25

Phytochemical Characterization of Stevia rebaudiana structure and function. Another well-defined category comprises sphingolipids (SPs), which contain a long-chain base as their core structure. Saccharolipids (SLs) comprise lipids in which fatty acyl groups are linked directly to a sugar backbone. The final category comprises polyketides (PKs), which are a diverse group of metabolites from plant and microbial sources125 (Figure 12). Although they are different in their chemical composition, they all share one characteristic, which is solubilization in non-polar solvents, such as chloroform and hexane.

O O O OH OH O H

O Fatty acyls: hexadecanoic acid Glycerolipids: 1-hexadecanoyl-2-(9Z-octadecenoyl)-sn-glycerol O O H OH P O O O N+ OH OH O H NH H

O O Glycerophospholipids: 1-hexadecanoyl-2-(9Z-octadecenoyl)-sn-glycero-3-phosphocholine Sphingolipids: N-(tetradecanoyl)-sphing-4-enine

H

H H

H H OH HO

Sterol lipids:cholest-5-en-3-b-ol Prenol lipids: 2E,6E-farnesol O O OH NH O HO O O O O O N O H O HN P O P O O O O HO HO HO O O O H

OHOH

Polyketides: aflatoxin B1

Saccharolipids:UDP-3-O-(3R-hydroxy-tetradecanoyl)-a-D-N-acetylglucosamine

Figure 12.Examples of lipid categories.

Lipids are one of the most important metabolites of the organism. Essential fatty acids and fat- soluble vitamins, which are required by organism, can be supplied from lipids. Terpenes and steroids like vitamin A, D, E, K, cholic acid, and steroid hormones are related with nutrition,

Hande Karaköse Jacobs University Bremen 26

Phytochemical Characterization of Stevia rebaudiana metabolism, and regulation function. Lipids on the surface of the organisms protect its surface from mechanical damages and heat losing, and for the cells, it is closely related to cell recognition 126. Apart from their biological functions, fatty acids, triacylglycerols and phospholipids are the primary classes of lipids of interest in foods.

In general, lipid analysis involves three basic steps which are; 1) extraction of lipids from the sample, 2) analytical separation, 3) identification and quantification of lipids.

Lipids are mainly extracted from cells, plasma, and tissues. The components obtained depend on the method of extraction used, especially the solvent. All lipids have a polar head and a nonpolar tail. Therefore, mixture of chloroform and methanol in a two-step extraction was chosen to obtain better dissolution of lipids. This approach was developed in 1950s by Folch127. Lipids of all major classes could be recovered via chloroform/methanol extraction, typically according to the Folch, or Lees, and Sloane Stanley or Bligh and Dyer protocols, in which they are mostly enriched in the chloroform phase 128. The most widely used method for the extraction of solid samples is Soxhlet extraction. Purified lipid extracts are susceptible to oxidation. They should be dissolved in a non-polar solvent (e.g., hexane or chloroform) and stored at −20°C in a glass container in a nitrogen atmosphere. They can be stored in refrigeration temperature (0–4°C) for short-term 125. Several separation techniques have been used for the determination of lipids. Long-chain fatty acids have been determined by gas chromatography (GC) or liquid chromatography (LC). Supercritical fluid chromatography (SFC) 129 and thin layer chromatography (TLC) 130 was also utilized for lipid separation. Traditionally, lipids have been analyzed using gas chromatographic (GC) separation with flame- ionization detection (FID) or mass spectrometry (MS) detection. Compounds must be thermally stable with high vapor pressure to be volatilized during the injection in to the GC. Therefore lipids have to be converted into derivatives with lower boiling points, such as alcoholic esters. Lipids can be analyzed after hydrolysis, derivatization, or pyrolysis with GC technique. Fatty acid methyl ester derivatization is the most common method used for analysis with GC technique. Transesterification is one mechanism that can be employed to form FAMEs from fatty-acid esters in foods. Alkali- or acid-catalyzed transesterification procedures can be used to form FAMEs in a methanolic medium. The separation of FAMEs is usually achieved on highly polar liquid phases and the analytes are separated according to their chain length and degree of

Hande Karaköse Jacobs University Bremen 27

Phytochemical Characterization of Stevia rebaudiana saturation. The identification of the fatty acids in the sample can be achieved with comparison of the retention times of FAME reference compounds. 125. Although the derivatization solves the problem of sample volatility, problems with ester formation may include the incomplete conversion of fatty acids to FAME, loses of highly volatile short chain fatty acids or formation of contaminations which can overlap with the FAME peak in the GC chromatogram131. The wide choice of mobile and stationary phase makes selectivity extremely powerful in HPLC. Depending on the mobile phases and stationary phases used, there are two modes in LC; which are normal phase (NP) and reversed phase (RP). RP-LC methods have been developed utilizing both aqueous and non-aqueous solutions for lipid analysis, whereas in NP-LC, non-polar solvents are used for separation. RP-LC separates lipids according to their fatty acyl composition and in NP-LC separation occurs on the basis of their class of compounds125. Lipids with molecular weights of 100 – 2000 Da can be detected by LC-MS126. Electrospray ionization-mass spectrometry (ESI-MS) has been successfully applied to lipid analysis. Especially, the combination of chromatographic techniques with MS provides the technical support for the analysis of lipids and accelerates the emergence of the lipidomics. Lipidomics, focuses on the global analysis of lipids and their metabolites. The concept lipidomics was raised by Han et al. in 2003126, 132. Strategies currently used in lipidomics include direct-infusion ESI–MS and ESI–MS/MS, LC coupled with ESI–MS or MS/MS, and MALDI combined with Fourier transform ion cyclotron resonance MS (MALDI–FTICR–MS) or time-of-flight–MS (MALDI–TOF–MS). In addition, for some classes of lipids, LC coupled with APCI–MS was also used133. The analytes can be directly injected to MS without prior separation and soft ionization. ESI-MS has the advantage that the structural identification of lipids is more straightforward using different MS/MS experiments such as precursor ion scan, product ion scan and neutral loss scan. However, ion suppression can be the major complication in the direct injection MS experiments. MALDI-MS is a laser-based, soft-ionization method that is often used for analysis of large proteins, but has also been used successfully with lipids. Important advantages of MALDI–MS in lipid analysis are the speed of analysis and simplicity of operation: the analytes are ionized under relative soft conditions by laser desorption using an ultraviolet-absorbing matrix. One important disadvantage of MALDI is the presence of a lot of background in the lower mass range due to the matrix molecules. In addition, MALDI–MS is generally less quantitative compared to ESI–MS technique133.

Hande Karaköse Jacobs University Bremen 28

Phytochemical Characterization of Stevia rebaudiana

Currently, a wide range of analytical techniques are used for analyzing the lipids, however none of them provides a global lipid profiling. Therefore, the more different techniques we use, the wider aspect we can have about our sample.

Hande Karaköse Jacobs University Bremen 29

Phytochemical Characterization of Stevia rebaudiana

3. RESEARCH OBJECTIVE

The project aims to carry out the chemical profiling of plant called Stevia rebaudiana. Stevia produces, as its main secondary metabolite diterpene glycosides (steviol glycosides), which are natural sweeteners. As a sweetener it has two advantages. Firstly, as a terpene it does not cause allergic reactions unlike most peptide based sweeteners. Secondly, stevia grows almost on any soils, in particular on fields where tobacco used to be grown. Due to pH of soil become acidic after tobacco cultivation, very few plants have this specialty and the stevia cultivation in European Union would offer the tobacco farmers an alternative crop. Most importantly, the advantages and health benefits of natural sweeteners from stevia make it promising as a novel food in near future.

Within the EU project (DIVAS) Stevia rebaudiana was cultivated in variety of locations in the Mediterranean region of Europe over a period of two years using seven different botanical varieties of stevia. The leaves were harvested between two and three times annually producing a total of 166 different samples of Stevia rebaudiana leaves. The objective of the project was to provide scientific basis for a future specification of stevia as novel food in Europe. For this purpose, chemical specification of stevia leaves has been carried out. Methods have been developed and used to study the major classes of secondary metabolites. Analysis of the chemical composition of stevia leaves will allow definition of upper and lower limits of all relevant stevia plant constituents that are appropriate to chemical analysis including proteins, lipids and secondary plant metabolites comprising polyphenols and terpene glycosides.

As stevia may be cultivated within various regions of EU with different soil and climatic conditions it is important to know whether an EU-common specification will be achieved and how stevia leaves from regions outside EU can be distinguished on a scientific basis. For this purpose, dataset obtained from stevia leaves from various parts of Europe were analyzed statistically.

Hande Karaköse Jacobs University Bremen 30

Phytochemical Characterization of Stevia rebaudiana

4. STEVIOL GLYCOSIDES ANALYSIS BY LC-MS 4.1. Overview

Steviol glycosides extracted from Stevia rebaudiana leaves were subjected to LC- tandem MS and LC-TOF MS for characterization. The separation of steviol glycosides were compared using C18 reverse phase column and hydrophilic interaction chromatography (HILIC) column. Additionally tandem MS data of steviol glycosides is presented using an ion trap instrument, taking advantage of low energy collision induced fragmentation and multi stage fragmentation up to MS4 to identify steviol glycosides. LC-TOF method using HILIC column was validated and used for quantification of steviol glycosides in 166 stevia leaves extracts harvested in Europe.

4.2.Materials & Methods 4.2.1. Extraction Method

Extraction of steviol glycosides, as well as phenolic acids, lipids and volatile terpenes were achieved using soxhlet extraction system. Soxhlet conditions were optimised with respect to solvent volume, extraction cycles and time. Prior to the steviol glycoside extraction a chloroform extraction was carried out to remove the lipid fraction. In addition to the steviol glycosides the methanolic fraction contained phenolics in quantitative amounts used for quantification.

Extraction of Steviol glycosides and Phenolic acids: Two grams of S. rebaudiana leaves were immersed in liquid nitrogen, ground in a hammer mill, and extracted first with 150 mL of chloroform in a Soxhlet apparatus (Buchi B-811 extraction system) for 2 h and then with 150 mL of methanol for another 2 h. Solvents were removed from the methanolic extract in vacuo, and extracts were stored at - 20 oC until required.

4.2.2. LC-MS Analysis of Steviol glycosides

Compound identification was carried out using high resolution mass spectrometry (HR-MS) and tandem MS using an ion trap mass spectrometer. A HR-MS using an ESI-TOF-MS experiment allowed determination of molecular formulae based on the accurate mass measurements. Molecular formulas were in general accepted if an error below 5 ppm was experimentally observed, as accepted by all peer reviewed chemistry journals.

Hande Karaköse Jacobs University Bremen 31

Phytochemical Characterization of Stevia rebaudiana

LC-TOF MS: The LC equipment (Agilent 1100 series, Bremen, Germany) comprised a binary pump, an autosampler with a 100 μL loop, and a diode array detector with a light-pipe flow cell (recording at 210 nm and scanning from 200 to 600 nm). This was interfaced with a MicroTOF Focus mass spectrometer (Bruker Daltonics) fitted with an ESI source. The MS parameters were: nebulizer 1.6 bar, dry gas 12.0 L/min, dry temperature 220 0C. The MicroTOF was operated in negative ion mode and the mass range was 150 – 1200 m/z. Internal calibration was achieved with 10 mL of 0.1 mol/L sodium formate solution injected through a six-port valve prior to each chromatographic run. Calibration was carried out using the enhanced quadratic calibration mode.

LC-MSn (tandem MS): The LC equipment (Agilent 1100 series, Bremen, Germany) comprised a binary pump, an autosampler with a 100 μL loop, and a diode array detector with a light-pipe flow cell (recording at 210 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 Auto-MSn mode to obtain fragment ions 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 V, starting at 30% and ending at 200%. MS operating conditions (negative mode) were capillary temperature of 365 oC, a dry gas flow rate of 10 L/min, and a nebulizer pressure of 50 psi.

4.2.3. HPLC conditions

HILIC conditions: Separation was achieved on a 4,6 x 150 mm Dionex Acclaim Mixed Mode Wax-1 column with 5 μm particle size. Solvent A was 10 mM ammonium formate buffer at pH 3 and solvent B was acetonitrile (ACN). Solvents were delivered at a total flow rate of 0.5 mL/min and the column temperature was set to 40 oC. 5 μL of samples in 80% ACN/water were injected in to LC-MS system, unless stated otherwise. The isocratic profile was 85 %ACN and 15% water (10 mM ammonium formate buffer).

Reverse phase (C18) conditions: Separation was on a 250 x 3 mm C18 column (Varian Pursuit XRS) with 5 μm particle size. Solvent A was water/formic acid (1000+0.005 v/v), and solvent B was acetonitrile (ACN). Solvents were delivered at a total flow rate of 0.5 mL/min and the column temperature was set to 25 oC. 5 μL of samples were injected in to LC-MS system,

Hande Karaköse Jacobs University Bremen 32

Phytochemical Characterization of Stevia rebaudiana unless stated otherwise. The gradient profile was 10 to 80% B in 60 min and a return to 10% B at 65 min and 5 min isocratic to re-equilibrate.

4.2.4. Calibration Curve of Steviol Glycoside Standards

Stock solutions of the standard compounds of stevioside, rebaudioside A and steviolbioside were prepared in 80% ACN/water. A series of standard solutions was injected (5μL) into the LC-MS system. The areas of the peaks of each standard from extracted ion chromatograms (EIC) were used to make the respective standard curves.

4.2.5. Method Validation

Selectivity of the method was determined by comparing the chromatograms of leaf extract and reference compounds. Precision was determined by intra and inter-day measurements with three different concentration of standard solution of stevioside and rebaudioside A on the HILIC column and evaluated by the relative standard deviation (%RSD). Accuracy of the method was determined by spiking two different stevia leaf extracts with three different amounts of stevioside and rebaudioside A, separately and RSD was calculated. Quantification was achieved by applying calibration curve equations obtained by the least square method.

4.2.6. Solid Phase Extraction (SPE) of Steviol glycosides Extraction of stevia leaves: 0.8 g of stevia pulverized leaves were sonicated and heated with 30 mL of ACN/water (70:30 v/v) for 15 minutes. Then, the extract was filtrated through a 0.45 μm filter.

SPE Material I Cartridges were filled with the steviaclean stationary phase specially produced for Stevia rebaudiana (Knauer GmbH) in the amounts of 0.2 g, 0.4 g and 0.6 g. Each was condinitioned with water (1mL) and 3 mL of ACN/water (90:10 v/v). 1mL of stevia extract was loaded on the cartridge. The steviol glycosides were eluted with 2 mL of ACN/water (90:10 v/v). The eluate was filtered and subjected to HPLC analysis with amino column (Knauer, Eurospher

100 NH2, 5 μm, 150 x 3mm). 5 μL of samples were injected in to LC-MS system. Solvents were delivered at a total flow rate of 1.0 mL/min and the column temperature was set to 35 oC. UV detection was at 210 nm. 134

Hande Karaköse Jacobs University Bremen 33

Phytochemical Characterization of Stevia rebaudiana

SPE Material II Bond Elut C18 cartridges were conditioned with 3 mL of ACN/water (90:10 v/v). 1 mL of stevia extract was loaded on the cartridges. The steviol glycosides were eluted with ACN/water (90:10 v/v). The eluate was subjected to HPLC analysis with amino and HILIC columns.

4.3. Results & Discussion

An optimized analytical method was developed for steviol glycoside analysis in stevia leaf methanol extracts using LC-MS. Typically for steviol glycoside extraction, solvent extraction using hot water or acetonitrile is employed followed frequently by further solid phase extraction (SPE) sample clean up. The challenge of steviol glycoside extraction lies in the different solubilities of steviol glycosides in aqueous and organic solvents. Good solubility of all steviol glycosides has been reported for water, however co-extraction of phenolic constituents and carbohydrates exacerbate separation problems and therefore analysis. For this reason, first, optimization to use different organic solvents for steviol glycoside extraction from dried leaf material using Soxhlet extraction was performed. Despite reports that rebaudioside A shows moderate methanol solubility, we first optimized for Soxhlet extraction using methanol. Prior to Soxhlet extraction 2 g of dried leaves were treated with liquid nitrogen and crushed and milled using a blade mill. Extraction times, extraction cycles and extraction volume were optimized by multiple extraction experiments and it was concluded that using 2 g of leaf material allowed for a reproducible amount of steviol glycosides being extracted from notoriously heterogenic plant material.

For development of the LC-MS method, a standard reversed phase C18 column to a HILIC column was compared. C18 columns have the advantage that steviol glycosides and all phenolic constituents can be analysed and quantified present in stevia leaves, however retention times are long and selectivity is not satisfactory. Co-elution of rebaudioside A with stevioside was observed using a C18 column (Figure 13). At retention times up to 25 min. a total of twelve chlorogenic acids and nine flavanone glycosides were detected from the analysis of commercially obtained stevia leaves (please refer to article attached in appendix).

Hande Karaköse Jacobs University Bremen 34

Phytochemical Characterization of Stevia rebaudiana

Intens 7 x10. Chlorogenic acids and flavonoids Diterpene glycosides 1.25

1.00 stevioside 0.75

0.50 RebA

0.25

0.00 0 5 10 15 20 25 30 35 40 45 Time [min]

Figure 13.Total ion chromatogram in negative ion mode using a C-18 column of methanolic Stevia rebaudiana extract showing phenolics (chlorogenic acids, flavonoids and steviol glycosides).

A total of eight steviol glycosides could be observed at retention times from 30 to 42 minutes with rebaudioside A and stevioside co-eluting. Efficient separation and resolution of this isomer pair and rubusoside/steviolbioside were achieved with HILIC column using acetonitrile/water (10 mM ammonium formate) as solvent in the HPLC method. In contrast to C18 column, the elution order is inverted on the HILIC column. The more glucose units attached to the backbone structure resulted in later retention times on HILIC. Thus, it was not possible to detect steviol on the HILIC column. A base peak chromatogram (BPC) is presented in Figure 14 showing baseline separation of all naturally occurring steviol glycosides. Characterization of steviol glycosides was achieved by ion-trap mass spectrometry with selected ion monitoring (SIM), and confirmation of elemental composition was provided by ESI-TOF measurements (Table 1).

Stevioside Steviolbioside

Rubusoside Dulcoside RebA A RebC

Figure 14.Base peak chromatogram of steviol glycosides obtained using HILIC column.

Hande Karaköse Jacobs University Bremen 35

Phytochemical Characterization of Stevia rebaudiana

4.3.1. Identification of Steviol Glycosides Compound identification was carried out using high resolution mass spectrometry (HR-MS) followed by tandem MS using an ion trap mass spectrometer. HR-MS values for all steviol glycosides are given in Table 1.

OR1

CH3 CH2

H3C COOR General structure of steviol glycosides Table 1.High resolution mass spectrometrical data of stevia terpene glycosides in negative ion mode from LC-TOF MS analysis

Compound R R1 Molecular Experimental Theoretical Relative Error Formula m/z (M-H+)- m/z (M-H+)- (ppm)

Steviol H H C20H30O3 317.0819 317.0717 9.0 Steviolbioside 2 1 H glc - glc C32H50O13 641.3181 641.3179 0.4

Rubusoside Glc glc C32H50O13 641.3166 641.3179 2.0

2 1 Stevioside Glc glc - glc C38H60O18 803.3751 803.3707 5.5

2 1 Rebaudioside A Glc glc3 - glc C44H70O23 965.425 965.4235 1.6

1glc 2 1 Rebaudioside B H glc3 - glc C38H60O18 803.368 803.3707 2.8

1glc 2 1 Rebaudioside C Glc glc3 - rham C44H70O22 949.427 949.4286 1.7 (Dulcoside B) 1glc 2 1 2 1 Rebaudioside D glc - glc glc3 - rham C50H80O28 1127.4726 1127.4763 3.3

1glc 2 1 2 1 Rebaudioside E glc - glc glc - glc C44H70O23 965.4199 965.4235 3.7

2 1 Rebaudioside F Glc glc3 - xyl C43H68O22 935.4097 935.4129 3.5

1glc 2 1 Dulcoside A Glc glc - rham C38H60O17 787.3732 787.3758 3.3

Hande Karaköse Jacobs University Bremen 36

Phytochemical Characterization of Stevia rebaudiana

Representative tandem MS spectra and fragmentation pathways are shown in Figure 15 and Figure 16 for rebaudioside A ([M-H+]- m/z 965) and rebaudioside E ([M-H+]- m/z 965). These two isomers were differing in their MS2 fragmentation, resulting with m/z 803 peak [M-H+-glc]- - with the loss of one glucose unit ([M-H2O] m/z 162) for rebaudioside A, and m/z 641 peak with the loss of two glucose units for rebaudioside E ([M-H+-2glc]-). It is most probable that the both rebaudiosides are losing the glucose first from carboxylic acid moiety due to the increased stability of the resulting resonance stabilized anion. In general steviol glycosides could be characterized up to MS4. Rebaudioside A loses one glucose unit each in MS3 and MS4 resulting with the m/z 479 and m/z 317 peaks corresponding to [M-H+-3glc]- and [M-H+-4glc]- ions, respectively.

OH OH OH OH OH OH HO O O

MS3 -162 O OH OH HO O O OH OH O OH HO OH MS4 -162 O O O O OH OH CH H C CH2 H C 2 3 HO 3 HO HO

HO HO HO Rebaudioside A O O Rebaudioside E O CH CH3 3 HO O O HO O O O OH OH OH MS2 -162 MS2 -324

Figure 15.Mechanism of fragmentation in tandem MS spectra of Rebaudioside A and Rebaudioside E illustrating how isomeric compounds can be distinguished by tandem MS.

Hande Karaköse Jacobs University Bremen 37

Phytochemical Characterization of Stevia rebaudiana

Intens x10. 7 965.6 -MS, 2 1001.6 1 0 1.5 803.4 -MS2 1.0 0.5 0.0 641.2 -MS3 0.5 317.1 413.1 479.1 0.0 -MS4 479.1 1 317.1 0 200 400 600 800 1000 m/z

Inten x10s. 5 641.2 -MS2 2 322.9 479.1 803.4 0 -MS3 1.0 479.1 0.5 317.0 0.0 4 -MS4 2 317.0

1

0 200 400 600 800 1000 m/z

Figure 16.Tandem MS spectra of Rebaudioside A (above) and Rebaudioside E (below) in negative ion mode.

Tandem mass spectra for rebaudioside D is presented in Figure 17. The structure losses one - 2 + - glucose unit ([M-H2O] m/z 162) in MS resulting with the [M-H -glc] ion with m/z 787 and in MS3 [M-H+-2glc]- with m/z 625. In MS4 rebaudioside D loses one rhamnose sugar unit ([M- - + - H2O] m/z 146) resulting with the [M-H -2glc - rham] with m/z 479 ion in negative mode. Further tandem MS data of steviol glycosides are presented in appendix A.

Intens. x10 7 949.6 -MS 1 985.6

0 787.4 -MS2 0.5

0.0 6 x10 -MS3 4 625.2 2 479.1 0 x10 6 -MS4 1 479.1 317.1 0 200 400 600 800 1000 m/z

Figure 17.Tandem MS spectra of Rebaudioside D in negative ion mode.

Hande Karaköse Jacobs University Bremen 38

Phytochemical Characterization of Stevia rebaudiana

4.3.2. Method Validation

The validation procedure involved determination of limit of quantitation (LOQ), determination of linear range for quantitation, repeatability studies including multiple injection and day to day repeatability. Additionally, inter sample repeatability experiments were carried out.

Sensitivity & Selectivity

Retention times of the reference compounds of rebaudioside A and stevioside were compared to the chromatograms obtained from the leaf extracts, apart from that, the accurate masses were obtained for each peak in the chromatogram from HR-MS measurements.

The calibration curve was linear in the range of 10 – 500 μg/mL for stevioside and 5 – 500 μg/mL for rebaudioside A and steviolbioside. The equations of calibration curves obtained by the least square method were as follows;

Stevioside: y = 35999x - 845055 R² = 0.9925 Rebaudioside A: y = 39360x + 287715 R² = 0.9935 Steviolbioside: y= 246929x –100000 R2 = 0.9985 where y is the peak area from the LC chromatogram and x is the μg/mL for rebaudiosideA and stevioside.

Precision

Precision was calculated based on intra and inter-day (n=3) repeatability. Standard solution of rebaudioside A at the concentrations of 5, 10, 100 and 500 μg/mL and 10, 50, 300 and 350 μg/mL for stevioside were measured on three different days on the HILIC column and the results were evaluated by calculating the %RSD. The repeatability of the inter-day measurements was in the range of 4.1 - 6.7 % for rebaudioside A and 1.9 – 5.4 % for stevioside.

Intra-day measurements were evaluated by calculating the %RSD of three injections of each concentration of 50, 200 and 300 μg/mL of rebaudioside A and 50, 250, 350 μg/mL of stevioside. Intra-day precision was in the range of 2.4 – 5.6 % for rebaudioside A and 0.9 – 3.7 % for stevioside.

Hande Karaköse Jacobs University Bremen 39

Phytochemical Characterization of Stevia rebaudiana

Limit of detection was determined as the concentration of the component providing signal to noise ratio (S/N) of three and for limit of quantification, the concentration resulting in a S/N of 10. According to that, for stevioside LOD was obtained as 2.5 ppm and LOQ as 10 ppm. For rebaudioside A, LOD was 5 ppm and LOQ was obtained as 10 ppm. For steviol glycoside LOD was 2 ppm and LOQ was 5 ppm.

Accuracy

The accuracy of the method was determined by calculating the relative error observed in a standard addition experiment. Stevia leaf extracts were spiked with different amounts of stevioside and rebaudioside A separately. The relative errors was in the range of 0.043 – 0.074 μg/mL for rebaudioside A and 0.056 – 0.14 μg/mL for stevioside

Comparison to UV data quantification

Calibration curves were also obtained from UV measurements (210 nm) for stevioside. The sensitivity of the method was less compared to results of LC-MS using EIC. The linearity range was 50 – 500 μg/mL. S/N of 2:1 was achieved with concentration of 75 ppm. The equation of calibration curve obtained by the least square method was y = 2.8799x – 90.019 (R2 = 0.9972). The %RSD of triple injections of 50, 250, 350 μg/mL of the standard was respectively, 19.1, 9.9, and 3.4.

4.3.3. Comparison to SPE sample clean up

Conventional methods for steviol glycoside quantification frequently employ SPE sample pretreatment followed by UV based quantification. Two problems might arise here, which have never been addressed: Firstly, do after SPE treatment analytes co-elute in the steviol glycosides not observed but co-quantified by UV and secondly, do SPE materials retain steviol glycosides. Two different SPE stationary phases reported in the literature were tested for presample treatment of stevia extracts prior to HPLC-MS analysis. SPE with material I was resulting with decrease in the peaks in the LC chromatogram especially with rebaudioside A, with the increased amount of material I in the cartridge (Figure 18). SPE cleanup with material II did not have a dramatic effect on the HPLC analysis with both amino and HILIC columns and gave the similar results if compared with material I based procedure (Figure 19).

Hande Karaköse Jacobs University Bremen 40

Phytochemical Characterization of Stevia rebaudiana

Consequently, great care is required when using SPE based protocols since retention of rebaudioside A, the most lipophilic steviol glycoside, might lead to non-satisfactory accuracy.

stevioside 0.2 g Material I

Rebaudioside C Rebaudioside A

0.6 g Material I

Figure 18.Total ion chromatograms for comparison of different amounts of material I in SPE cleanup procedure.

0.2 g Material II

0.2 g Material I

Figure 19.Total ion chromatograms for comparison of SPE cleanup of the stevia extract with materials I and II cartridges.

Hande Karaköse Jacobs University Bremen 41

Phytochemical Characterization of Stevia rebaudiana

Additionally, it was shown that for both SPE materials I and II if using identical HPLC- conditions to those reported in the literature using an MS detector co-elution of several analytes with Rebaudioside A, E and Dulcoside were observed (appendix B, page 122). The nature of the co-eluting analytes remains largely unknown, however, it can be anticipated that some of them display UV absorption at 210 nm, again leading to non-satisfactory accuracy. These critical assessments of SPE based LC-UV quantification methods clearly show that this type of method require urgent improvement and amendments and are inferior to LC-MS based methods without SPE sample pretreatment.

4.3.4. Quantification of Steviol Glycosides

Steviol glycoside levels were quantified in all 166 samples made available within the project. Quantification of steviol glycosides were performed using the validated method with HILIC column. For three selected steviol glycosides (rebaudioside A, stevioside and steviol), calibration curves were obtained using six-point calibration from the extracted ion chromatogram (EIC) of LC-TOF measurement. The quantities of other steviol glycosides were calculated relatively according to stevioside values for each sample. Please refer to section 4.3.2 for the calibration curve data.

Data analysis reveals that there are distinct differences between the steviol glycoside profile in Stevia rebaudiana leaves in the seven different varieties analyzed and distinct differences between Stevia rebaudiana leaves from different origins. The average amounts and range (min&max values) of quantity of each steviol glycoside in all 166 samples is presented in Table 3. Detailed quantification data can be found in appendix B.

Table 3.Steviol glycosides values from 166 samples

Steviol glycoside Average (mg/100 g leaves) Range (mg/100 g leaves) Rebaudioside A 1.017 79 - 5336 Stevioside 6071 252 - 17509 Dulcoside A 239 5 - 680 Rubusoside 111 5 - 459 Rebaudioside C 282 26 - 820 Total 9036 554 - 18067

Hande Karaköse Jacobs University Bremen 42

Phytochemical Characterization of Stevia rebaudiana

Additionally, from the obtained quantification data, the samples were analyzed on a radar plot based on their variety and origin (Figure 20 and 21). Within the project seven defined botanical varieties of Stevia rebaudiana were cultivated and their steviol glycoside profile was determined. The distinct differences in the stevioside profile of different variety of stevia leaves can be easily recognized. From the radar plot (Figure 20), it can be seen that variety 5 and non-EU samples have the maximum value of stevioside, whereas variety 7 having the minimum value for stevioside but maximum value for rebaudioside A. Stevia rebaudiana was within this project cultivated in nine different locations within the EU. Additionally samples from outside the EU were available for comparison. In Figure 21, EU cultivated Stevia rebaudiana shows higher concentrations of stevioside (Amiflikeia and Argentinie having the maximum end values). However for rebaudioside A, Conaga cultivated stevia shows higher end concentration values.

1 10.000 8.000 Non-EU 2 6.000 4.000 RebA 2.000 Stevioside 7 0.000 3 Dulcoside A Rubusoside RebC

6 4

5

Figure 20.Radar plot of steviol glycoside concentrations varying between seven varieties (average values taken within +/- 3 σ) and in comparison to non-EU samples. Concentrations are given on radial axis in g/100g dry leaf material. Outer numbers are indicating the 7 varieties and non-EU samples; numbers inside the plots are indicating the concentrations.

Hande Karaköse Jacobs University Bremen 43

Phytochemical Characterization of Stevia rebaudiana

TCV 12.000 Conaga Uconor 10.000 8.000 APTTB 6.000 Agrinion RebA 4.000 2.000 Stevioside Amiflikeia 0.000 Toumpa Dulcoside A Rubusoside RebC Turkei Portugal

Granada Amfilia Argentinie

Figure 21.Radar plot of steviol glycoside concentrations varying between all origins (average values taken within +/- 3 σ) and in comparison to non-EU samples. Concentrations are given on radial axis in g/100g dry leaf material.

4.4. Conclusion

In conclusion, steviol glycosides were successfully analyzed and quantified using LC-MS technique. Stevia leaves can be analyzed using a variety of different LC-MS methods. While LC- MS on a C18 column allows analysis of steviol glycosides, however suffering from selectivity problems, analysis on a HILIC column coupled to ESI-TOF detection allows separation and quantification of all known naturally occurring steviol glycosides. Linear range, sensitivity and reproducibility were excellent. Using both high resolution MS and tandem MS on an ion trap instrument reliable structure confirmation can be carried out based on characteristic MSn fragment spectra of all steviol glycosides.

Distinct differences in the quantity of steviol glycosides within the stevia samples were observed. From the data it was observed that variety 7 is having the maximum concentration for rebaudioside A but minimum value for stevioside concrentration. EU origin cultivated stevia samples have the maximum concentration for stevioside but average value for rebaudioside A concentration.

Hande Karaköse Jacobs University Bremen 44

Phytochemical Characterization of Stevia rebaudiana

5. POLYPHENOLS in STEVIA REBAUDIANA 5.1. Overview

The hydroxycinnamate derivatives of S. rebaudiana have been investigated qualitatively and quantitatively by LC-MSn. Chlorogenic acids and flavonoid glycosides of Stevia rebaudiana from different origins in all around Europe with seven different botanical varieties were profiled and quantified. The correlation study between CGAs, differences between stevia samples and effect of origin and variety on the CGA profile were tested statistically from the obtained dataset.

5.2. Materials & Methods 5.2.1. Sample Preparation

Two grams of S. rebaudiana leaves was immersed in liquid nitrogen, ground in a hammer mill, and extracted first with 150 mL of chloroform in a Soxhlet apparatus (Buchi B-811 extraction system) for 2 h for removal of lipid fraction and then with 150 mL of methanol for another 2 h. Solvents were removed from the methanolic extract in vacuo, and extracts were stored at - 20 oC until required.

5.2.2. LC-MS Analysis of Polyphenols

LC-TOF MS: The LC equipment (Agilent 1100 series, Bremen, Germany) comprised a binary pump, an autosampler with a 100 μL loop, and a diode array detector with a light-pipe flow cell (recording at 254 nm and scanning from 200 to 600 nm). This was interfaced with a MicroTOF Focus mass spectrometer (Bruker Daltonics) fitted with an ESI source. The MS parameters were: nebulizer 1.6 bar, dry gas 12.0 L/min, dry temperature 220 0C. The MicroTOF was operated in negative ion mode and the mass range was 150 – 1200 m/z. Internal calibration was achieved with 10 mL of 0.1 mol/L sodium formate solution injected through a six-port valve prior to each chromatographic run. Calibration was carried out using the enhanced quadratic calibration mode.

LC-MSn: The LC equipment (Agilent 1100 series, Bremen, Germany) comprised a binary pump, an autosampler with a 100 μL loop, and a diode array detector with a light-pipe flow cell (recording at 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 Auto-MSn mode to obtain fragment ions 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 V, starting at 30% and ending at 200%. MS operating

Hande Karaköse Jacobs University Bremen 45

Phytochemical Characterization of Stevia rebaudiana conditions (negative mode) were capillary temperature of 365 oC, a dry gas flow rate of 10 L/min, and a nebulizer pressure of 50 psi.

HPLC: Separation was achieved on a 250 x 3 mm C18 column (Varian Pursuit XRS) with 5 μm particle size. Solvent A was water/formic acid (1000+0.005 v/v), and solvent B was acetonitrile (ACN). Solvents were delivered at a total flow rate of 0.5 mL/min and the column temperature was set to 25 oC. 5 μL of samples were injected in to LC-MS system, unless stated otherwise. The gradient profile was 10 to 80% B in 60 min and a return to 10% B at 65 min and 5 min isocratic to re-equilibrate.

5.2.3. Calibration Curve of Standard Compounds

Most abundant chlorogenic acid derivatives (3-CQA, 4-CQA, 5-CQA, 3,5-diCQA, 4,5 diCQA) and two flavonoid glycosides (quercetin-3-glycoside and kaempferol-7-glycoside) were chosen for calibration curves.

Stock solutions of the standard compounds were prepared in 80% ACN/water. A series of standard solutions was injected (5μL) into the LC-MS system. The areas of the peaks of each standard from extracted ion chromatograms (EIC) were used to make the respective standard curves. 5.2.4. Hydrolysis of Flavonoid Glycosides 5 mg crude extract was dissolved in 2 ml 2M HCl and heated at 90 0C for 40 min. Sample was then directly used for LC-MS or diluted with MeOH.

5.2.5. Statistical Analysis Statistical analyses of the data were performed using IBM SPSS 20. The distributions of the variables were tested for normality using the Kolmogorov-Smirnov test. Associations between the variables were investigated using both parametric (Pearson’s correlation) and non-parametric (Spearman’s correlation) techniques. Results were interpreted using the widely accepted 5% level of significance. To test whether there were differences on each chlorogenic acid with respect to its origin or variety, separate one-way ANOVA analyses was employed, followed by two post-hoc tests: Fisher’s Least Significant Difference (LSD) as the least conservative test where equal variances are assumed and Games-Howell test where non-equal variances are assumed for the multiple pair

Hande Karaköse Jacobs University Bremen 46

Phytochemical Characterization of Stevia rebaudiana wise comparison tests. All empirical results were interpreted using the widely accepted 5% level of significance (p < 0.05).

5.3. Results & Discussion

Stevia extract was analyzed by LC-MSn in the negative ion mode using an ESI ion-trap mass spectrometer, allowing assignments of compounds to regioisomeric level, and also by HR-MS using ESI-TOF in negative ion mode connected to LC, that allowed determination of molecular formulae based on the accurate mass measurements. Molecular formulas were in general accepted if an error below 5 ppm was experimentally observed, as accepted by all peer reviewed chemistry journals. In a second experiment tandem MS experiments were carried out and the observed fragmentation patterns compared to those of authentic reference materials (either obtained commercially or from our own laboratory). After obtaining multi-dimensional information of four parameters on chromatographic retention times, UV-spectra (UV-VIS DAD detector coupled to LC-MS system), HR-MS and tandem-MS data comparison to authentic reference material did allow compound identification. Peak assignments of CGAs have been made on the basis of structure diagnostic hierarchical keys previously developed81, 135.

Chlorogenic acids and flavonoids were quantified using an established reversed phase LC-MS method on a C-18 column using ESI-TOF-MS in the negative ion mode. All required analytes showed baseline separation with exception of the pair 3,4- and 4,5 dicaffeoyl quinic acid. A typical chromatogram using a reversed phase C-18 column of a stevia extract showing polar polyphenols at early retention times and more lipophilic steviol glycosides at later retention times is shown in Figure 22. Abbreviations and numbering of CGAs and flavonoids are presented in Figure 23, Table 4. In all cases quantitation was carried out using extracted ion chromatograms only. Additionally flavonoids were quantified using kaempferol-7-glucoside and quercetin-3-glucoside as reference standards, resulting for flavonoids in relative values rather than absolute values.

Hande Karaköse Jacobs University Bremen 47

Phytochemical Characterization of Stevia rebaudiana

Intens. x10 6 CGAs & Flavonoid glycosides Steviol glycosides

2.5

2.0 12

11 2 1.5

1.0 3 1 13 4

0.5 5 6 8 10 14 9 7 0.0 5 10 15 20 25 30 35 40 Time [min]

Figure 22.Base peak chromatogram in negative ion mode using a C18 column of methanolic Stevia rebaudiana extract showing phenolics (CGAs, flavonoids and steviol glycoside).

Table 4.Chromatographic and MS data on flavonoid glycosides and CGAs present in stevia leaves

Compound Number Compound* m/z [M-H+]- 1 3-caffeoylquinic acid (3CQA) 353 2 5-caffeoylquinic acid (5CQA) 353 3 4-caffeoylquinic acid (4CQA) 353 4 Rutin 609 5 Quercetin-galactoside 463 Kaempferol-glucopyranoside 6 447 Quercetin-rhamnoside 7 Quercetin-fructoside; Luteolin-glucuronide 461 8 Quercetin-pentoside 433 kaempferolxylosylglucoside 9 579 Naringin 10 Apigenin-galactoside 431 11 3,5-di-caffeoylquinic acid (3,5diCQA) 515 12 4,5-dicaffeoylquinic acid(4,5diCQA) 515 13 Quercetin-diglucoside-rhamnoside 771 14 Kaempferol-glucosylrhamnosyl-glucoside/galactoside 755 Kaempferol-rhamnopyranosyl-glucopyranoside(rutinoside) isomers 15 Quercetin-dirhamnoside 593 Apigenin-diglucoside/galactoside 16 Quercetin-trisaccharide 741 17 Kaempferol 3-rhamnopyranosyl-rhamnopyranosyl-glucopyranoside 739 *Compounds named for flavonoid glycosides are only possible structures, which were not identified or confirmed.

Hande Karaköse Jacobs University Bremen 48

Phytochemical Characterization of Stevia rebaudiana

OH OH OH HOOC OH 3 O OH O O OH O HOOC O

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

O O O OH O 5 O HOOC OH HOOC 4 O O 5 3 3 HOOC 4 O OH O O OH O O OH OH OH OH 3,5-diCQA 3,4-diCQA 4,5-diCQA

OH OH HO OH OH

HO OH O O HO O 5 O HOOC 4 O OH OH OH OH OH O OH cis-4,5-diCQA Kaempferol

OH OH OH OH OH HO O HO O HO O

OH OH O OH O OH O

Quercetin Luteolin Apigenin

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

Kaempferol-7-O-beta-D-glucoside Quercetin-3-O-beta-D-glucoside Figure 23.Structures of caffeoylquinic acids and flavonoid glycosides.

Hande Karaköse Jacobs University Bremen 49

Phytochemical Characterization of Stevia rebaudiana

5.3.1. Characterization of Chlorogenic acids Mono-caffeoylquinic acids (mono-CQA) and di-caffeoylquinic acids (di-CQAs) were identified in stevia using the hierarchial keys previously developed81. It is possible to discriminate between the isomers of caffeoylquinic acids and feruloylquinic acids by using LC-MSn. The fragmentation pattern depends on the stereochemical relationships between the substituents on the quinic acid moiety. Four peaks were detected at m/z 353.1 and assigned as well-known 3- CQA, trans-5-CQA, and 4-CQA and cis-5CQA. Three dicaffeoylquinic acid isomers were identified by their parent ion m/z 515.2 and were assigned as 3,5-diCQA, 3,4-diCQA, and 4,5- diCQA using the hierarchial keys81, 135. Two further peaks present as minor components showed fragmentation patterns similar to that of 4,5-diCQA, which were identified as cis isomers of 4,5- diCQA. Figures 24 - 26 show selected data including an extracted ion chromatogram for monocaffeoyl quinic acids and two tandem mass spectra for selected regioisomeric stevia caffeoyl quinic acids as typical secondary metabolites. All other fragmentation patterns are provided in appendix C.

Twenty-four hydroxycinnamic acid derivatives of quinic and shikimic acid were detected in the work using stevia leaves not cultivated in this project and the results have been published in the course of this project (Please refer to the article attached in appendix for detailed insight).

Hande Karaköse Jacobs University Bremen 50

Phytochemical Characterization of Stevia rebaudiana

Inte 2 EIC 353.0 -All MS x10ns. 8 2.0

1.5

1.0 1 3 Cis-2 0.5

0.0 0 10 20 30 40 50 Time [min]

OH OH HOOC OH OH OH 1 3 HO O OH O

O

O 5 OH HOOC OH HOOC O OH 1 O OH OH OH OH OH

Figure 24.Extracted ion chromatogram of m/z 353 of three mono-caffeoylquinic acids, from left to right: 3-Caffeoylquinic acid (1), 5-Caffeoylquinic acid (2) and 4-Caffeoylquinic acid (3) in negative ion mode.

All monoacyl CGA (m/z ~ 353) gave the expected parent ion (monoacyl CGA - H+) (Figure 25 and 26) in tandem MS analysis. 3-CQA and 5-CQA produce an MS2 base peak at m/z ~ 191 + - 3 corresponding to [quinic acid-H ] (Q1 ion) and in MS it fragments to Q2 ion with m/z 85.1 and + - [quinic acid – H2O - H ] (Q3 ion) at m/z 172.8 (Figure 25). 3-CQA might be discriminated by its MS2 peak at m/z ~135 and by slight difference of the intensity of m/z ~178 in MS2. It is easy to distinguish the 4-substititued CGA by it is dehydrated quinic acid moiety which gives MS2 base peak at m/z ~173.

Hande Karaköse Jacobs University Bremen 51

Phytochemical Characterization of Stevia rebaudiana

Intens. -MS x107 3 352.9 2 1 472.9 729.2 0 x107 O -MS2(352.9) 190.7 1.0 HOOC HO C CH OH H 0.5 OH OH HO Caffeic m/z 135 134.8 Q1 0.04 x10 -MS3(353.1->190.6) 6 85.1 O 172.8 O 4 HOOC OH 2 Q OH OH OH 2 Q3 0 200 400 600 800 1000 m/z Figure 25.Consecutively MS, MS2and MS3 spectra of 3-Caffeoylquinic acid in negative ion mode.

Intens. -MS x107 352.9 3 2 1 374.9 472.8 0 x106 -MS2(352.9) 6 172.7 4 2 0 x105 -MS3(353.1->172.7) 93.0 1.0

0.5 71.3 154.7 0.0 200 400 600 800 1000 m/z

Figure 26.Consecutively MS, MS2and MS3 spectra of 4-Caffeoylquinic acid in negative ion mode. The diacyl CGAs behave similarly, giving the parent ion [diacyl CGA – H+]- at m/z 515. DiCQAs loses the caffeic acid moiety in MS2 yielding a [diacyl CGA – cinnamoyl – H+]- and these ions are identical to the parent ions obtained from monoCQAs. 4,5-diCQA give dehydrated quinic acid moiety as base peak at m/z ~173 in MS3 as previously seen for 4-CQA. This ion was not observed for 3,5-diCQA, instead MS3 base peak was at m/z 191 as previously observed for 3- CQA (Figure 27 and 28). Thus, 3,5-diCQA can be distinguished easily from the 4-acylated caffeoylquinic acid isomers.

Hande Karaköse Jacobs University Bremen 52

Phytochemical Characterization of Stevia rebaudiana

Intens. -MS x108 515.0 1 1031.3 352.9 613.0 0 x108 352.9 -MS2(515.0) 0.5 190.7 0.0 x107 -MS3(515.3->352.9) 190.7 2 134.7 0 x105 -MS4(515.3->353.1->190.7) 2 126.8 1 85.1 172.7 0 200 400 600 800 1000 m/z Figure 27.Consecutively MS, MS2and MS3 MS4 spectra of 3,5-dicaffeoylquinic acid in negative ion mode.

Intens. -MS x108 515.0 1 1031.4 0 x108 -MS2(515.0) 352.9 0.5 172.7 254.8298.9 0.0 x107 -MS3(515.3->352.9) 172.7 1 134.8 0 x105 -MS4(515.3->353.1->172.8) 4 93.0 2 0 200 400 600 800 1000 m/z Figure 28.Consecutively MS, MS2and MS3 MS4 spectra of 4,5-dicaffeoylquinic acid in negative ion mode.

Hande Karaköse Jacobs University Bremen 53

Phytochemical Characterization of Stevia rebaudiana

5.3.2. Characterization of Flavonoid Glycosides Flavonoid glycosides are an important group of plant natural products in which different type of sugars are linked to an aglycone. Determination of the identity and position of linkage to the aglycone of sugars by mass spectrometry alone is challenging. From the mass spectra of flavonoid glycosides using tandem MS we can obtain molecular mass, structure of the aglycone (from its m/z), number of sugar rings and their configuration. Negative ion mode was selected for the analyses because previous results suggested that negative mode was more sensitive than positive ion mode. A total of twelve peaks in the chromatogram corresponding to flavonoid glycosides were identified. All compounds could be identified as belonging to this class of compounds due to their characteristic fragmentation patterns in tandem MS showing neutral losses of sugar moieties following by characteristic fragment spectra of the aglycones. The nature of the aglycones was further substantiated by hydrolysis of the total phenol extract followed by LC-MS analysis revealing that four flavonoid aglycones quercetin, kaempferol, luteolin and apigenin (Figure 29 and Table 4) are present in Stevia rebaudiana leaves.136

OH OH OH OH OH OH

HO O HO O HO O HO O

OH OH OH O OH O OH O OH O

Kaempferol Quercetin Luteolin Apigenin Mw 286 Mw 302 Mw 286 Mw 270

Figure 29.Chemical structure of four flavonoid aglycones identified in Stevia rebaudiana leaves.

Although the aglycone and the glycane were identified for an observed m/z, the accurate structure of the flavonoids glycoside could not be determined because identity and the site of connection of monosaccharide cannot be determined by LC-MS. The possible structures of compounds were determined by comparison of the mass spectral data obtained with literature data. However, the detailed chemical structure of these flavonoids could not be established unambiguously by LC-MS. None of the compounds present were shown to be identical to any of the twelve reference standards used commercially or to reference compounds available in our laboratory. A preparative LC isolation and full structure elucidation of the compounds was

Hande Karaköse Jacobs University Bremen 54

Phytochemical Characterization of Stevia rebaudiana outside the scope and timeline of the project but needs further attention. A table showing the chromatographic and MS characteristics of the flavonoids was given previously in Table 4 and extracted ion chromatogram and tandem mass spectra for selected compound is shown in Figure 30 and 31. All other tandem mass spectral data of flavonoid glycosides are presented in appendix C. The possible fragmentation pattern and ion nomenclature of flavonoid glycosides is illustrated on luteolin-7-O-rutinoside in Figure 32.

Intens. EIC 447.0 -All MS x108 3 1.25

1.00

0.75 1 0.50

0.25 2

0.00 5 1 1 2 2 3 Time 0 5 0 5 0 [min]

Figure 30.Extracted ion chromatogram of m/z 447.0 in negative ion mode.

Intens. [%] -MS 1 447.0 100 50 0 -MS2(447.0) 284.8 100 50 0 -MS3(447.3->286.8) 216.7 100 174.7 50 0 100 200 300 400 500 600 700 m/z

Figure 31.An example of tandem MS spectra for compound 1, revealing its identity as kaempferol glucopyranoside.

Hande Karaköse Jacobs University Bremen 55

Phytochemical Characterization of Stevia rebaudiana

Y 0

Z0 B0

X0 OH CH2OH

O O O OH OH

OH OH

A1 B OH O 1 A 0 Figure 32.Fragmentation illustration on luteolin-7-glucoside137

The most useful fragmentation for flavonoid aglycone identification are the cleavage of two C— C bonds of the C-ring, resulting in structural information for A and B ions (Figure 32). These ions can be rationalized by retro-Diels–Alder (RDA) reactions and are the most diagnostic fragments for flavonoid identification since they provide information on the number and type of substituents in the A- and B-rings. The flavonoid aglycone fragment ions can be designated according to the nomenclature proposed by Ma et al 138, 139.

5.3.3. Quantification of Chlorogenic acids & Flavonoid Glycosides Phenolics including chlorogenic acids and flavonoids were quantified using an established reversed phase LC-MS method on a C-18 column using ESI-TOF-MS in the negative ion mode. Standard solutions were analyzed using the same chromatographic method as used for stevia leaf extracts as indicated before (chapter 5.2.2). The calibration curves were obtained by the external standard method on six levels of concentration of reference compounds. For quantitation the six most abundant chlorogenic acid derivatives (3-CQA, 4-CQA, 5-CQA. 3,4-diCQA, 3,5-diCQA, 4,5 diCQA) were chosen and calibration curves were obtained with excellent linearities. In all cases quantitation was carried out using extracted ion chromatograms only. Cis isomers of 5- CQA and 4,5-diCQA were quantified based on the corresponding calibration curves of trans isomers, resulting in relative values of cis isomers (please refer to appendix D for quantification data, page 140). Additionally flavonoids were quantified using kaempferol-7-glycoside (k7g) and

Hande Karaköse Jacobs University Bremen 56

Phytochemical Characterization of Stevia rebaudiana quercetin-3-glycoside (q3g) as reference standards, resulting for flavonoids in relative values rather than absolute values (appendix D).

The linearity and the equations of calibration curves obtained by the least square method were as follows:

Reference compound Linearity (μg/mL) Equation R2 3-CQA 1-100 82912x+346549 0.99 5-CQA 5-500 49062x+907693 0.99 4-CQA 1-150 129889x+395550 0.99 3,5-diCQA 10-200 138064x+7000000 0.98 4,5-diCQA 10-400 138243x+4000000 0.98 K7g 50-1000 26105x+6000000 0.98 Q3g 2-150 97003x+2000000 0.98

Where y is the peak area from the LC chromatogram and x is the μg/mL for the reference compound.

5.3.3.1. Sample Variation

From the obtained quantification data, a series of statistical analysis was carried out. As a first step, for each variety, origin and harvest average values and standard deviations were determined. Additionally, minimum and maximum values for each sample subgroups are given in the tables.

Within the project seven defined botanical varieties of Stevia rebaudiana were cultivated and their phenolic profile was determined. From the data variations between different batches and average values averaged over all samples from a single variety can be compared. Additionally variations in single compound quantities, quantities of groups of compounds (e.g. mono-caffeoyl quinic acids, dicaffeoyl quinic acids) or ratios of two single compounds can be compared.

From the data for example it can be seen that the average concentration of all monocaffeoyl quinic acids remains rather constant over all varieties (2.123 - 2.686 g/100g), whereas a more spread of data is observed for dicaffeoyl quinic acids (1.484 – 2.432 g/100g). Varieties 5, 6, 7 and 3 show on average increased levels of chlorogenic acids compared to varieties 2. Average values are given in Table 5, for detailed quantification data please refer to appendix D. All EU

Hande Karaköse Jacobs University Bremen 57

Phytochemical Characterization of Stevia rebaudiana

Stevia rebaudiana varieties show considerable higher levels of chlorogenic acids and flavonoids if compared to some samples obtained from outside the EU. Variations can be displayed in a radar plot shown in Figure 33, 34, 35. Here the lower chlorogenic acid content in non-EU samples as well as in variety 2 can be appreciated.

Table 5.Average values (taken within +/- 3 σ) for chlorogenic acids in seven different varities

3CQA 4CQA 5CQA Total mono 3,5-diCQA 4,5-diCQA TotaldiCQA Variety (g/100g (g/100g (g/100g (g/100g (g/100g (g/100g (g/100g leaves) leaves) leaves) leaves) leaves) leaves) leaves)

1 0.279 0.096 1.921 2.262 0.925 1.159 1.969

2 0.267 0.094 1.804 2.208 0.880 1.208 1.807

3 0.290 0.114 2.267 2.650 1.101 1.213 2.022 4 0.294 0.118 2.258 2.616 1.101 1.259 2.020

5 0.279 0.105 2.108 2.466 1.365 1.313 2.432

6 0.261 0.105 2.100 2.686 1.309 1.396 2.272

7 0.178 0.113 2.262 2.507 1.277 1.488 2.108

Non-EU 0.195 0.092 1.836 2.123 0.881 0.700 1.484

Hande Karaköse Jacobs University Bremen 58

Phytochemical Characterization of Stevia rebaudiana

1 2.500

Non-EU 2.000 2 1.500 1.000 3-CQA 0.500 4-CQA 7 0.000 3 5-CQA 3,5 diCQA 4,5-diCQA

6 4

5

Figure 33.Radar plot of individual chlorogenic acid concentrations varying between seven varieties (average values taken within +/- 3 σ) and in comparison to non-EU samples. Concentrations are given on radial axis in g/100g dry leaf material. Outer numbers indicating the 7 varieties and non-EU samples; numbers inside the plot are indicating the average concentrations of individual chlorogenic acids.

1 3 2.5 Non-EU 2 2 1.5 1 0.5 monoCQA 7 0 3 diCQA

6 4

5

Figure 34.Radar plot of total mono- and di-acyl quinic acids (chlorogenic acids) concentrations varying between seven varieties (average values taken within +/- 3 σ) and in comparison to non- EU samples. Concentrations are given on radial axis in g/100g dry leaf material. Outer numbers indicating the 7 varieties and non-EU samples; numbers inside the plots are indicating the concentrations.

Hande Karaköse Jacobs University Bremen 59

Phytochemical Characterization of Stevia rebaudiana

Non-EU

7

6

5 di CQA 4 mono CQA 3

2

1

0.000 0.500 1.000 1.500 2.000 2.500 3.000

Figure35. Bar plot of total mono- and di-acyl quinic acids (chlorogenic acids) concentrations varying between seven varieties (average values taken within +/- 3 σ) and in comparison to non- EU samples. Concentrations are given on radial axis in g/100g dry leaf material.

Variations by Origin

Stevia is capable of growing almost anywhere under poor soil conditions, in particular where tobacco used to grown. There are only few plants possessing this feature. Therefore, stevia can serve as an alternative crop to the tobacco farmers and discourage tobacco cultivation inside EU.

Stevia rebaudiana was within this project cultivated in nine different locations (Figure 36) within the EU (e.g. Conaga,TCV, Italy; Granada, Spain; and Toumpa, Agrinio, Amfikleia, Greece). Additionally samples from outside the EU were available for comparison (e.g. Paraguay, Argenitine). Stevia leaves obtained from these origins were analyzed for studying the effect of growth conditions (e.g. sun, soil, and climate) on the metabolite profile.

According to the literature polyphenol concentrations are due to their physiological function as UV protection agents a direct function of growth altitude and climatic conditions, in particular sunshine hours. Accordingly variations of chlorogenic acid concentrations between different origins should be expected.

Hande Karaköse Jacobs University Bremen 60

Phytochemical Characterization of Stevia rebaudiana

Figure 36.Map showing the origins of stevia cultivation within the project.

Indeed the data reveal significant variations in CGA concentrations varying from 3.090 -1.637 g/100g for total monocaffeoyl quinic acids and 2.890 - 1.144 g/100g for dicaffeoyl quinic acids. EU cultivated Stevia rebaudiana shows concentrations of CGAs nicely sandwiched between extreme values at both ends observed in samples from outside the EU (e.g. highest for Argentinian samples with an average value of 2.890 g/100g dicaffeoyl quinic acids and APTTB and Portugal samples with a lowest average value of 1.448 g/100g and 1.144 g/100g respectively). Again a radar plot shown in Figure 37 was used to display variations between different origins. Average values are given in Table 6.

Hande Karaköse Jacobs University Bremen 61

Phytochemical Characterization of Stevia rebaudiana

TCV 4.000 Conaga Uconor 3.000 APTTB 2.000 Agrinion 1.000 mono CQA Amiflikeia 0.000 Toumpa di CQA

Turkei Portugal

Granada Amfilia Argentinie

Figure 37.Radar plot of total mono- and di-acyl quinic acids (chlorogenic acids) concentrations varying between all origins (average values taken within +/- 3 σ) and in comparison to non-EU samples. Concentrations are given on radial axis in g/100g dry leaf material.

A bar plot shown in Figure 38 allows further direct comparison between samples of different origins.

Conaga APTTB Amiflikeia Turkei Granada Argentinie di CQA Amfilia mono CQA Portugal Toumpa Agrinion Uconor TCV 0.000 0.500 1.000 1.500 2.000 2.500 3.000 3.500

Figure 38.Bar plot of total mono- and di-acyl quinic acids (chlorogenic acids) concentrations varying between all origins (average values taken within +/- 3 σ) and in comparison to non-EU samples. Concentrations are given on radial axis in g/100g dry leaf material.

Hande Karaköse Jacobs University Bremen 62

Phytochemical Characterization of Stevia rebaudiana

Table 6.Average values (taken within +/- 3 σ) for chlorogenic acids between origins

3CQA 5CQA 4CQA Total mono 3,5-diCQA 4,5-diCQA TotaldiCQA Origin (g/100g (g/100g (g/100g (g/100g (g/100g (g/100g (g/100g leaves) leaves) leaves) leaves) leaves) leaves) leaves) TCV 0.224 2.015 0.101 2.336 0.909 0.987 1.757 Uconor 0.323 2.060 0.119 2.575 1.373 1.480 2.079 Agrinion 0.278 2.078 0.104 2.604 1.233 1.339 2.460 Toumpa 0.269 2.450 0.123 2.780 1.351 1.255 2.471 Portugal 0.202 1.580 0.079 1.861 0.555 0.589 1.144 Amfilia 0.342 2.208 0.110 2.674 0.898 1.043 1.941 Argentinie 0.284 2.446 0.122 2.883 1.435 1.520 2.890 Granada 0.197 1.855 0.093 2.145 0.686 0.703 1.389 Turkei 0.104 1.837 0.092 2.093 0.727 0.859 1.586 Amiflikeia 0.441 2.646 0.133 3.090 1.302 1.254 2.335 APTTB 0.206 1.363 0.068 1.637 0.746 0.703 1.448 Conaga 0.310 2.525 0.126 2.956 0.993 1.410 2.328

Variation between harvests

Stevia rebaudiana was harvested three times within this project. From the data for example it can be seen that the average concentration of all dicaffeoyl quinic acids and monocaffeoyl quinic acids remains rather constant over first and second harvests, whereas a significant decrease can be observed in the third harvests for monocaffeoyl quinic acids. Overall, for individual compounds and for total amounts, second harvests are having the highest values 2.595g /100g for monocaffeoyl quinic acids and third harvests are for dicaffeoylquinic acids 2.242 g/100 g (Table 7). However, decrease in the quantities of CGAs with the later harvest could be observed if the three harvests are compared while the origin and the variety of stevia are kept constant (Table 8). In particular mono-CQAs are showing higher decreases from harvest I to harvest III, whereas diCQAs values are rather constant within chosen three origins with same variety presented in Table 8.

Hande Karaköse Jacobs University Bremen 63

Phytochemical Characterization of Stevia rebaudiana

Table 7.Average values (taken within +/- 3 σ) for chlorogenic acids between harvests

Total 3CQA 4CQA 5CQA 3,5-diCQA 4,5-diCQA TotaldiCQA monoCQA Harvest (g/100g (g/100g (g/100g (g/100g (g/100g (g/100g (g/100g leaves) leaves) leaves) leaves) leaves) leaves) leaves)

I 0.306 0.110 2.096 2.562 1.004 1.164 1.967

II 0.276 0.113 2.251 2.595 1.209 1.182 2.142

III 0.145 0.091 1.828 2.064 1.226 1.155 2.242

Table 8.Comparison of average values (taken within +/- 3 σ) for chlorogenic acids between three harvests of same variety and origin

Sample no. Origin Variety Harvest 3CQA 5CQA 4CQA Totalmono 3,5diCQA 4,5diCQA TotaldiCQA

71 Agrinion 3 I 0.413 3.266 0.163 3.843 1.008 1.282 2.290

123 Agrinion 3 II 0.247 2.626 0.131 3.004 1.570 1.342 2.913

162 Agrinion 3 III 0.168 1.372 0.069 1.608 0.974 0.938 1.912

98 Toumpa 3 I 0.491 4.099 0.205 4.795 1.208 1.535 2.743

127 Toumpa 3 II 0.175 2.490 0.124 2.789 1.650 1.271 2.921

126 Toumpa 3 III 0.187 1.961 0.098 2.246 1.684 1.567 3.251

137 TCV 3 I 0.441 3.153 0.158 3.752 1.578 1.266 2.844

139 TCV 3 II 0.267 2.692 0.135 3.094 1.854 1.155 3.010

161 TCV 3 III 0.161 1.777 0.089 2.026 0.987 1.077 2.064

Hande Karaköse Jacobs University Bremen 64

Phytochemical Characterization of Stevia rebaudiana

5.3.3.2. Flavonoid quantification

A selection of the two major flavonoids identified within the chromatogram was quantified as their glycosylated derivatives using calibration curves for closely related compounds quercetin- 3-glycoside and kaempferol-7-glycoside for all 166 samples. Additionally, for ten selected samples a flavonoid hydrolysis was carried out using acidic methanol with subsequent quantification of the four flavonoid aglycones quercetin, apigenin, luteolin and kaempferol (please refer to chapter 5.3.2 for identification). Relative values for flavonoids based on LC-MS data are contained within Tables 9 and 10. Data for the flavonoid hydrolysis are given in Table 11.

Table 9.Flavonoid glycosides average values for two major flavonoids in samples between origins determined by LC-MS directly from extracts without hydrolysis. Reference compounds used were quercetin-3-glucoside and kaempferol-7-glucoside. Total flavone value designates addition of all intensities of all flavonoid signal in chromatogram referenced to kaempferol-7- glucoside.

Kaempferol-7-glucoside Quercetin-3-glucoside Total flavones Origin (g/100g leaves) (g/100g leaves) (g/100g leaves) TCV 1.653 0.089 3.630 Uconor 2.762 0.103 6.397 Agrinion 3.431 0.047 7.865 Toumpa 3.602 0.063 8.170 Portugal 3.792 0.051 9.122 Amfilia 4.001 0.073 8.142 Argentinie 4.461 0.053 9.698 Granada 3.094 0.062 7.491 Turkei 2.885 0.088 7.444 Amiflikeia 2.956 0.066 6.984 APTTB 3.782 0.052 9.730 Conaga 2.506 0.101 5.726

Hande Karaköse Jacobs University Bremen 65

Phytochemical Characterization of Stevia rebaudiana

Table 10.Flavonoid glycosides average values between varieties

Kaempferol-7-glucoside Quercetin-3-glucoside Total flavones Variety (g/100g leaves) (g/100g leaves) (g/100g leaves)

1 2.860 0.073 7.224

2 2.090 0.099 5.401

3 2.955 0.062 7.412

4 3.095 0.104 6.640

5 2.904 0.076 8.083

6 2.850 0.045 7.682

7 2.164 0.155 5.432

Non-EU 3.064 0.097 7.638

Table 11.Values for flavonoids quercetin, kaempferol, luteolin and apigenin determined after hydrolysis of total polyphenol fraction using HCl/MeOH, determined by LC-MS. n.d. indicatesthat value was outside linear range of method140.

Sample Kaempferol Quercetin Luteolin Apigenin Origin Variety Harvest Year no (mg/100g) (mg/100g) (mg/100g) (mg/100g)

21 TCV 1 II 28.09.2010 82.5 75 12 4

32 Uconor 2 I 11.08.2010 n.d 78 8 3

41 Uconor 3 I 11.08.2010 70 638 9 n.d.

47 Toumpa 3 II 10.09.2010 108 463 n.d. 5

61 Amfilia 3 II 15.09.2010 90 455 n.d n.d.

87 Uconor 7 I 13.07.2011 190 737.5 14 n.d.

89 Uconor 4 I 2011 65 353 n.d. n.d.

94 Uconor 5 I 13.07.2011 88 417.5 n.d. n.d.

114 Uconor 6 II 2011 n.d 445 7 2

115 Conaga 3 I 2011 73 195 8 4

123 Agrinion 3 II 2011 73 318 n.d. n.d.

Hande Karaköse Jacobs University Bremen 66

Phytochemical Characterization of Stevia rebaudiana

If individual figures for flavonoids from LC-MS data are compared to those obtained by hydrolysis an obvious differences in values is apparent, even if the g/100 g dry leaves values are corrected for the increased molecular weight of the flavonoids. A direct comparison must be viewed with great care since firstly the literature hydrolysis method has been validated only for flavonoid glucosides. However, in the previous identification section we have shown that compounds present in stevia are hexosides but not glucosides, and can hence show a greatly reduced hydrolysis kinetic. Secondly, the quantitative data without hydrolysis are based on the use of reference standards that are chemically different from the compounds present in the leaf. Again it can be anticipated that ion enhancement effects in LC-MS might contribute to higher levels of flavonoids determined here. However, the values obtained are highly useful since they give a detailed insight into relative variations between flavonoids between different varieties and origins and give a guideline towards absolute and accurate values.

5.3.3.3. Principal Component Analysis (PCA)

A principle component analysis (PCA) based on the LC-MS dataset of stevia phenols was carried out to allow differentiation between different stevia varieties and geographic origins. Figure 39a shows a representative analysis with score and loading plots where within the scores plot every spot corresponds to a single leave sample with clear groupings apparent allowing distinction between varieties based on their phenolic profile. The loading plot reveals in each data point a pair of retention time and m/z ratio, therefore defining individual compounds whose quantities can be used as phytochemical markers for variety distinction.

A second PCA analysis was carried out with an aim to distinguish EU cultivated samples from non-EU cultivated samples. For this purpose 20 EU and 20 non EU samples were subjected to a full PCA analysis. Score and loading plots are shown in Figure 39b.

From the score plot it can be seen that the samples fall in three groupings. Group A from South American samples can clearly be distinguished from all other samples based on their high diCQA content (from score plot).

A second group B contains exclusively European samples from the Uconor cooperative. The final group C contains both EU and non-EU samples e.g. from Turkey, Ucraine and India, which

Hande Karaköse Jacobs University Bremen 67

Phytochemical Characterization of Stevia rebaudiana group together. The sample distinction information from the loading plot suggests that a combination of rebaudioside A concentrations and diCQA concentrations allows distinction here.

PC 4 PC 4

1 1

0 0

-1 -1

-1.0 0.0 1.0 PC 2 -1.0 0.0 1.0 PC 2

Figure 39a.PCA analysis of phenol profile of 35 stevia leaf LC-MS datasets. Score plot is on the left with each colour representing a different stevia variety and loading plot on the right with each data point corresponding to individual compounds allowing differentiation.

PC 2 PC 2

B 0.2 0.2 Reb A 3,5 diCQA 0.0 0.0

-0.2 A -0.2 C -0.4 -0.4

-0.5 0.0 0.5 1.0 PC 1 -0.5 0.0 0.5 1.0 PC 1

Figure39b.PCA analysis of phenol profile of 40 stevia leaf LC-MS datasets (red points non-EU samples, blue points EU samples). Score plot is on the left with each colour representing a different stevia variety and loading plot on the right with each data point corresponding to individual compounds allowing differentiation.

Hande Karaköse Jacobs University Bremen 68

Phytochemical Characterization of Stevia rebaudiana

5.3.4. Statistical Evaluation of Quantification Data of Polyphenols in Stevia

From the obtained data, a series of statistical analysis was carried out. For each variety, origin and harvest average values and standard deviations were determined. Additionally, the statistical pattern and type of statistical distribution of the data was analyzed for each subgroup. The correlation studies were performed by Pearson's correlation, with the significance value of p < 0.05. Significant differences among stevia leaves for each variable were assessed with analysis of variance (ANOVA).

5.3.4.1. Statistical Spread of Data

The distribution of the dataset is an essential step for examination of data in statistical analysis. The most important and useful distribution of data is Gaussian (normal) distribution. A normal distribution can be easily characterized by observing its symmetrical bell shaped curve on a histogram (Figure 40). Skewness and kurtosis values (< 1) show also that the data is normally distributed (Table 12). The Kolmogorov-Smirnov (KS) test was also used for the analysis of data distribution. In this test, the significance value above 0.05 means the data is normally distributed.

Each mono and di-CQAs as well as total mono and di-CQAs quantities obtained from 166 stevia samples showed normal distribution as judged by the KS test. In contrast quantitative data for cis-5 CQA and cis-4,5 diCQA were found to exhibit a non-Gaussian distribution. The KS test result, mean values, standard deviations, skewness and kurtosis of the curve for each CGA is represented in Table 12. Histogram of 5-CQA is presented as an example in Figure 40.

Figure 40.Histogram of 5-CQA, showing the normal distribution of the dataset.

Hande Karaköse Jacobs University Bremen 69

Phytochemical Characterization of Stevia rebaudiana

Table 12.Descriptive statistics of caffeoylquinic acids

Statistics

3-CQA 5-CQA 4-CQA 3,5-diCQA 4,5-diCQA Total diCQA Total monoCQA

Mean 0.295 2.488 0.124 1.201 1.237 2.435 2.907 Std. Deviation 0.144 0.852 0.426 0.512 0.485 0.937 1.009 Skewness 0.411 0.151 0.151 0.049 0.476 0.111 0.137 Kurtosis -0.217 0.365 0.372 -0.848 0.170 -0.513 0.212 Minimum 0.002 0.193 0.010 0.173 0.210 0.311 0.205 Maximum 0.719 4.986 0.249 2.476 2.609 4.575 5.608 KS test, Asymp sig. (2 tailed) 0.521 0.951 0.916 0.468 0.746 0.584 0.817

5.3.4.2. Correlations

The correlation analysis was performed to measure the degree of relation between two chosen CGAs. There are several different correlation methods, the most commons are Pearson (parametric) test, which is used in the case of normally distributed populations and the Spearman (nonparametric) test, in which there is no requirement for the assumption of normality or homogeneity of variance. The main result obtained from these tests is the correlation coefficient (r) and it ranges from -1.0 to +1.0. The closer “r” is to -/+1, the more closely the two variables are related.

Correlations of CGAs grouped as mono to monoCQA, diCQA to diCQA and as well as monoCQA to diCQAs were tested according to Pearson correlation. However, correlations of cis derivatives were tested according to Spearman correlation due to their non-Gaussian data distribution. Overall, correlation was observed for all CGAs with each other (Table 13). However, the strong correlation was observed between the mono-CQAs and 3,5-diCQA with 4,5-diCQA (Pearson correlation of 0.762 and 0.791 respectively) and 5-CQA with 4,5-diCQA as well as 4-CQA with 4,5-diCQA (Pearson correlation of 0.573 for both cases). The rest of the correlations were slightly weaker. It needs to be pointed out that, all correlation coefficients closer to +1, indicates that increasing quantity of e.g. 5-CQA leads to quantity increase in the 4,5-diCQA. The square of the correlation coefficient (r) is the percent of the variation in one variable which is related to the variation in the other. In the case of correlation of 3,5-diCQA with 4,5-diCQA 62% of the variance is related (or is correlated) (Figure 41).

Hande Karaköse Jacobs University Bremen 70

Phytochemical Characterization of Stevia rebaudiana

Table 13.Correlation coefficients of mono and di-CQAs Correlations 3-CQA 5-CQA 4-CQA 3,5-diCQA 4,5-diCQA Pearson Correlation (r) 1 0.762** 0.762** 0.416** 0.514** 3-CQA Sig. (2-tailed) 0.000 0.000 0.000 0.005 R2 0.581 0.581 0.173 0.264 Pearson Correlation (r) 0.762** 1 1.000** 0.532** 0.573** 5-CQA Sig. (2-tailed) 0.000 - 0.000 0.000 0.000 R2 0.581 - 1 0.283 0.328 Pearson Correlation (r) 0.762** 1.000** 1 0.532** 0.572** 4-CQA Sig. (2-tailed) 0.000 0.000 - 0.000 0.000 R2 0.581 1 - 0.283 0.327 Pearson Correlation (r) 0.416** 0.532** 0.532** 1 0.791** 3,5-diCQA Sig. (2-tailed) 0.000 0.000 0.000 - 0.000 R2 0.173 0.283 0.283 - 0.625 Pearson Correlation (r) 0.514** 0.573** 0.572** 0.791** 1 4,5-diCQA Sig. (2-tailed) 0.000 0.000 0.000 0.000 - R2 0.264 0.328 0.327 0.625 - ** Correlation is significant at the 0.01 level (2-tailed).

a) b)

Hande Karaköse Jacobs University Bremen 71

Phytochemical Characterization of Stevia rebaudiana

c) d)

Figure 41.Graph showing the correlation between a) 3,5-diCQA/4,5-diCQA b) 3-CQA/5-CQA, c) 5-CQA/4,5-diCQA and d) 4-CQA/3,5-diCQA.

Furthermore, another study was performed on whether the formation of cis-caffeoylquinic acids plays a role in agricultural practice and whether they would serve as a useful marker for UV exposure of plant tissues. Agricultural parameters including climatic conditions were obtained through records of the nearest official weather station.

To investigate the biosynthetic origin of the cis-CQA derivatives, initially the statistical pattern and type of statistical distribution of the quantitative data was analyzed. Both 5-CQA and 3,4 di- CQA concentration show a Gaussian distribution profile over all samples analyzed as judged by the Kolmogorov Smirnov test (section 5.3.4.1). In contrast quantitative data for cis-5 CQA and cis-3,4 diCQA were found to exhibit a non-Gaussian distribution.

For a large data set of quantitative data for three mono- and four di-CQA concentrations, correlation of all chlorogenic acids isomers correlate linearly with each other pointing to a common stimulus of biosynthetic production in the plant (Figure 41). The correlation linearity test was applied to cis derivatives of 5-CQA and 4,5-diCQ with their trans derivatives and there was no linear concentration dependency (Figure 42). This result indicates towards two distinct stimuli and pathways of their biosynthetic production.

In contrast concentrations of the two cis-4,5-diCQA isomers and concentration of cis-5-CQA and the cis-4,5 diCQA derivatives show a linear concentration dependency (Table 14 & Figure 42). This should be interpreted as all cis derivatives sharing the same external stimulus for

Hande Karaköse Jacobs University Bremen 72

Phytochemical Characterization of Stevia rebaudiana

production. This stimulus is different from that of trans-CQA biosynthesis. Due to that the stimulus of cis-CQA production is direct irradiation by UV light of exposed leaves (appendix E).

Figure 42.Linear dependency of cis-5-CQA with 5-CQA and two isomers of cis-4,5-diCQAs.

Table 14.Correlation coefficients of cis-isomers according to Spearman’s rule Correlations Cis-5CQA Cis-4,5diCQA1 Cis-4,5diCQA2 Correlation Coefficient 1.000 0.265** 0.183* Cis-5CQA Sig. (2-tailed) . 0.003 0.041 N 126 126 126 Correlation Coefficient 0.265** 1.000 0.731** Spearman's rho Cis-4,5diCQA1 Sig. (2-tailed) 0.003 . 0.000 N 126 126 126 Correlation Coefficient 0.183* 0.731** 1.000 Cis-4,5diCQA2 Sig. (2-tailed) 0.041 0.000 . N 126 126 126 ** Correlation is significant at the 0.01 level (2-tailed). * Correlation is significant at the 0.05 level (2-tailed).

In a dataset comprising 120 samples of Stevia rebaudiana grown in different locations, linear correlations of trans-mono and di-CQAs with each other indicates a common stimulus in production. However, non-linear correlation between cis and trans-CQAs and linear correlation

Hande Karaköse Jacobs University Bremen 73

Phytochemical Characterization of Stevia rebaudiana between the cis-CQAs convincingly demonstrates that the production of cis and trans CQA derivatives must follow two distinctly different pathways, where trans-CQAs production is not UV dependent stimulus.

To substantiate this hypothesis, available UV irradiation data with cis-CQA concentrations were correlated. As a first set of data, changes of cis-CQA concentrations in different harvests were studied. For two set of samples three harvests were carried out from June to September of crops grown at the same locations. For the last harvest in September the number of sunshine hours affecting the plants was always considerably lower if compared to the two early harvests from June to August. Sunshine hours were available through the weather databases of the nearest airport weather station (www.weather.org and www.weather.online.co.uk). Representative data are shown in Figure 43 where in two bar charts the total cis-CQA concentration is given for three harvests in two different locations.

mg/100g 400 350 300 250 200 150 100 50 0 Harvest 1 A Harvest 1B Harvest 2A Harvest 2B Harvest 3A Harvest 3B

Figure 43.Amount of 5-CQA in mg/100g dry leaves from three harvests from location A (TCV) and location B (Amfilikeia) during 2011.

Hande Karaköse Jacobs University Bremen 74

Phytochemical Characterization of Stevia rebaudiana

A more global picture including all available data was obtained by the log of trans/cis-5-CQA concentrations against the number of sunshine hours in the month prior to sample collection for six locations were in 2010 and 2011 a total of ten harvests were collected. The data plotted in this manner display a linear relationship with the quotient trans/cis-5-CQA concentration clearly depending on the number of sunshine hours. The more sunshine hours the plant leaves were exposed to the smaller quotient, so the higher the relative amount of cis-5-CQA concentrations (Figure 44).

log C(trans)/C (cis) vs sunshine hours in month 3

2.5

2

1.5

1

0.5

0 0 50 100 150 200 250 300 350 400

Figure 44.log of trans/cis-5-CQA concentrations against the number of sunshine hours in the month for a total of ten harvests from six locations.

Hande Karaköse Jacobs University Bremen 75

Phytochemical Characterization of Stevia rebaudiana

5.3.4.3. Analysis of Variance (ANOVA)

One way ANOVA was performed to study the influence of origin and, in another test, variety on the mono and di-CQA content of stevia leaves cultivated in various regions of EU with different soil and climatic conditions. Post-hoc test was applied to find out which origin or varieties differ, if any.

ANOVA is a statistical analysis used for identifying factors that are influencing a given dataset. It is a statistical technique for comparing mean values for multiple independent populations. ANOVA analysis selects the most discriminating variables by calculating an F factor which is proportional to the ratio of the ‘within-group’ variance to the ‘between groups’ variance. The higher this ratio, the more the groups are significantly different from each other141. The One-Way ANOVA, used in this study, compares the mean of one or more groups based on one independent factor. However, there are two assumptions to be met within the dataset. First one is, the data should be normally distributed (the data distribution was analyzed in chapter 5.3.4.1). The second assumption is that the variances of the groups to be compared should be similar. This can be checked by Levene’s test (test of homogeneity of variances). If the significance value is greater than 0.05 (found in the Sig. column, e.g. Table 15) then there is homogeneity of variances and therefore, the assumption of homogeneity of variance is met. If the Levene's test is significant (lower than 0.05), it would mean that there is no similar variances and the assumption of homogeneity of variance is not met, therefore it would be necessary to use an adjusted test such as the Welch statistic (the Robust Tests of Equality of Means) within ANOVA. In general, when setting up the analysis, it is common and advantageous to select a test for either situation since we do not know if the assumption is met or violated.

Influence of Origin: Levene’s test was performed to assess whether the assumption of homogeneity of variance between groups is met within the data. Test of homogeneity results revealed that for dependent variable 3-CQA and 4,5-diCQA

Levene’s test is not significant (F8/116 = 1.783, p=0.087), (F8/116 = 1.430, p=0.191),respectively. However, for 5-CQA, 4CQA and 3,5-diCQA Levene’s test was significant at the level of 0.05 (Table 15), which states that the variances in the different groups of origins are different (the groups are not equal variance on the dependent variable, that is origin in this case) and therefore,

Hande Karaköse Jacobs University Bremen 76

Phytochemical Characterization of Stevia rebaudiana a modified procedure which do not assume equality of variance (Welch statistic) was also applied to these data along with ANOVA.

Table 15. Results of test of homogeneity of variances

Levene df1 df2 Sig. Statistic cqa3 1.783 8 116 0.087 cqa5 2.058 8 116 0.046 cqa4 2.073 8 116 0.044 totalmono 2.146 8 116 0.037 diCQA35 2.018 8 116 0.050 diCQA45 1.430 8 116 0.191 totaldi 1.733 8 116 0.098

*df: degrees of freedom

ANOVA results (Table 16) did not reveal significant differences between the average CGA content of stevia leaves of different origins. The data only hints at some differences in the average content of stevia leaves from different origin destinations. These, however, can be considered only marginally significant at the 5% level (p= 0.054 for e.g. 4-CQA). Moreover, from the ANOVA output, robust test of equality of means (Welch test) is considered for the CQAs that are not meeting the assumption of homogeneity of variance. Like the ANOVA test, if the significance value is less than 0.05 in Welch test, then it means there are statistically significant differences between groups.

Hande Karaköse Jacobs University Bremen 77

Phytochemical Characterization of Stevia rebaudiana

Table 16.ANOVA results for effect of origin on stevia CGA content

Robust Test of Equality of Means One Way Anova Compound F Sig. (p) Compound F Sig. (p) 3-CQA 1.022 0.138 3-CQA Welch 1.823 0.132 5-CQA 1.850 0.054 5-CQA Welch 1.588 0.183 4-CQA 1.853 0.054 4-CQA Welch 1.590 0.182 Totalmono-CQA 1.985 0.064 Totalmono Welch 1.960 0.101 3,5-diCQA 1.767 0.106 3,5-diCQA Welch 3.135 0.017

4,5-diCQA 1.727 0.127 4,5-diCQA Welch 2.666 0.034

Total-diCQA 1.881 0.089 Total-diCQA Welch 3.120 0.017

From the results of Welch test and One-Way ANOVA, the conclusion would be that at least two of the group means is significantly different from the each other. Beyond this, it is necessary to conduct a post-hoc comparison test to see exactly which pairs of groups are significantly different. There are varieties of post hoc tests available. The most commonly used LSD test is applied for CQAs (3CQA, 4,5-diCQA) which are meeting the assumption of homogeneity of variance. Games-Howell post hoc test is applied for the CQAs (5CQA, 4CQA, 3,5diCQA) violating the assumption of homogeneity of variance (equal variances not assumed). Multiple comparison (post hoc) results obtained from LSD test (Fisher's Least Significant Difference) revealed that the 3-CQA amounts are significantly different at the 0.05 level between the origins of Agrinion&Granada, Granada&TCV, Granada&Uconor and Granada & non-EU origins. 4,5-diCQAs resulted to be significantly different at the 0.05 level for origin of Granada and the all other origins except Portugal.

Multiple comparison (post hoc) results obtained from Games-Howell test revealed that there is no significant difference at the 0.05 level between the origins for 5-CQA and 4-CQA. However, significant difference observed for 3,5-diCQA between the origins of Amiflikeia and Toumpa with the significance value of 0.008.

Hande Karaköse Jacobs University Bremen 78

Phytochemical Characterization of Stevia rebaudiana

Influence of Variety: Test of homogeneity results revealed that for dependent variable 3-CQA

(F7/117 = 1.207, p=0.304), 3,5-diCQA (F7/117 = 0.545, p=0.799) and 4,5-diCQA (F7/117 = 1.193, p=0.312) Levene’s test was not significant. However, for 5-CQA and 4-CQA Levene’s test was significant at the level of 0.05 (p=0.018) (Table 17). Same as in the analysis of origins, Welch test was also applied to data for the cases of violation of homogeneity of variances.

Table 17. Test of homogeneity of variances

Levene Statistic df1 df2 Sig.

cqa3 1.207 7 117 0.304 cqa5 2.542 7 117 0.018 cqa4 2.549 7 117 0.018 totalmono 2.242 7 117 0.036 diCQA35 0.545 7 117 0.799 diCQA45 1.193 7 117 0.312 totaldi 1.090 7 117 0.374

*df:degrees of freedom

ANOVA results did not reveal significant differences between the average CGA content of stevia leaves of different varieties. Moreover, Welch test did not reveal any significant difference at the level of 0.05. The results of ANOVA test and Welch test is presented in Table 18.

Table 18.ANOVA results for effect of variety on stevia CGA content

Robust Test of Equality of Means One Way ANOVA Compound F Sig. (p) Compound F Sig. (p) 3-CQA Welch 0.926 0.500 3-CQA 0.849 0.549 5-CQA Welch 0.428 0.878 5-CQA 0.459 0.862 4-CQA Welch 0.433 0.874 4-CQA 0.464 0.859 Totalmono-CQA 0.463 0.860 Totalmono Welch 0.373 0.911 3,5-diCQA 1.901 0.075 3,5-diCQA Welch 1.637 0.161 4,5-diCQA 1.446 0.194 4,5-diCQA Welch 1.250 0.305 Total-diCQA 1.577 0.149 Total-diCQA Welch 1.402 0.238

Hande Karaköse Jacobs University Bremen 79

Phytochemical Characterization of Stevia rebaudiana

Post-hoc tests with Games-Howell method resulted with no pair-wise differences for 5-CQA and 4-CQA for each variety. However for 3-CQA with the LSD test, the mean difference between variety 1-5 and variety 3-5 was significant at the 0.05 level (p=0.04). The varieties 4-6, 5-7 and 6-7 had significant difference (p=0.04) when 4,5-diCQA quantity in stevia leaves was taken as a dependent variable. Variety 6 showed significant difference with varieties 2, 3, 4 and 7 for 3,5- diCQA quantity in stevia leaves.

5.4. Conclusion

Mono-CQAs and di-CQAs including their corresponding cis-isomers and flavonoid glycoside of 120 samples (after removal of outliers and duplicates) of Stevia rebaudiana grown in different locations and with different botanical varieties were profiled and quantified successfully. For the first time a full quantitative data set from a large number of samples from different origins and varieties, in which the full profile of all secondary metabolite quantities was determined in any agricultural plant was obtained.

With the obtained dataset, differences between stevia leaves and influence of the origin of cultivation and the botanical varieties on the CQAs profile were studied successfully with the most common used statistical method ANOVA. Pair-wise comparisons of varieties and origins for each CQA were achieved by less conservative statistical post hoc test (LSD) and Games- Howell post hoc test to determine exact pair of variety/origin that are differentiating from each other.

Correlation studies showed that the production of cis and trans CQA derivatives must follow two distinctly different pathways and the stimulus of cis-CQA production is direct irradiation by UV light exposed leaves. Therefore, cis-CQA concentrations may serve as useful markers of UV exposure of plant material in agricultural practices.

Finally, with the PCA, it was possible to differentiate the EU and non-EU cultivated stevia leaves according to their rebaudioside A and diCQAs profile.

Hande Karaköse Jacobs University Bremen 80

Phytochemical Characterization of Stevia rebaudiana

6. LIPID ANALYSIS of STEVIA 6.1. Overview

Lipid profile of stevia leaves was determined for 46 chosen samples by GC-MS. Total lipid amounts of each sample were determined gravimetrically after chloroform extraction. Identification was achieved for the fatty acids by comparison of retention times and mass spectra with the commercially obtained fatty acid methyl esters (FAME) reference mixture. This study was complemented by a MALDI-TOF-MS anaylsis for further lipid identification. Additionally, a steam distillation extract was subjected to GC-MS analysis for the analysis of volatile terpenes. Identification of the terpenes was achieved by software based NIST library search.

6.2. Materials & Methods 6.2.1. Extraction

Two grams of stevia leaves was immersed in liquid nitrogen, ground in a hammer mill, and extracted with 150 mL of chloroform in a Soxhlet apparatus (Buchi B-811 extraction system) for 2 h. The solvent was removed in vacuo and total lipid amount was determined gravimetrically.

6.2.2. Methyl Ester Formation

Total lipid extract was dissolved in 2 mL of chloroform after gravimetric determination. 200 μL of methanolic potassium hydroxide solution (2 mol/L) and 1 g of sodium hydrogen sulfate monohydrate (NaHSO4) were added to 1 mL of lipid extract solution in chloroform to form the methy esters of lipids for GC analysis.

6.2.3. GC-FID

GC analyses were performed on GC-2010 (Shimadzu, Kyoto, Japan) equipment with flame ionization detector and split/splitless injector. Injector temperature was at 290 °C and samples were injected using autosampler (1 µL) with split ratio of 1:10. Capillary columns was used Rxi - 5 ms (15 m × 0.25 mm, with film thickness of 0.25 µm) Restek. The temperature program was raised from 80 °C (1min) up to 300 °C at rate 5°C/min, and the total run time was 50 mins. Helium was used as carrier gas at flow rate of 5 mL/min. Detector temperature was set at 310 °C. To form the flame, hydrogen gas flow, 40 mL/min, and air gas flow, 400 mL/min, were used. GC solution 2.10 software was used for data collection, and calculation of all parameters.

Hande Karaköse Jacobs University Bremen 81

Phytochemical Characterization of Stevia rebaudiana

6.2.4. GC-MS

GC-MS analyses have been carried out with a Varian CP-3800 gas chromatograph (Palo Alto, CA, USA) equipped with a split/splitless injector and coupled to a Saturn 2000 ion-trap Varian mass spectrometer (MS). Data acquisition was performed using a Star Toolbar system (Varian). Samples were injected manually with split ratio of 1:10 at 290 °C. The compounds were analyzed on a 30 m, 0.25 mm I.D. fused-silica capillary column coated with a 0.1µm layer of poly (5% phenyl/95% dimethylpolisiloxane) (Rxi-5Sil MS, Restek) using helium as the carrier gas at flow rate of 1.3 mL/min, respectively. The oven temperature was heated from 80 °C (1 min) to 300 °C at the rate 5 °C/min and the total run time was 50 min. For the MS, the electron multiplier was set to 1350 V and ionization was accomplished by electron impact (EI). The transfer line temperature was set at 300 °C whilst 244 °C and 120 °C were the temperatures used for the trap and the manifold, respectively. Mass spectra were recorded from m/z 40–600.

6.2.5. Calibration Curve of FAME

The quantitative analyses have been initiated by generation of calibration curves for FAME standard mixture from the range of C14:0/C14:1 – C24:0/C24:1. Calibration curves were generated by plotting peaks areas of FAME standard mixture (Marine Oil FAME Mix (20 components) from Restek) at different concentrations as function of peak areas. In doing so, FAME standard solution was diluted with n-heptane to a series of standards with concentrations of 5, 50, 100, 250, 500, 750, 1000, 1250, 1500, 2000, 3000 μg/L.

6.2.6. MALDI-TOF MS

As matrix solution 5g/L 2,5-dihydroxybenzoic acid (2,5-DHB) solution in acetonitrile containing 0.1% trifluoroacetic acid (TFA) was used due to robustness of DHB to impurities. 1 μl aliquot of the organic extracts of stevia was mixed with 1 μl of the matrix solution on the maldi target (MPT AnchorChip 600-384 target, Bruker Daltonics) and the matrix crystals were allowed to air- dry.

MALDI-TOF spectra were acquired on an Autoflex II MALDI-TOF-TOF mass spectrometer (Bruker Daltonics) equipped with a 337 nm nitrogen laser. The instrument was operated in the reflector mode: source, 19.00 kV; lens, 8.95 kV; and reflector, 20 kV, using an optimized ion extraction delay time of 80 ns. The laser frequency was set at 25 Hz with 50 laser shots per

Hande Karaköse Jacobs University Bremen 82

Phytochemical Characterization of Stevia rebaudiana acquisition. The laser strength was kept about 40% above threshold to obtain optimum signal to noise ratio. Spectra were obtained by summing, on average, 200 laser shots. Spectra were acquired in the mass range 0–2500 amu. The instrument was externally calibrated in the enhanced quadratic calibration mode prior to acquisition using a peptide tune-mix sample (Bruker Daltonics).

6.3. Results & Discussion 46 Stevia rebaudiana samples were chosen for lipid analysis. Firstly, total lipids were determined gravimetrically after extraction with a non-polar solvent. Since hexane extraction provided values below 1 weight %, a Soxhlet chloroform extraction was selected for this purpose. The total lipid sample was in a following step subjected to GC-MS analysis. Table 19 shows total lipid values obtained for 46 representative samples comprising at least three samples from all seven varieties and samples from all origins. Secondly the total lipid fraction was subjected to basic hydrolysis and derivatisation followed by GC-MS analysis to identify the individual fatty acid spectrum of all seven stevia varieties and of representative samples from all origins. Derivatisation chosen included the formation of fatty acid methylesters. Identification of fatty acid methylesters was achieved through GC-MS by comparison of retention time and mass spectra with a commercial certified reference compound mixture. A representative GC chromatogram is shown in Figure 45 for total lipid profile of stevia after chloroform extraction and in Figure 46, the GC chromatogram of FAME standard mixture can be observed142, 143. Figure 47 presents the fragmentation of fatty acid by electron impact ionization obtained in GC- MS measurement of stevia leaves chloroform extract. Additionally a steam distillation extract was obtained and subjected to GC-MS analysis to profile the volatile terpenes and to compare the lipid composition with chloroform extract.

Hande Karaköse Jacobs University Bremen 83

Phytochemical Characterization of Stevia rebaudiana

Figure 45.GC.MS chromatogram of total lipid extracts from Stevia rebaidiana leaves from sample (Uconor, Var.4).

C16:1

C18:1 C18:1 C18:2 C18:3

C18:0 C20:0 C20:4 C20:1 C20:3 C14:1 C20:2

C20:5 C16:0 C22:0

C14:0 C22:1 C22:6

Figure 46.GC-MS chromatogram of FAME standard mixture.

Hande Karaköse Jacobs University Bremen 84

Phytochemical Characterization of Stevia rebaudiana

Figure 47.Representative EI-MS spectra obtained from GC-MS measurement of stevia extract.

Hande Karaköse Jacobs University Bremen 85

Phytochemical Characterization of Stevia rebaudiana

Table 19.Total lipid values in weight % from 46 samples Sample no Origin Variety Harvesting Year % 119 Agrinion 4 II 2011 3.42 120 conaga 6 I 2011 7.71 121 Conaga 7 I 2011 4.19 122 Agrinion 6 II 2011 5.14 123 Agrinion 3 II 2011 5.45 124 Toumpa 7 I 2011 3.05 125 Toumpa 4 III 2011 4.43 126 Toumpa 3 III 2011 6.67 127 Toumpa 3 II 2011 5.59 128 Agrinion 4 II 2011 0.50 129 Toumpa 4 II 2011 4.32 132 Agrinion 3 II 2011 5.92 133 Toumpa 5 II 2011 4.26 134 Toumpa 6 II 2011 4.15 135 Toumpa 7 II 2011 3.92 136 TCV 5 I 30.06.2011 1.68 137 TCV 3 I 30.06.2011 5.48 138 TCV 6 I 30.06.2011 5.16 139 TCV 3 II 11.08.2011 7.23 140 TCV 6 II 17.08.2011 4.90 141 TCV 4 II 24.08.2011 4.09 142 TCV 7 I 18.08.2011 2.11 143 TCV 4 I 07.07.2011 3.42 144 TCV 5 II 17.08.2011 8.23 147 Amiflikeia 5 I 2011 3.25 148 Amiflikeia 6 I 2011 4.23 149 Amiflikeia 1 I 2011 4.27 154 Toumpa 5 I 2011 3.24 8 TCV 4 I 04.08.2010 1.52 49 Toumpa 3 I 30.07.2010 3.37 43 Agrinion 2 II 20.09.2010 2.63 75 Granada 3 I 09.09.2010 3.41 54 Amfilia 4 II 15.09.2010 2.71 30 Uconor 4 II 07.05.2010 0.54 53 Amfilia 4 I 04.08.2010 2.99 5 TCV 1 I 04.08.2010 0.89 16 TCV 2 I 04.08.2010 0.98 19 TCV 3 II 10.09.2010 6.74 6 TCV 2 I 04.08.2010 5.11 10 TCV 2 II 28.09.2010 7.05 14 TCV 2 I 04.08.2010 2.27 32 Uconor 2 I 11.08.2010 3.14 43 Agrinion 2 II 20.09.2010 3.69 37 Uconor 4 I 11.08.2010 2.60 34 Uconor 4 II 07.09.2010 5.13 36 Uconor 4 I 11.08.2010 1.46

Hande Karaköse Jacobs University Bremen 86

Phytochemical Characterization of Stevia rebaudiana

For determination of the fatty acid profile GC-FID was applied using the GC-MS data as a reference point. Oleic acid was found to be the predominant fatty acid in stevia with an average ratio of saturated to unsaturated fatty acids of 1:3.8. Among the saturated fatty acids 16:0 palmitic acid was the predominant acid with varying quantities of stearic acid present (24 - 2 %). Saturated fatty acids with shorter and longer chain length were not identified. In terms of unsaturated fatty acids 16:1 was identified next to three co-eluting isomers of 18:1, with oleic acid being the predominant compound. Values stated for 18:1 acids represent a sum over all three isomers. A series of polyunsaturated fatty acids were identified as well comprising 16:2, 18:3 and 18:2 acids with variable quantities. Values are given in Table 20.

Table 20.Quantities of polyunsaturated fatty acids Sample Origin Variety Harvest Year % 16:2 %16:1 %16:0 %18:3 % 18:2 % 18:1 % 18:0

S10 TCV 2 II 28.09.2010 15.45 9.97 12.04 15.56 10.25 35.09 1.64 S6 TCV 2 I 04.08.2010 20.41 20.52 11.42 2.51 7.57 34.59 2.98 S14 TCV 2 I 04.08.2010 22.03 14.73 9.50 13.49 6.87 30.39 2.99 S34 Uconor 4 II 07.09.2010 9.29 1.71 12.26 6.00 11.29 52.06 7.39 S37 Uconor 4 I 11.08.2010 3.54 2.10 14.70 8.41 12.44 55.97 2.84 S5 TCV 1 I 04.08.2010 2.09 0.72 14.59 4.70 13.04 57.64 7.23 S10 TCV 2 II 28.09.2010 14.56 9.02 13.01 11.90 10.32 36.11 5.09 S16 TCV 2 I 04.08.2010 4.00 0.51 14.56 5.50 12.61 55.03 7.78 S19 TCV 3 II 10.09.2010 1.21 0.57 12.57 8.35 12.77 48.69 15.83 S32 Uconor 2 I 11.08.2010 2.18 1.11 14.86 6.80 11.80 53.53 9.71 S36 Uconor 4 I 11.08.2010 5.15 3.38 14.18 9.03 13.75 46.55 7.96 S43 Agrinion 2 II 20.09.2010 1.49 0.39 11.12 10.81 10.28 49.80 16.11 S45 Portugal 1 I 07.07.2010 1.42 0.68 10.35 11.98 10.40 40.32 24.86 S121 Conaga 7 I 2011 7.91 4.77 14.20 6.00 13.59 47.71 5.82 S122 Agrinion 6 II 2011 8.44 5.68 15.47 4.10 9.43 52.31 4.57 S123 Agrinion 3 II 2011 10.69 6.36 13.41 8.63 9.44 48.57 2.90 S124 Toumpa 7 I 2011 0.93 0.46 21.06 5.77 16.42 54.17 1.20 S127 Toumpa 3 II 2011 9.05 5.67 16.23 6.39 13.06 43.56 6.03 S136 TCV 5 I 30.06.2011 1.11 0.61 14.03 6.84 10.42 52.11 14.88 S143 TCV 4 I 07.07.2011 8.05 5.38 14.04 4.43 11.29 50.25 6.56 S147 Amiflikeia 5 I 2011 8.27 5.48 15.26 8.71 8.41 45.93 7.94 S148 Amiflikeia 6 I 2011 10.93 7.22 15.29 7.20 9.95 42.27 7.13 S149 Amiflikeia 1 I 2011 7.26 4.38 16.83 4.95 10.70 49.61 6.27 S154 Toumpa 5 I 2011 0.63 0.39 15.93 4.80 11.48 60.90 5.86 S120 Conaga 6 I 2011 1.71 0.55 14.79 5.37 11.58 58.93 7.07 S142 TCV 7 I 18.08.2011 2.23 1.31 17.64 6.13 15.83 51.55 5.30 Average 6.93 4.37 14.21 7.48 11.35 48.22 7.46

Hande Karaköse Jacobs University Bremen 87

Phytochemical Characterization of Stevia rebaudiana

It should be noted that in plants fatty acid composition is reported to be a consequence of climatic conditions, in particular temperature, rather than botanical variation. The total lipid content determined is well in line with values from other leafy dietary plants. For example values for green tea has been determined as 3-5% total lipids, spinach for 4.5 % total lipids and the botanically related Asteraceae plant lettuce at 4-6 % total lipids. The average fatty acid distribution and their structures are shown in Figure 48 and Figure 49.

O O

OH OH

16:0 Palmitic acid (hexadecanoic acid) 18:0 Stearic acid (octadecanoic acid)

O O

OH OH

16:1 Palmitoleic acid (hexadec-9-enoic acid) 18:1 Oleic acid (octadec-9-enoic acid)

O HO O

HO

16:2 9,12-Hexadecadienoic acid 18:2 Oleic acid (9,12 - octadecadienoic acid)

O OH

18:3 Linoleic acid (9,12,15 - octadecatrienoic acid)

Figure 48.Structures of fatty acids in Stevia rebaudiana extract.

Hande Karaköse Jacobs University Bremen 88

Phytochemical Characterization of Stevia rebaudiana

60

50

40

30

20

10

0 % 16:2 %16:1 %16:0 %18:3 % 18:2 % 18:1 % 18:0 % X:Y

Figure 49.Fatty acid profile of average stevia leaf in %. X:Y denominates the number of carbon atoms in the fatty acid (X) and the number of double bonds in the fatty acid (Y). Data were obtained from 46 samples after lipid hydrolysis, derivatisation and GC-MS analysis

Average fatty acid and lipid values determined were as follows:

• Total lipids: 4.52 % (0.89-8.23 %) • Ratio saturated/unsaturated fatty acids: 1:3.8 (1:6-1:1.8)

This work was complemented by a MALDI-TOF-MS anaylsis of the lipid fraction allowing identification of selected intact lipids, mainly triacylglycerides of oleic acid. A representative MALDI-MS spectrum is shown in Figure 50.

Hande Karaköse Jacobs University Bremen 89

Phytochemical Characterization of Stevia rebaudiana

Intens. 0_O20\1: +MS [%] 872.3 100 Ether type lipid: 1,2-dioleoyl-3-oleylglycerol (C57H106O5)

H2C O (CH2)8 CH CH (CH2)7 CH3

478.8 80 229.4 HC O CO (CH2)7 CH CH (CH2)7 CH3

H2C O CO (CH2)7 CH CH (CH2)7 CH3

60

361.5

40

593.9 20

326.7

533.8

0 200 400 600 800 1000 1200 1400 1600 1800 2000 m/z

Figure 50.MALDI-MS spectrum of total lipid extract in positive ion mode using 2,5-DHB as a matrix. Main peak at m/z 872.3 corresponding to trioleic acid glyceride.

Finally direct GC-MS analysis of the lipid fraction was carried out supplemented by a GC-MS analysis of a steam distillation extract of Stevia rebaudiana leaves. This analysis allowed positive identification of twelve mono- and sesquiterpenes using a NIST library search. The NIST library search identification of compounds was made if the database spectrum showed a match with the experimental spectrum with a NIST score of 800 or above. Additionally a library A database search was carried out and compounds accepted if a library A score of above 25 % and a NIST score of above 800 was observed. A NIST score of above 800 is generally accepted in the literature for positive compound identification. All compounds identified were additionally present in both steam distillation extract and total lipid fraction. Selected structures identified are shown below in Figure 51 and retention times with NIST score for few terpenes are presented in Table 21.

It should be noted that previously an additional 25 volatile terpenes have been reported in Stevia rebaudiana leaves. These compounds are shown in the appendix F, but their presence could not be confirmed in this study144.

Hande Karaköse Jacobs University Bremen 90

Phytochemical Characterization of Stevia rebaudiana

O O

limonene sabinene terpinene myrtenal 1,8 cineol alpha pinene -pinene

O

cumin aldehyde 3-carene copaene humulene gamma cadinene (caryophyllene)

Figure 51.Chemical structures of terpenes identified in Stevia rebaudiana leaves.

Table 21.Retention time and NIST scores of some terpenes identified in stevia extract

Terpene RT (min) NIST score m/z sabinene 9.5 838 136

α-pinene 10.4 920 136

3-carene 10.3 925 136 caryophyllene 10.5 727 204

γ-cadinene 11.2 887 204

copaene 12.8 848 204

In addition to the mono- and sesquiterpenes the GC-MS data show a series of triterpenes at longer retention times. NIST search clearly indicates their identity as triterpenes with low NIST scores for steroid and related structures, however, no match in the database allowed positive compound identification. None of the terpenes identified was reported to show a toxicologically problematic profile. Compounds have been reported in many additional dietary plants. General volatile terpene levels in the plant are very low. The total FID or TIC (total ion current in MS

Hande Karaköse Jacobs University Bremen 91

Phytochemical Characterization of Stevia rebaudiana data) are around 3-5 % of the total intensity of the lipid fraction, which allows an estimate 1.5 - 2.5 × 10-3 weight % of total volatile lipids.

Additionally the GC-MS data revealed the presence of giberillic acid (structure below) a well known plant hormone in the lipid fraction. To confirm its existence, methylester formation of reference compound of steviol (having the closest chemical structure) was performed and subjected to GC-MS analyses (Figure 52). The retention times and mass spectral data were compared between stevia extract and the steviol standard. Furthermore, increase in the intensity was observed after standard steviol addition to the stevia extract.

OR

CH3

RO2C CH3

steviol

O OH OC HO COOH

giberillic acid

Figure 52.GC chromatogram of methylesterified steviol and stevia extract.

Hande Karaköse Jacobs University Bremen 92

Phytochemical Characterization of Stevia rebaudiana

Further secondary metabolite search

The groups of Nash and Fleet recently reported on the isolation of novel indolizidine alkaloid they named steviamine from Stevia rebaudiana leaves.145 Within the LC-MS data available ions corresponding to its m/z ratio was searched, however, an ion corresponding to this mass at a significant level was not found in any sample analyzed in neither positive nor negative ion mode (above S/N 20). The absence of the ion can be confirmed later on in future studies with an authentic reference samples requested from the Fleet group.

N HO HO OH Steviamine (not found)

6.4. Conclusion

Lipid profile and quantification data for chosen stevia samples were successfully obtained. Steam distillation and chloroform extracts of stevia leaves analyzed on GC-MS and GC-FID did not have significant differences in lipid profile. Furthermore, volatile terpenes were identified by NIST library search.

Hande Karaköse Jacobs University Bremen 93

Phytochemical Characterization of Stevia rebaudiana

7. PROTEOMICS of STEVIA 7.1. Overview

The main aim of stevia leave protein analysis was to purify, separate and sequence stevia leaf proteins with an aim to identify potentially allergenic proteins. Isolation and purification of stevia leaf proteins were achieved using 2D gel electrophoresis. The resulting 2D gels were stained with coumassie blue staining to visualise and identify individual proteins. Susbsequent trypsin digestion and MALDI-TOF MS analysis was carried out on selected proteins. For MALDI-MS sample preparation anchor chip MALDI targets were used in conjunction with 2,5- DHB (2,5-dihydroxy benzoic acid) matrices. Using data base search algorithms the obtained mass spectrometrical data were used in attempt to sequence the proteins.

7.2. Materials & Methods

7.2.1. Extraction of Proteins

Fresh plant tissues were crushed in pestle and mortar in presence of liquid nitrogen in order to prevent the degradation of protein by the release of protease enzyme. The leaves were crushed in fine powder, and to precipitate the proteins, the powder was suspended in TCA extraction buffer at -20 oC for overnight (Table 22). The proteins precipitated as white flakes after the overnight incubation. The supernatant is collected and centrifuged at 5000g for 30 min. The supernatant was discarded, and the protein pellets settled at bottom was washed with ice cold acetone and centrifuged again at 5000g for 15 min. The washing step was repeated for 3 to 4 times and then this protein pellet was dried by passing nitrogen gas at slow stream, after the drying process these pellets can be stored at -80°C for few months for later identification processes (Figure 53).116

Table22. Amount and properties of chosen stevia leaves

Sample Tissue Extraction buffer Approx.protein weighed Harvest Variety Origin Number (g) (mL/g) after extraction (mg) 8 I 4 TCV 10 20 20 10 I 2 TCV 10 20 15 36 I 4 Uconor 10 20 12 7 II 7 TCV 10 20 15

Hande Karaköse Jacobs University Bremen 94

Phytochemical Characterization of Stevia rebaudiana

Figure 53.Extraction procedure of proteins. Red arrow shows the process of TCA extraction and Blue arrow follows the phenol extraction116.

7.2.2. Protein Analysis SDS-PAGE Protocol: The protein pellet (200μg) was resuspended in IEF buffer/sample buffer (125 μL for 7 cm strip). 100 μL of IEF buffer with the protein sample was vortexed for 10 min at 10000 rpm and the supernatant was collected. The remaining volume of 25 μL of IEF buffer was added on the remaining pellet and vortexed for 20 mins and centrifuged at 10000 rpm. The supernatants were collected and added to the previous collected supernatant. The collected supernatant (IEF buffer with protein sample) was added in the rehydration tray uniformly and then the IPG strip (strip for isoelectric focusing, ph range was 3 - 10 and pH 4 - 7) was placed in the rehydration tray and it was kept overnight for sample absorption in to the strip.

Hande Karaköse Jacobs University Bremen 95

Phytochemical Characterization of Stevia rebaudiana

For running the IPG strip, the voltage gradient of 250 volt for 30 min, and then to 3500 volts for 4 hours was used. Equilibration buffer I: 6M urea, 0.375M Tris-HCl pH8.8, 2% SDS, 20% Glycerol, 2%(w/v) DTT.

Equilibration buffer II : 6M urea, 0.375M Tris-HCl pH8.8, 2% SDS, 20% Glycerol, 2.5%(w/v) iodoacetamide.

Sample buffer: CHAPS (Roche), Pharmalyte (pH 3-10) (GE Healthcare Lifesciences), dithiothreitol (DTT) (Sigma-Aldrich), Serdolit MB-1 (Serva), urea, Pefabloc® (VWR), Thiourea (Fluka)146 Rehydration buffer was purchased from Biorad ReadyPrep™ 2D starter kit.

2D SDS-PAGE Protocol: The gel casting chamber was filled from the bottom to a height of about 2 cm below the top of the glass plates with separation gel and stacking gel (Table 23) on the top. The gels were carefully overlaid with 1.0-1.5 mL buffer-saturated 2-butanol to allow for complete polymerization. In order to ensure good contact between the IPG strip and the gel an agarose solution was added. The agarose solution was kept at 70 °C and added first on top of the gel. Immediately after, the equilibrated strip was placed on the gel. The IPG strip gel was then subjected to electric field at 121volt and 45Amp for 1 to 2 hours for separation of proteins according to their molecular weight. The gel was stained overnight in Coomassie brilliant blue and later on destained with dd H2O. The protein bands were excised for destaining and trypsin digestion147.

Table 23.Preparation of separation and stacking gel for 2D SDS PAGE

A. 12.5 % Separating Gel B. 4% Stacking gel Water 3.3 ml Water 6.1 ml

30% Acrylamide 1.3 ml 30% Acrylamide 4.2 ml 1.5M Tris (pH 8.8) 2.5 ml 0.5M Tris (pH 6.8) 2.5 ml 10% SDS 100 μl 10% SDS 100 μl 10% APS 50 μl 10% APS 50 μl TEMED 5.0 μl TEMED 10 μl

Hande Karaköse Jacobs University Bremen 96

Phytochemical Characterization of Stevia rebaudiana

Destaining gel pieces excised from Coomassie stained 2D SDS gel: The protein bands were excised from 2D-SDS and the each excised band was cut in to smaller pieces. The gel pieces were transferred to a microcentrifuge tube and 100 μL of 100 mM ammonium bicarbonate/acetonitrile (1:1, v/v) was added, incubated with occasional vortexing for 30 min. 500 μL of neat acetonitrile was added and incubated at room temperature with occasional vortexing, after gel pieces become white and shrink, the acetonitrile was removed. The destained gel pieces were then subjected to in-gel digestion. Alternatively, they can be stored at -20 0C for a few weeks until they needed.

Trypsin Digestion: 50 μL of trypsin buffer (Promega gold mass spectrometry grade, Germany) was added on the destained gel pieces and it was left in an ice bucket for 30 min, and more trypsin buffer was added if all solution was absorbed. After 90 mins, 10 μL of ammonium bicarbonate buffer was added to cover the gel pieces and to keep them wet during the enzymatic cleavage. The gel pieces were kept in an air circulation thermostat for incubation overnight at 37 0C. The tubes were chilled to room temperature, and 1μL aliquot of supernatant was directly used for MALDI-TOF MS analysis. For further analysis 10 μL of 0.1 % (v/v) TFA was added in to the tube, vortexed and centrifuged at 10,000 rpm, aliquot was withdrawn and dried down in a vacuum centrifuge and stored at -20 0C till it was needed for further MS/MS analysis.

7.2.3. MALDI-TOF MS

MALDI-TOF MS analysis was carried out on selected trypsin digested proteins. For MALDI- MS sample preparation anchor chip MALDI targets (MPT AnchorChip 600-384 target, Bruker Daltonics.) were used in conjunction with DHB (dihydroxy benzoic acid) matrices (Sigma Aldrich). Using data base search algorithms the obtained mass spectrometrical data were used in attempt to sequence the proteins.

As matrix solution 5g/L 2,5-dihydroxybenzoic acid (DHB) solution in acetonitrile containing 0.1% trifluoroacetic acid (TFA) was used since DHB is more robust to impurities. 1 μl aliquot of the organic extracts of stevia was mixed with 1 μl of the matrix solution on the maldi target (MPT AnchorChip 600-384 target, Bruker Daltonics) and the matrix crystals were allowed to air- dry.

Hande Karaköse Jacobs University Bremen 97

Phytochemical Characterization of Stevia rebaudiana

MALDI-TOF spectra were acquired on an Autoflex II MALDI-TOF-TOF mass spectrometer (Bruker Daltonics) equipped with a 337 nm nitrogen laser. The instrument was operated in the reflector mode: source, 19.00 kV; lens, 8.95 kV; and reflector, 20 kV, using an optimized ion extraction delay time of 80 ns. The laser frequency was set at 25 Hz with 50 laser shots per acquisition. The laser strength was kept about 40% above threshold to obtain optimum signal to noise ratio. Spectra were obtained by summing, on average, 200 laser shots. Spectra were acquired in the mass range 0–7500 amu. The instrument was externally calibrated in the enhanced quadratic calibration mode prior to acquisition using a peptide tune-mix sample (Bruker Daltonics).

7.3. Results and Discussion 7.3.1. SDS Results

SDS gel was performed to separate the proteins according to their molecular weight, but SDS gel resulted with only one big band which corresponds to molecular weight of 55 kDa (Figure 54) and other bands appeared as a light background, which can be due to insufficient separation or due to low concentration of proteins with low range molecular weight. The bands on SDS gel was subjected to trypsin digestion and analyzed by MALDI-TOF. The results from MALDI were not sufficient for characterization of stevia proteins, thus, 2D-SDS separation of proteins were performed.

There was no significant difference observed in the protein profile of four chosen stevia samples, judged by their SDS gel, 2D gel bands and MALDI-TOF analysis.

In addition, only for one stevia sample phenol extraction method was tested and SDS gel bands were compared with TCA/acetone extraction. Essentially, there was no benefits obtained from phenol extraction, though the procedure was more time consuming.

Hande Karaköse Jacobs University Bremen 98

Phytochemical Characterization of Stevia rebaudiana

Figure 54.SDS Gel, for sample number 8 TCV harvest I, loaded on gel at different concentrations 1mg/mL and 0.5mg/mL.

7.3.2. 2D-SDS

Proteins extracted by TCA/acetone method was separated by 2D SDS−PAGE are shown in Figure 55. Essentially, the same results were obtained when the samples were prepared by phenol precipitation (data not shown). Resolved protein spots were more concentrated at 55kDa band, thus, more interest was given on these protein spots for further analysis by MALDI-TOF. The selected spots were cut and treated with trypsin digestion separately and subjected to MALDI-TOF analysis.

Hande Karaköse Jacobs University Bremen 99

Phytochemical Characterization of Stevia rebaudiana

70kD 2 4 A 55kD 8 3 A 1 40kD 9 A 11 00 35kD 9 A 6 10 25kD 00 A 5 9

Figure 55.2D-SDS separation of stevia total protein extract. 7cm strip of pH 4-7, where spot 1 and 2 are at 55kDa, and spot 5 at ~15kDa

7.3.3. MALDI-TOF MS Results

Proteins isolated by 2-D electrophoresis were digested by trypsin, and the resulting peptides were mass analyzed by MALDI-TOF. The masses obtained from the spectral data were compared with expected values computed from sequence database entries according to the enzyme's cleavage specificity. The results were scored, and the ranking suggests the protein being identified or not. Enzyme cleavage specificity, number of detected cleavage peptides, and mass accuracy are the critical parameters148. Moreover, peptide mass tolerance was set to 50 ppm; the mode of proteolytic digestion was chosen as ‘trypsin digestion’ the searching database used was NCBI and the searching taxonomy was “other green plants”. These parameters play a very crucial role in MALDI –TOF for an exact surveillance of related protein sequence and reducing the false positive results. The list of peptide masses were transferred into the peptide mass fingerprint (PMF) search program Mascot. The result of a peptide mass search with Mascot contains a lot of information. First, the probability based score is very important. A protein is identified with a score higher than 100. Second, the full protein summary report should be considered. By clicking onto the accession number of the first hit more detailed protein information is displayed. The nominal mass must be in accordance with the experimental data obtained from the gel electrophoresis. If this is not the case, protein fragments or adducts should be considered. Furthermore, the sequence coverage

Hande Karaköse Jacobs University Bremen 100

Phytochemical Characterization of Stevia rebaudiana

(SC) and the number of mass values matched (MM) are very important. The difference between the number of mass values searched and the number of mass values matched should be as small as possible149.

The results in Figure 56, shows that the PMF spectra generated has a close proximity to ribulose-1,5-bisphosphate carboxylase (RuBisCO) enzyme150, the result has the score of 232 with the expect value of 5.3E-18 at 50 ppm, and the sequence coverage is 45% (Figure 56 & 57). The sequence which appears in bold black represents the matched peaks. Each matched peak (m/z) defines a particular type of amino acid sequence, which is identified by database search. The MS/MS fragmentation was processed for the unmatched and matched peaks with respect to RuBisCO enzyme (Figure 58).

Hande Karaköse Jacobs University Bremen 101

Phytochemical Characterization of Stevia rebaudiana

Protein sequence coverage: 45%

Matched peptides shown in bold black. 1 KDYKLTYYTP EYETKDTDIL AAFRVTPQPG VPPEEAGAAV AAESSTGTWT 51 TVWTDGLTSL DRYKGRCYGI EPVPGEDNQY IAYVAYPLDL FEEGSVTNMF 101 TSIVGNVFGF KALRALRLED LRIPTAYVKT FDGPPHGIQV ERDKLNKYGR 151 PLLGCTIKPK LGLSAKNYGR ACYECLRGGL DFTKDDENVN SQPFMRWRDR 201 FLFCAEAIYK AQAETGEIKG HYLNATAGTC EDMMKRAVFA RELGVPIVMH 251 DYLTGGFTAN TSLAHYCRDN GLLLHIHRAM HAVIDRQKNH GMHFRVLAKA 301 LRMSGGDHIH SGTVVGKLEG EREITLGFVD LLRDDFIETD RSRGIYFTQD 351 WVSLPGVLPV ASGGIHVWHM PALTEIFGDD SVLQFGGGTL GHPWGNAPGA 401 VANRVALEAC VQARNEGRDL ATEGNEIIRE ATKWSPELAA ACEVWKEIKF 451 EFQAMDTLDG DKDKDKKR

Figure 56.MALDI-TOF MS spectra and mass list of trypsin digested 2D-SDS spot (spot number 8) and Mascot search result showing the sequence information for RuBisCO enzyme with the score of 45%. The sequence which appears in bold black represents the matched peaks (known in the database and matched with experimental data) and sequences which are in black represents the unmatched peaks (unknown peaks).

Hande Karaköse Jacobs University Bremen 102

Phytochemical Characterization of Stevia rebaudiana

Figure 57.Mascot search result of MALDI spectra, the score is 232 and expectation rate is 5.3e- 18; this data is generated by using Matrix science which acts as search engine. NCBI database and other green plant taxonomy was selected, the peptide mass tolerance was kept at 50 ppm and two partials. The gene bank accession number is “gi| 38146633|”.

MALDI TOF peptide mapping of the stevia protein yielded partial identification of the sequence from MASCOT database search. To prove the sequence obtained from the database and to have more information on the unknown peaks MS/MS fragmentation of chosen peaks with significant intensity was performed.

The MS/MS fragmentation of one of the unmatched peak m/z 842 was performed and the peptide was tried to be sequenced by de-novo sequencing and database search. The de-novo sequence for this peptide resulted with the sequence of AVAETVPR, however when this sequence was subjected for MASCOT search, the result was 100% sequence coverage for a hypothetical protein without any correlation with Stevia rebaudiana. Furthermore, the MS/MS fragmentation of one of the matched peak with m/z 1230 was performed (Figure 58) and again the peptide was sequenced by de-novo sequencing (Figure 59) and the sequence obtained was subjected to MASCOT database search and indeed, the suggested sequence which was DLATEGNEIIR resulted in 100% sequence coverage with that of RuBisCO sequence belonging to Stevia rebaudiana as it was shown in Figure 56, with the gene bank accession number of gi| 38146633|.

Hande Karaköse Jacobs University Bremen 103

Phytochemical Characterization of Stevia rebaudiana

Figure 58.MS/MS de novo sequencing of the m/z 1230. Series of y and b fragments are labeled.

y1 y10 y7 y6

b2

b1 b10

Figure 59.Structure and fragmentation of m/z 1230 based on de novo sequencing.

Hande Karaköse Jacobs University Bremen 104

Phytochemical Characterization of Stevia rebaudiana

The database information on plant proteomics is too limited. The only plant which has entire genome had been sequenced is Arabidopsis thaliana. Apart from that genome of few plants, such as rice maze tomato and wheat had been sequenced. Currently, there is no plant genetically related to the Asteraceae family available in the databases. Therefore, only insufficient sequence coverage is being obtained from the database searches. The missing parts of the sequence were partially covered by studying the tandem mass spectra of unknown peaks and sequencing by de- novo method. However, successful de novo sequencing requires full sequence coverage, thus demanding better quality spectra than those typically used for data base searching and sequences obtained by de novo needs confirmation and this can be done either by chemically synthesizing the obtained sequence and comparing their mass spectrical information or by complete sequencing of the Stevia rebaudiana genome.

7.4. Conclusion

This study was the first attempt for sequencing leaf proteins of a plant from Asteraceae family and for Stevia rebaudiana. MALDI-TOF has given a breakthrough in this research and once again proved to be a very crucial technique in field of proteomics, for the first time ever Stevia proteins are being characterized. The total of 75 peaks were generated by MALDI-TOF out of which 33 matched peaks yielded protein score of 232 and 45% of sequence coverage of RuBisCO enzyme, and 42 of them were considered to be unmatched with the native sequence. The satisfactory match obtained in the database with the only protein sequence established in stevia available, clearly indicates that the method employed was valid. However, further studies are necessary to have the entire sequence of proteins exist in stevia leaves.

Hande Karaköse Jacobs University Bremen 105

Phytochemical Characterization of Stevia rebaudiana

8. SUMMARY

In summary, a unique and novel dataset comprising the full analysis of around sixty secondary metabolites from an agricultural plant using seven different varieties from nine different origins were obtained. In addition to this, correlation studies between the phytochemical information and climatic metadata obtained within the project on growth conditions provide a new insight in to agricultural plant science in general.

In more detail, secondary metabolite profile of 166 Stevia rebaudiana leave samples is analyzed and ten volatile terpenes have been identified and fatty acid profile and quantities have been obtained. First attempt to sequence leaf proteins of a plant from Asteraceae family and for Stevia rebaudiana was performed. Proteins were separated and analyzed successfully and efficient results were obtained.

Furthermore, around fourty phenolic secondary metabolites, of the class of chlorogenic acids and flavonoid glycosides were identified and quantified. Additionally, ten steviol glycosides have been analyzed and quantified. For a total of ten compounds accurate quantitative data have been obtained for all 166 samples and for a further fourty compounds relative concentration variations. The data allow a full description of variations between plants of different varieties and of different origins. Both for phenolics and steviol glycosides significant variations between origins, varieties and harvests have been observed as well as variations between stevia samples cultivated in EU and outside.

As conclusion, the quantitative data allow a scientifically sound and state of the art specification of stevia leaves for licenscing as a novel food in the EU.

Hande Karaköse Jacobs University Bremen 106

Phytochemical Characterization of Stevia rebaudiana

9. REFERENCES

1. M. S. Bertoni, Revista de Agronomia de l'Assomption, Editon edn., 1899, vol. 1:35. 2. C. C. Shock, in California Agriculture, University of California, Editon edn., 1982, vol. 36, pp. 4-5. 3. N. Kolb, J. L. Herrera, D. J. Ferreyra and R. F. Uliana, Journal of Agricultural and Food Chemistry, 2001, 49, 4538-4541. 4. I. Prakash, G. E. DuBois, J. F. Clos, K. L. Wilkens and L. E. Fosdick, Food and Chemical Toxicology, 2008, 46, S75-S82. 5. C. Gardana, P. Simonetti, E. Canzi, R. Zanchi and P. Pietta, Journal of Agricultural and Food Chemistry, 2003, 51, 6618-6622. 6. N. A. Ibrahim, S. El-Gengaihi, H. Motawe and S. A. Riad, European Food Research and Technology, 2007, 224, 483-488. 7. Y. Oshima, J. Saito and H. Hikino, Tetrahedron, 1986, 42, 6443-6446. 8. Y. Oshima, J.-I. SAITO and H. HIKINO, Phytochemistry, 1988, 27, 624-626. 9. K. Yasukawa, A. Yamaguchi, J. Arita, S. Sakurai, A. Ikeda and M. Takido, Phytotherapy Research, 1993, 7, 185-189. 10. A. D. Kinghorn, Stevia The Genus Stevia, CRC Press, 2001. 11. J. Brandle, Agriculture and Agri-Food Canada, Editon edn., 2004, Retrieved 2006. 12. M. Bridel and R. Lavielle, Academie des Sciences Paris Comptes Rendus (Parts 192), Editon edn., 1931, pp. 1123–1125. 13. C. Hellfritsch, A. Brockhoff, F. Stähler, W. Meyerhof and T. Hofmann, Journal of Agricultural and Food Chemistry, 2012, 60, 6782-6793. 14. H.U. Mitchell, Sweeteners and Sugar Alternatives in Food Technology, Wiley, 2008. 15. C. Trocho, R. Pardo, I. Rafecas, J. Virgili, X. Remesar, J. A. Fernández-López and M. Alemany, Life Sciences, 1998, 63, 337-349. 16. B. A. Magnuson, G. A. Burdock, J. Doull, R. M. Kroes, G. M. Marsh, M. W. Pariza, P. S. Spencer, W. J. Waddell, R. Walker and G. M. Williams, in Crit Rev Toxicol, United States, Editon edn., 2007, vol. 37, pp. 629-727. 17. A. G. Renwick, Food Additives and Contaminants, 2006, 23, 327-338. 18. V. L. Grotz and I. C. Munro, Regulatory Toxicology and Pharmacology, 2009, 55, 1-5. 19. I. M. Baird, N. W. Shephard, R. J. Merritt and G. Hildick-Smith, Food and Chemical Toxicology, 2000, 38, Supplement 2, 123-129. 20. Saccharin warning AP via Telegraph-Herald. 1973-05-22. Retrieved 2011-06-09. 21. J. Whysner and G. M. Williams, Pharmacology & Therapeutics, 1996, 71, 225-252. 22. E. Dybing, Toxicology, 2002, 181–182, 121-125. 23. R. L. Myers and R. L. Myers, The 100 most important chemical compounds: a reference guide, Westport, Conn: Greenwood Press. 24. L. J. Navia, R. E. Walkup, N. M. Vernon, D. S. Neiditch, Production of sucralose without intermediate isolation of crystalline sucralose-6-ester, US Patent 536/124, Mar 12, 1996. 25. L. Burnett, Artificial Sweetener controversies from saccharin to sucralose. leda.law.harvard.edu. Retrieved 20 December 2010 26. H. Misra, M. Soni, N. Silawat, D. Mehta, B. Mehta and D. Jain, J Pharm Bioallied Sci, Editon edn., 2011, vol. 3, pp. 242-248. 27. J. Hawke, in Nexus magazine, Editon edn., February–March 2003, Retrieved 20 December 2010, vol. 10. 28. J. M. Pezzuto, C. M. Compadre, S. M. Swanson, D. Nanayakkara and A. D. Kinghorn, Proceedings of the National Academy of Sciences, 1985, 82, 2478-2482. 29. P. J., Elsevier, New Trends in Natural Products Chemistry, Editon edn., 1986, pp. 371-385. 30. E. Procinska, B. A. Bridges and J. R. Hanson, Mutagenesis, 1991, 6, 165-167.

Hande Karaköse Jacobs University Bremen 107

Phytochemical Characterization of Stevia rebaudiana

31. H. R. Compadre CM, Nanayakkara NP, Pezzuto JM, Kinghorn AD., Biomed Environ Mass Spectrom, Editon edn., 1988, vol. 15, pp. 211-222. 32. M. Matsui, K. Matsui, Y. Kawasaki, Y. Oda, T. Noguchi, Y. Kitagawa, M. Sawada, M. Hayashi, T. Nohmi, K. Yoshihira, M. Ishidate and T. Sofuni, Mutagenesis, 1996, 11, 573-579. 33. C. Toskulkao, L. Chaturat, P. Temcharoen and T. Glinsukon, Drug Chem Toxicol, 1997, 20, 31-44. 34. L. Xili, B. Chengjiany, X. Eryi, S. Reiming, W. Yuengming, S. Haodong, H. Zhiyian, Food Chem Toxicol, 1992, 30, 957-965. 35. Y. Aze, K. Toyoda, K. Imaida, S. Hayashi, T. Imazawa, Y. Hayashi, M. Takahashi, Eisei Shikenjo Hokoku, 1991, 48-54. 36. K. Toyoda, H. Matsui, T. Shoda, C. Uneyama, K. Takada, M. Takahashi, Food and Chemical Toxicology, 1997, 35, 597-603. 37. R. M. Oliveira-Filho, O. A. Uehara, C. A. Minetti and L. B. Valle, Gen Pharmacol, 1989, 20, 187- 191. 38. C. Wasuntarawat, P. Temcharoen, C. Toskulkao, P. Mungkornkarn, M. Suttajit, T. Glinsukon, Drug Chem Toxicol, 1998, 21, 207-222. 39. Official Journal of the European Union, Editon edn., 11 November 2011, vol. 1131/2011, p. 26. 40. V. Chatsudthipong and C. Muanprasat, Pharmacology & Therapeutics, 2009, 121, 41-54. 41. C. Toskulkao, M. Sutheerawatananon, C. Wanichanon, P. Saitongdee and M. Suttajit, Journal of Nutritional Science and Vitaminology, 1995, 41, 105-113. 42. J. M. C. Geuns, J. Buyse, A. Vankeirsbilck, E. H. M. Temme, F. Compernolle and S. Toppet, Journal of Agricultural and Food Chemistry, 2006, 54, 2794-2798. 43. G. Brahmachari, L. C. Mandal, R. Roy, S. Mondal and A. K. Brahmachari, Archiv der Pharmazie, 2011, 344, 5-19. 44. E. Koyama, K. Kitazawa, Y. Ohori, O. Izawa, K. Kakegawa, A. Fujino, Food and Chemical Toxicology, 2003, 41, 359-374. 45. A. Wheeler, A. C. Boileau, P. C. Winkler, J. C. Compton, I. Prakash, X. Jiang, D. A. Mandarino, Food and Chemical Toxicology, 2008, 46, S54-S60. 46. N. Totté, L. Charon, M. Rohmer, F. Compernolle, I. Baboeuf and J. M. C. Geuns, Tetrahedron Letters, 2000, 41, 6407-6410. 47. N. Totté, W. V. d. Ende, E. J. M. Van Damme, F. Compernolle, I. Baboeuf, J. M. C. Geuns, Canadian Journal of Botany, 2003, 81, 517-522. 48. J. E. Brandle, A. Richman, A. K. Swanson and B. P. Chapman, Plant Molecular Biology, 2002, 50, 613-622. 49. T. Humphrey, A. Richman, R. Menassa and J. Brandle, Plant Molecular Biology, 2006, 61, 47-62. 50. J. E. Brandle and P. G. Telmer, Phytochemistry, 2007, 68, 1855-1863. 51. A. Hemmerlin, J. L. Harwood and T. J. Bach, Progress in Lipid Research, 2012, 51, 95-148. 52. J. Metivier and A. M. Viana, Journal of Experimental Botany, 1979, 30, 805-810. 53. P. Mauri, G. Catalano, C. Gardana and P. Pietta, ELECTROPHORESIS, 1996, 17, 367-371. 54. P. Nishiyama, M. Alvarez and L. G. E. Vieira, Journal of the Science of Food and Agriculture, 1992, 59, 277-281. 55. H. Mizukami, K. Shiiba and H. Ohashi, Phytochemistry, 1982, 21, 1927-1930. 56. M. Jamaluddin Ahmed and R. M. Smith, Journal of Separation Science, 2002, 25, 170-172. 57. U. Woelwer-Rieck, C. Lankes, A. Wawrzun and M. Wüst, European Food Research and Technology, 2010, 231, 581-588. 58. H. C. Makapugay, N. P. D. Nanayakkara and A. D. Kinghorn, Journal of Chromatography A, 1984, 283, 390-395. 59. , P. Karásek, M. Roth, K Beneš K řík Č ký Analytical and Bioanalytical Chemistry, 2007, 388, 1847-1857. 60. J. Liu and S. F. Y. Li, Journal of Liquid Chromatography, 1995, 18, 1703-1719.

Hande Karaköse Jacobs University Bremen 108

Phytochemical Characterization of Stevia rebaudiana

61. T něk Nep ı mp x A m ı nd ček Journal of Food Composition and Analysis, 2001, 14, 383-388. 62. V. J. Y. Minne, F. Compernolle, S. Toppet and J. M. C. Geuns, Journal of Agricultural and Food Chemistry, 2004, 52, 2445-2449. 63. e -Rieck, W. Tomberg and A. Wawrzun, Journal of Agricultural and Food Chemistry, 2010, 58, 12216-12220. 64. B. Zimmermann, U. Woelwer-Rieck and M. Papagiannopoulos, Food Analytical Methods, 2012, 5, 266-271. 65. F. Cacciola, P. Delmonte, K. Jaworska, P. Dugo, L. Mondello and J. I. Rader, Journal of Chromatography A, 2011, 1218, 2012-2018. 66. J. Pol, B. Hohnova and T. Hyotylainen, Journal of Chromatography A, 2007, 1150, 85-92. 67. C. Gardana, M. Scaglianti and P. Simonetti, Journal of Chromatography A, 2010, 1217, 1463- 1470. 68. J.M.C. Geuns, Validated techniques of analysis. In Stevia and Steviol Glycosides, Euprint:Heverlee, Belgium, 2010; ISBN 978-90-742-53116, p61-113. 69. D. M. W.-D. Eric S. Grumbach , Pamela C. Iraneta , Bonnie Alden , Jeffrey R. Mazzeo, in LCGC North America, Editon edn., 2004, vol. 22, p. 1010. 70. M. Murata, E. Tanaka, E. Minoura and S. Homma, Bioscience, Biotechnology, and Biochemistry, 2004, 68, 501-507. 71. L. Bravo, Nutrition Reviews, 1998, 56, 317-333. 72. M. Naczk and F. Shahidi, Journal of Chromatography A, 2004, 1054, 95-111. 73. H. Wang, G. Cao and R. L. Prior, Journal of Agricultural and Food Chemistry, 1996, 44, 701-705. 74. M. Naczk and F. Shahidi, Journal of Pharmaceutical and Biomedical Analysis, 2006, 41, 1523- 1542. 75. C. Manach, A. Scalbert, C. Morand, C. Rémésy and L. Jiménez, The American Journal of Clinical Nutrition, 2004, 79, 727-747. 76. A. Crozier, M. E. J. Lean, M. S. McDonald and C. Black, Journal of Agricultural and Food Chemistry, 1997, 45, 590-595. 77. A. J. Stewart, S. Bozonnet, W. Mullen, G. I. Jenkins, M. E. J. Lean and A. Crozier, Journal of Agricultural and Food Chemistry, 2000, 48, 2663-2669. 78. L. Gennaro, C. Leonardi, F. Esposito, M. Salucci, G. Maiani, G. Quaglia and V. Fogliano, Journal of Agricultural and Food Chemistry, 2002, 50, 1904-1910. 79. N. Balasundram, K. Sundram and S. Samman, Food Chemistry, 2006, 99, 191-203. 80. M. N. Clifford, Journal of the Science of Food and Agriculture, 2000, 80, 1033-1043. 81. M. N. Clifford, K. L. Johnston, S. Knight and N. Kuhnert, Journal of Agricultural and Food Chemistry, 2003, 51, 2900-2911. 82. M. N. Clifford and J. R. Ramirez-Martinez, Food Chemistry, 1990, 35, 13-21. 83. V. Marques and A. Farah, Food Chemistry, 2009, 113, 1370-1376. 84. J. H. Chen and C.-T. Ho, Journal of Agricultural and Food Chemistry, 1997, 45, 2374-2378. 85. M. Lepelley, G. Cheminade, N. Tremillon, A. Simkin, V. Caillet and J. McCarthy, Plant Science, 2007, 172, 978-996. 86. M. R. Olthof, P. C. H. Hollman and M. B. Katan, The Journal of Nutrition, 2001, 131, 66-71. 87. Y. Kono, K. Kobayashi, S. Tagawa, K. Adachi, A. Ueda, Y. Sawa and H. Shibata, Biochimica Et Biophysica Acta-General Subjects, 1997, 1335, 335-342. 88. M. R. Luzia, K. C. C. DaPaixao, R. Marcilio, L. C. Trugo, L. M. C. Quinteiro and C. A. B. DeMaria, International Journal of Food Science and Technology, 1997, 32, 15-19. 89. A. W. Herling, H. J. Burger, D. Schwab, H. Hemmerle, P. Below and G. Schubert, American Journal of Physiology-Gastrointestinal and Liver Physiology, 1998, 274, G1087-G1093.

Hande Karaköse Jacobs University Bremen 109

Phytochemical Characterization of Stevia rebaudiana

90. J. Shearer, A. Farah, T. de Paulis, D. P. Bracy, R. R. Pencek, T. E. Graham and D. H. Wasserman, Journal of Nutrition, 2003, 133, 3529-3532. 91. H. Hemmerle, H. J. Burger, P. Below, G. Schubert, R. Rippel, P. W. Schindler, E. Paulus and A. W. Herling, Journal of Medicinal Chemistry, 1997, 40, 137-145. 92. W. J. Arion, W. K. Canfield, F. C. Ramos, P. W. Schindler, H. J. Burger, H. Hemmerle, G. Schubert, P. Below and A. W. Herling, Archives of Biochemistry and Biophysics, 1997, 339, 315-322. 93. K. L. Johnston, M. N. Clifford and L. M. Morgan, American Journal of Clinical Nutrition, 2003, 78, 728-733. 94. M. D. del Castillo, J. M. Ames and M. H. Gordon, Journal of Agricultural and Food Chemistry, 2002, 50, 3698-3703. 95. W. E. Robinson, M. Cordeiro, S. AbdelMalek, Q. Jia, S. A. Chow, M. G. Reinecke and W. M. Mitchell, Molecular Pharmacology, 1996, 50, 846-855. 96. W. E. Robinson, M. G. Reinecke, S. AbdelMalek, Q. Jia and S. A. Chow, Proceedings of the National Academy of Sciences of the United States of America, 1996, 93, 6326-6331. 97. B. McDougall, P. J. King, B. W. Wu, Z. Hostomsky, M. G. Reinecke and W. E. Robinson, Antimicrobial Agents and Chemotherapy, 1998, 42, 140-146. 98. A. Trute, J. Gross, E. Mutschler and A. Nahrstedt, Planta Medica, 1997, 63, 125-129. 99. H. F. Stich, M. P. Rosin and L. Bryson, Mutation Research, 1982, 95, 119-128. 100. N. Hanlon, N. Coldham, A. Gielbert, N. Kuhnert, M. J. Sauer, L. J. Kingi and C. Ioannides, British Journal of Nutrition, 2008, 99, 559-564. 101. A. Stalmach, W. Mullen, D. Barron, K. Uchida, T. Yokota, C. Cavin, H. Steiling, G. Williamson and A. Crozier, Drug Metabolism and Disposition, 2009, 37, 1749-1758. 102. A. Farah, M. Monteiro, C. M. Donangelo and S. Lafay, Journal of Nutrition, 2008, 138, 2309-2315. 103. H. M. Merken and G. R. Beecher, Journal of Agricultural and Food Chemistry, 2000, 48, 577-599. 104. P.-G. Pietta, Journal of Natural Products, 2000, 63, 1035-1042. 105. E. de Rijke, P. Out, W. M. A. Niessen, Journal of Chromatography A, 2006, 1112, 31-63. 106. A z nić F Buc Ž M eš A M n B Nig ić Molecules, Molecules, Editon edn., 2009, vol. 14, pp. 2466-2490. 107. P. C. H. Hollman and M. B. Katan, Food and Chemical Toxicology, 1999, 37, 937-942. 108. I. Ignat, I. Volf and V. I. Popa, Food Chemistry, 2011, 126, 1821-1835. 109. I.-S. Kim, M. Yang, O.-H. Lee and S.-N. Kang, LWT - Food Science and Technology, 2011, 44, 1328- 1332. 110. S. Ghanta, A. Banerjee, A. Poddar and S. Chattopadhyay, Journal of Agricultural and Food Chemistry, 2007, 55, 10962-10967. 111. U. Wölwer-Rieck, Journal of Agricultural and Food Chemistry, 2012, 60, 886-895. 112. G. Li, R. Wang, A. J. Quampah, Z. Rong, C. Shi and J. Wu, Journal of Agricultural and Food Chemistry, 2011, 59, 13065-13071. 113. J. M. González-Buitrago, L. Ferreira, M. Isidoro-García, C. Sanz, F. Lorente and I. Dávila, Clinica Chimica Acta, 2007, 385, 21-27. 114. S. Rogers, M. Girolami, W. Kolch, K. M. Waters, T. Liu, B. Thrall and H. S. Wiley, Bioinformatics, 2008, 24, 2894-2900. 115. V. Dhingra, M. Gupta, T. Andacht and Z. F. Fu, International Journal of Pharmaceutics, 2005, 299, 1-18. 116. C. M. D. Tal Isaacson, Ramu S Saravanan,Yonghuae, Carmen Catala, Montserrat Saladie & Jocelyn KC Rose, Nature protocol, Editon edn., 13 July 2006. 117. R. M. LoPachin, R. C. Jones, T. A. Patterson, W. Slikker Jr and D. S. Barber, NeuroToxicology, 2003, 24, 761-775. 118. J. A. Reynolds and C. Tanford, Proc Natl Acad Sci U S A, 1970, 66, 1002-1007. 119. V. Neuhoff, N. Arold, D. Taube and W. Ehrhardt, Electrophoresis, 1988, 9, 255-262.

Hande Karaköse Jacobs University Bremen 110

Phytochemical Characterization of Stevia rebaudiana

120. J. E. Shively, EXS, 2000, 88, 99-117. 121. D. Lin, D. L. Tabb, Biochimica et Biophysica Acta (BBA) - Proteins&Proteomics, 2003, 1646, 1-10. 122. X. Han, A. Aslanian and J. R. Yates Iii, Current Opinion in Chemical Biology, 2008, 12, 483-490. 123. R. Krüger, A. Pfenninger, I. Fournier, M. Glückmann and M. Karas, Analytical Chemistry, 2001, 73, 5812-5821. 124. J. B. Ohlrogge and J. G. Jaworski, Annual Review of Plant Physiology and Plant Molecular Biology, 1997, 48, 109-136. 125. A. Carrasco-Pancorbo, N. Navas-Iglesias and L. Cuadros-Rodríguez, TrAC Trends in Analytical Chemistry, 2009, 28, 263-278. 126. C. Zhu, P. Hu, Q.-L. Liang, Y.-M. Wang and G.-A. Luo, Chinese Journal of Analytical Chemistry, 2009, 37, 1390-1396. 127. J. Folch, S. Arsove and J. A. Meath, Journal of Biological Chemistry, 1951, 191, 819-831. 128. V. Matyash, G. Liebisch, T. V. Kurzchalia, A. Shevchenko and D. Schwudke, Journal of Lipid Research, 2008, 49, 1137-1146. 129. A. Nomura, J. Yamada, T. Yarita and T. Maeda, The Journal of Supercritical Fluids, 1995, 8, 329- 333. 130. T. White, S. Bursten, D. Federighi, R. A. Lewis and E. Nudelman, Analytical Biochemistry, 1998, 258, 109-117. 131. K. Eder, Journal of Chromatography B: Biomedical Sciences and Applications, 1995, 671, 113- 131. 132. X. Han and R. W. Gross, in J Lipid Res, United States, Editon edn., 2003, vol. 44, pp. 1071-1079. 133. C. Hu, R. van der Heijden, M. Wang, J. van der Greef, T. Hankemeier and G. Xu, Journal of Chromatography B, 2009, 877, 2836-2846. 134. K. U. Borstel René, LCGC chromatography online, Application note, Editon edn., 2011. 135. M. N. Clifford, S. Knight and N. Kuhnert, Journal of Agricultural and Food Chemistry, 2005, 53, 3821-3832. 136. A. J. Day, F. Mellon, D. Barron, G. Sarrazin, M. R. A. Morgan, Free Radical Research, 2001, 35, 941-952. 137. A. Plazonic, F. Bucar, Z. Males, A. Mornar, B. Nigovic and N. Kujundzic, in Molecules, Switzerland, Editon edn., 2009, vol. 14, pp. 2466-2490. 138. F. Cuyckens and M. Claeys, Journal of Mass Spectrometry, 2004, 39, 1-15. 139. Y.L. Ma, Q.M. Li, H. Van den Heuvel, M. Claeys, Rapid Communications in Mass Spectrometry, 1997, 11, 1357-1364. 140. J. M. Harnly, R. F. Doherty, G. R. Beecher, J. M. Holden, D. B. Haytowitz, S. Bhagwat, S. Gebhardt, Journal of Agricultural and Food Chemistry, 2006, 54, 9966-9977. 141. C. J. Schwartz, The mixed-model ANOVA: The truth, the computer packages, the books. Part I: Balanced data, The American Statistician, 1993. 142. L. Zelles and Q. Y. Bai, Soil Biology & Biochemistry, 1993, 25, 495-507. 143. G. Gutnikov, Journal of Chromatography B-Biomedical Applications, 1995, 671, 71-89. 144. A. Martelli, C. Frattini and F. Chialva, Flavour and Fragrance Journal, 1985, 1, 3-7. 145. A. L. Thompson, A. Michalik, R. J. Nash, F. X. Wilson, R. van Well, P. Johnson, G. W. J. Fleet, C. Y. Yu, X. G. Hu, R. I. Cooper and D. J. Watkin, Acta Crystallographica Section E-Structure Reports Online, 2009, 65, O2904-U2232. 146. G. B. A. Görg, O. Drews, A. Köpf, C. Lück, G. Reil, W. Weiss, A Laboratory Manual, Technische Universität München, Editon edn., 2003. 147. A. Shevchenko, H. Tomas, J. Havlis, J. V. Olsen and M. Mann, Nat. Protocols, 2007, 1, 2856-2860. 148. V. Egelhofer, K. Büssow, C. Luebbert, H. Lehrach and E. Nordhoff, Analytical Chemistry, 2000, 72, 2741-2750.

Hande Karaköse Jacobs University Bremen 111

Phytochemical Characterization of Stevia rebaudiana

149. B. Thiede, W. Höhenwarter, A. Krah, J. Mattow, M. Schmid, F. Schmidt and P. R. Jungblut, Methods, 2005, 35, 237-247. 150. J. L. Panero and V. A. Funk, Molecular Phylogenetics and Evolution, 2008, 47, 757-782.

Hande Karaköse Jacobs University Bremen 112

Phytochemical Characterization of Stevia rebaudiana

APPENDIX

A. Tandem mass spectra of steviol glycosides in negative ion mode.

Stevioside m/z 803 [M-H+]-

Intens. -MS2(803.4) x10 7 641.2 1.0 0.5 0.0 x10 6 -MS3(803.8->641.2) 6 479.1 4 2 317.0 0 x10 6 -MS4(803.8->641.5->479.3) 2 317.0

1

0 200 400 600 800 1000 m/z

Steviolbioside m/z 641 [M-H+]-

Intens. -MS2(641.2) x10 7 479.1 2 1 317.0

07 x10 -MS3(641.6->479.4) 1.0 317.0

0.5

0.0 2 -MS4(641.6->479.4->317.1) 1 0 -1 200 400 600 800 1000 m/z

Hande Karaköse Jacobs University Bremen 113

Phytochemical Characterization of Stevia rebaudiana

Rebaudioside F m/z 935 [M-H+]-

Intens. x10 6 -MS2(935.5) 773.4 1.0 0.5

0.0 x10 5 -MS3(936.0->773.3), 3 611.2 2 1 479.1 317.1 413.1 0 x10 4 -MS4(936.0->773.7->613.4) 6 479.1 4 2 317.0 0 200 400 600 800 1000 m/z

Dulcoside A m/z 787 [M-H+]- m/z 823[M+Cl-]-

Intens. -MS2(823.4) x10 7 625.3 1.0 0.5 0.0 x106 -MS3(824.0->625.2) 4 479.1

2 317.0

05 x10 -MS4(824.0->625.5->479.3) 6 317.0 4 2 0 200 400 600 800 1000 m/z

Rubusoside m/z 641[M-H+]-

Intens. -MS2(641.2) x107 479.1 2 317.1

07 x10 -MS3(641.6->479.4) 317.0 1.0 0.5 0.0 2 -MS4(641.6->479.4->317.2) 1 0 -1 200 400 600 800 1000 m/z

Hande Karaköse Jacobs University Bremen 114

Phytochemical Characterization of Stevia rebaudiana

B. Quantification data of steviol glycosides

Stdev(σ) 0.741 3.077 0.152 0.080 0.154 3.200 g/100g RebA Stevioside DulcosideA Rubusoside RebC sum leaves min 0.079 0.252 0.005 0.005 0.026 0.554 average 1.017 7.314 0.264 0.122 0.319 9.036 max 5.336 17.509 0.680 0.459 0.820 18.067

20.000 allsamples 18.000 16.000 14.000

12.000 min 10.000 average 8.000 max 6.000 4.000 2.000 0.000 RebA stevioside DulcosideA Rubusoside RebC sum

Bar plot showing minimum, maximum and average values obtained from all 166 stevia samples within within +/- 3 σ

Hande Karaköse Jacobs University Bremen 115

Quantity of Steviol Glycosides RebA Stevioside DulcosideA Rubusoside RebC Sum Sample No. Origin Variety Harvesting Year ( g/100g) (g/100g) ( g/100g) ( g/100g) (g/100g) ( g/100g) 1 TCV 3 II 10.09.2010 0.596 6.168 0.140 0.128 0.196 7.229 2 TCV 4 II 14.09.2010 1.411 7.781 0.263 0.113 0.409 9.975 3 TCV 3 I 04.08.2010 1.203 7.980 0.299 0.086 0.383 9.952 4 TCV 1 I 04.08.2010 1.400 8.197 0.132 0.138 0.447 10.313 5 TCV 1 I 04.08.2010 1.592 7.608 0.103 0.101 0.518 9.921 6 TCV 2 I 04.08.2010 1.983 2.962 0.186 0.042 0.366 5.538 7 TCV 2 I 04.08.2010 1.959 2.863 0.104 0.049 0.476 5.450 8 TCV 4 I 04.08.2010 1.591 4.773 0.273 0.118 0.409 7.164 9 TCV 4 II 14.09.2010 1.007 5.390 0.300 0.128 0.297 7.122 10 TCV 2 II 28.09.2010 2.154 7.360 0.179 0.107 0.399 10.199 11 TCV 2 II 28.09.2010 1.607 6.803 0.210 0.101 0.308 9.030 12 Pojava 4 II 14.09.2010 0.565 6.316 0.235 0.083 0.211 7.411 13 TCV 3 I 04.08.2010 0.632 5.648 0.254 0.080 0.215 6.827 14 TCV 2 I 04.08.2010 2.058 3.585 0.073 0.038 0.345 6.100 15 TCV 1 I 01.09.2010 1.001 5.178 0.075 0.072 0.256 6.582 16 TCV 2 I 04.08.2010 1.569 3.314 0.073 0.035 0.503 5.494 17 TCV 3 I 04.08.2010 0.717 5.769 0.217 0.048 0.223 6.975 18 TCV 4 I 04.08.2010 0.800 5.322 0.169 0.047 0.248 6.586 19 TCV 3 II 10.09.2010 0.550 6.003 0.178 0.097 0.178 7.006 20 TCV 2 II 28.09.2010 1.622 4.889 0.078 0.051 0.316 6.956 21 TCV 1 II 28.09.2010 0.853 7.590 0.117 0.073 0.322 8.955 22 TCV 3 II 10.09.2010 0.362 7.808 0.308 0.136 0.154 8.768 23 TCV,Pojana 4 I 04.08.2010 0.640 6.297 0.165 0.048 0.254 7.404 24 TCV,pojana 3 I 04.08.2010 0.292 7.049 0.380 0.074 0.150 7.945 25 TCV 1 I 01.09.2010 0.923 5.300 0.065 0.075 0.256 6.619 26 TCV 4 II 14.09.2010 0.738 5.554 0.154 0.063 0.209 6.719 27 TCV 4 I 04.08.2010 0.657 5.138 0.158 0.041 0.196 6.190

Hande Karaköse Jacobs University Bremen 116

Quantity of Steviol Glycosides

Sample No. Origin Variety Harvesting Year RebA Stevioside DulcosideA Rubusoside RebC sum 28 TCV 2 II 28.09.2010 1.669 4.870 0.266 0.058 0.820 7.684 29 TCV 3 II 10.09.2010 0.437 6.549 0.255 0.125 0.169 7.535 30 Uconor 4 II 07.05.2010 0.562 5.699 0.209 0.065 0.185 6.720 31 Uconor 3 I 11.08.2010 0.337 3.769 0.297 0.075 0.160 4.638 32 Uconor 2 I 11.08.2010 1.955 3.881 0.048 0.034 0.407 6.325 33 Uconor 2 II 07.05.2010 1.672 4.530 0.060 0.045 0.503 6.810 34 Uconor 4 II 07.09.2010 0.722 5.217 0.124 0.064 0.202 6.328 35 Uconor 4 I 11.08.2010 0.404 4.799 0.280 0.071 0.143 5.696 36 Uconor 4 I 11.08.2010 0.619 3.780 0.104 0.045 0.195 4.745 37 Uconor 4 I 11.08.2010 1.032 5.085 0.087 0.047 0.286 6.537 38 Uconor 3 I 11.08.2010 0.540 6.065 0.235 0.076 0.201 7.117 39 Uconor 4 I 11.08.2010 0.797 3.773 0.122 0.044 0.223 4.959 40 Uconor 2 I 11.08.2010 1.471 2.947 0.075 0.041 0.479 5.015 41 Uconor 3 I 11.08.2010 0.664 5.185 0.139 0.057 0.268 6.313 42 Uconor 3 II 07.09.2010 0.214 5.704 0.281 0.125 0.127 6.451 43 Agrinion 2 II 20.09.2010 1.561 4.908 0.173 0.080 0.325 7.047 44 Toumpa 1 I 30.07.2010 0.444 4.896 0.206 0.083 0.166 5.795 45 Portugal 1 I 07.07.2010 0.600 6.732 0.051 0.064 0.252 7.700 46 Amfilia 1 I 04.08.2010 0.266 5.301 0.240 0.078 0.133 6.018 47 Toumpa 3 II 10.09.2010 0.365 4.107 0.126 0.054 0.143 4.794 48 Agrinion 2 I 09.08.2010 1.048 4.093 0.133 0.053 0.242 5.568 49 Toumpa 3 I 30.07.2010 0.355 5.134 0.190 0.071 0.146 5.895 50 Agrinion 4 I 09.08.2010 0.789 3.533 0.064 0.026 0.251 4.664 51 Agrinion 4 II 20.09.2010 1.034 4.449 0.143 0.066 0.243 5.935 52 Portugal 4 I 26.06.2010 0.563 4.573 0.124 0.103 0.188 5.551 53 Amfilia 4 I 04.08.2010 0.434 4.741 0.200 0.071 0.155 5.601 54 Amfilia 4 II 15.09.2010 0.491 4.528 0.147 0.094 0.200 5.460 55 Argentinie 2009 - - 2009 0.514 5.264 0.289 0.220 0.216 6.504

Hande Karaköse Jacobs University Bremen 117

Quantity of Steviol Glycosides

Sample No. Origin Variety Harvesting Year RebA Stevioside DulcosideA Rubusoside RebC sum 56 Paragan2009 - - 2009 0.373 3.750 0.178 0.162 0.200 4.663 57 Argentinien2010 - - 2010 0.552 7.212 0.337 0.240 0.252 8.593 58 Amfilia 2 II 15.09.2010 1.125 5.026 0.119 0.107 0.302 6.680 59 Agrinion 1 II 20.09.2010 1.044 7.864 0.348 0.116 0.362 9.735 60 Amfilia 2 I 04.08.2010 1.114 6.732 0.265 0.083 0.323 8.518 61 Amfilia 3 II 15.09.2010 0.770 6.073 0.184 0.127 0.328 7.481 62 Toumpa 1 I 30.07.2010 0.444 4.396 0.567 0.211 0.210 5.829 63 Amfilia 1 II 15.09.2010 0.695 3.428 0.366 0.187 0.257 4.933 64 Portugal 3 I 26.06.2010 1.768 6.869 0.546 0.339 0.578 10.100 65 Toumpa 1 II 10.09.2010 0.703 3.745 0.455 0.173 0.366 5.442 66 Agrinion 1 I 09.08.2010 0.673 3.569 0.357 0.107 0.327 5.033 67 Amfilia 3 I 04.08.2010 1.100 7.131 0.190 0.097 0.366 8.884 68 Toumpa 2 II 10.09.2010 0.419 8.115 0.458 0.384 0.190 9.566 69 Agrinio 3 II 20.09.2010 0.910 9.167 0.488 0.151 0.274 10.990 70 Toumpa 4 II 10.09.2010 0.733 7.944 0.342 0.123 0.268 9.410 71 Agrinion 3 I 09.08.2010 0.693 7.053 0.327 0.123 0.292 8.488 72 Toumpa 4 I 30.07.2010 0.806 5.995 0.222 0.060 0.243 7.326 73 Granada 2 I 15.09.2010 1.728 6.826 0.073 0.110 0.486 9.224 74 Granada 4 I 15.09.2010 0.779 6.875 0.261 0.080 0.332 8.327 75 Granada 3 I 09.09.2010 0.484 7.916 0.406 0.099 0.225 9.130 76 Equador - - - 0.647 7.599 0.005 0.005 0.067 8.323 77 DZ - - - 0.319 7.042 0.048 0.050 0.082 7.541 78 Kenia 2010 - - 2010 0.615 7.866 0.019 0.051 0.123 8.674 79 Paraguay 2009 - - 2009 0.193 10.950 0.078 0.049 0.094 11.364 80 Argentinien 2010 - - 2010 0.185 12.898 0.110 0.055 0.100 13.349 81 Indien 2010 - - 2010 0.556 14.084 0.054 0.053 0.162 14.907 82 Argenitinien 2009 - - 2009 0.245 12.551 0.091 0.048 0.026 12.962 83 Hohenheim 2010 - - 2010 0.443 17.509 0.074 0.016 0.026 18.067 118

Quantity of Steviol Glycosides

Sample No. Origin Variety Harvesting Year RebA Stevioside DulcosideA Rubusoside RebC Sum 84 Krim - - - 0.190 0.252 0.026 0.046 0.040 0.554 85 Turkei PS1 2010 - - 2010 0.500 12.696 0.028 0.021 0.214 13.282 86 Turkei PS2 2010 - - 2010 0.446 8.683 0.022 0.016 0.196 9.202 87 Uconor 7 I 13.07.2011 0.523 1.058 0.041 0.012 0.029 1.664 88 Agrinon 4 I 2011 0.721 2.416 0.049 0.011 0.031 3.228 89 Uconor 4 I 2011 0.079 4.269 0.299 0.107 0.188 4.943 90 Agrinion 3 I 2011 0.421 8.658 0.068 0.016 0.049 9.210 91 Agrinion 5 I 2011 0.396 3.335 0.177 0.061 0.272 4.241 92 Uconor 6 I 13.07.2011 0.455 2.968 0.124 0.066 0.191 3.804 93 Uconor 3 I 13.07.2011 0.151 3.060 0.306 0.056 0.075 3.648 94 Uconor 5 I 13.07.2011 0.639 2.590 0.137 0.041 0.170 3.577 95 Agrinion 6 I 2011 0.790 3.817 0.151 0.055 0.282 5.096 96 Toumpa 1 I 2011 0.576 5.840 0.372 0.182 0.263 7.234 97 Toumpa 2 I 2011 1.101 4.465 0.254 0.203 0.282 6.305 98 Toumpa 3 I 2011 0.501 5.638 0.356 0.171 0.168 6.833 99 Toumpa 4 I 2011 0.943 4.585 0.228 0.127 0.306 6.189 100 Amiflikeia 4 II 2011 1.603 6.729 0.563 0.103 0.530 9.529 101 Amiflikeia 5 II 2011 1.611 10.507 0.275 0.148 0.533 13.073 102 Amiflikeia 4 II 2011 1.753 10.559 0.403 0.141 0.452 13.309 103 Amiflikeia 3 II 2011 1.088 10.636 0.491 0.211 0.374 12.799 104 Amiflikeia - II 2011 0.802 11.020 0.306 0.146 0.305 12.578 105 Amiflikeia 6 II 2011 1.827 9.539 0.282 0.169 0.512 12.330 106 Amiflikeia 1 II 2011 1.053 10.517 0.457 0.233 0.337 12.597 107 APTTB 3 I 2011 1.029 10.030 0.420 0.149 0.307 11.935 108 APTTB 5 I 2011 1.101 9.921 0.350 0.154 0.333 11.859 109 APTTB 6 I - 1.282 8.761 0.268 0.164 0.358 10.833 110 APTTB 4 I - 1.078 8.955 0.286 0.140 0.398 10.856 111 Uconor 5 II 2011 1.133 8.765 0.337 0.108 0.444 10.786 119

Quantity of Steviol Glycosides

Sample No. Origin Variety Harvesting Year RebA Stevioside DulcosideA Rubusoside RebC sum 112 Uconor 4 II 2011 1.589 7.906 0.180 0.086 0.417 10.178 113 Uconor 3 II 2011 0.646 8.769 0.375 0.207 0.266 10.263 114 Uconor 6 II 2011 1.110 10.151 0.297 0.101 0.473 12.133 115 Conaga 3 I 2011 0.694 9.683 0.413 0.100 0.285 11.175 116 Conaga 5 I 2011 1.856 8.011 0.247 0.055 0.429 10.599 117 Conaga 4 I 2011 0.997 10.457 0.395 0.098 0.384 12.331 118 Agrinion 5 II 2011 1.601 9.733 0.349 0.149 0.467 12.298 119 Agrinion 4 II 2011 1.173 9.625 0.351 0.148 0.379 11.676 120 Conaga 6 I 2011 2.222 8.436 0.346 0.064 0.587 11.654 121 Conaga 7 I 2011 4.901 1.148 0.680 0.021 0.680 7.429 122 Agrinion 6 II 2011 1.397 9.888 0.306 0.147 0.547 12.285 123 Agrinion 3 II 2011 1.140 10.231 0.490 0.181 0.434 12.477 124 Toumpa 7 I 2011 5.336 0.966 0.050 0.018 0.711 7.082 125 Toumpa 4 III 2011 1.000 10.844 0.405 0.159 0.326 12.734 126 Toumpa 3 III 2011 0.897 11.795 0.596 0.271 0.340 13.898 127 Toumpa 3 II 2011 0.743 10.671 0.524 0.249 0.330 12.517 128 Agrinion 4 II 2011 1.447 8.980 0.337 0.155 0.461 11.380 129 Toumpa 4 II 2011 0.934 9.641 0.335 0.191 0.376 11.477 130 Toumpa 3 II 2011 0.722 10.616 0.547 0.190 0.303 12.377 131 Toumpa 4 II 2011 1.235 9.892 0.349 0.146 0.466 12.089 132 Agrinion 3 II 2011 0.956 9.886 0.383 0.222 0.407 11.855 133 Toumpa 5 II 2011 0.770 10.641 0.427 0.194 0.387 12.420 134 Toumpa 6 II 2011 1.030 9.752 0.290 0.159 0.464 11.695 135 Toumpa 7 II 2011 4.590 1.515 0.017 0.023 0.599 6.745 136 TCV 5 I 30.06.2011 0.836 7.947 0.139 0.165 0.366 9.454 137 TCV 3 I 30.06.2011 0.525 9.503 0.411 0.224 0.281 10.943 138 TCV 6 I 30.06.2011 0.824 7.083 0.222 0.142 0.377 8.648 139 TCV 3 II 11.08.2011 0.642 9.780 0.498 0.184 0.246 11.350 120

Quantity of Steviol Glycosides

Sample No. Origin Variety Harvesting Year RebA Stevioside DulcosideA Rubusoside RebC sum 140 TCV 6 II 17.08.2011 0.922 8.626 0.302 0.116 0.537 10.504 141 TCV 4 II 24.08.2011 1.090 8.332 0.285 0.125 0.405 10.236 142 TCV 7 I 18.08.2011 3.729 1.099 0.019 0.031 0.613 5.491 143 TCV 4 I 07.07.2011 0.670 8.299 0.291 0.129 0.440 9.828 144 TCV 5 II 17.08.2011 1.215 10.728 0.410 0.127 0.594 13.073 145 Amiflikeia 3 I 2011 0.574 10.429 0.490 0.207 0.269 11.970 146 Amiflikeia 4 I 2011 1.167 10.350 0.310 0.195 0.422 12.444 147 Amiflikeia 5 I 2011 1.440 10.736 0.297 0.230 0.456 13.159 148 Amiflikeia 6 I 2011 1.233 10.463 0.355 0.225 0.430 12.706 149 Amiflikeia 1 I 2011 0.771 12.365 0.529 0.383 0.287 14.335 150 Amiflikeia 2 I 2011 1.175 11.315 0.397 0.334 0.425 13.646 151 Amiflikeia 4 I 2011 0.867 11.958 0.501 0.312 0.299 13.937 152 Amiflikeia 3 I 2011 0.711 10.987 0.623 0.459 0.313 13.093 153 Toumpa 3 I 2011 0.704 11.494 0.580 0.354 0.249 13.381 154 Toumpa 5 I 2011 0.902 10.066 0.316 0.160 0.387 11.831 155 Toumpa 6 I 2011 1.033 7.601 0.185 0.093 0.381 9.293 156 Toumpa 4 I 2011 1.233 10.828 0.336 0.296 0.441 13.133 157 Toumpa 4 I 2011 1.060 9.909 0.352 0.132 0.368 11.821 158 Toumpa 3 I 2011 0.720 10.251 0.379 0.184 0.289 11.823 159 TCV 6 III 2011 1.222 9.232 0.326 0.169 0.464 11.413 160 TCV 5 III 2011 1.069 10.200 0.389 0.191 0.493 12.341 161 TCV 3 III 2011 0.819 10.650 0.478 0.176 0.361 12.485 162 Agrinion 3 III 2011 1.239 10.607 0.425 0.142 0.495 12.908 163 TCV 4 III 2011 0.608 8.321 0.342 0.165 0.296 9.731 164 Agrinion 6 III 2011 1.720 12.198 0.455 0.198 0.713 15.283 165 Agrinion 5 III 2011 1.365 12.007 0.393 0.170 0.635 14.570 166 Agrinion 4 III 2011 1.778 12.302 0.540 0.224 0.704 15.549

Hande Karaköse Jacobs University Bremen 121

Phytochemical Characterization of Stevia rebaudiana

Co-elution of steviol glycosides in the UV-chromatogram at 210 nm (LC-MS measurement with Knauer amino column):

3

2 1

Co-elution of rebaudioside A with stevioside or rebaudioside B:

Intens. -MS, 11.2min #670 x105 965.5 1.0 1

0.8

0.6

0.4 1001.4

0.2 300.2 803.4 196.0 401.2 518.2 0.0 200 400 600 800 1000 m/z Co-elution of rebaudioside C with dulcoside A:

Intens. -MS, 7.6min #452 x104 949.5 2

3

2

247.0 359.1 300.2 1 179.1 985.4 787.4 406.1 480.2 0 200 400 600 800 1000 m/z

Hande Karaköse Jacobs University Bremen 122

Phytochemical Characterization of Stevia rebaudiana

Co-elution of stevioside with steviolbioside:

Intens. -MS, 6.1min #366 x105 803.4 8 3

6

4

2 641.3

0 200 400 600 800 1000 m/z Extracted ion chromatogram of rebaudioside A (HILIC)

Intens. -MS, 27.0min #1612 x105 965.4

3

2

1 389.3

457.2 1011.4 1079.4 528.2 0 200 400 600 800 1000 m/z

Hande Karaköse Jacobs University Bremen 123

Phytochemical Characterization of Stevia rebaudiana

C. Tandem mass spectra of CGAs and flavonoid glycosides in negative ion mode

3-caffeoylquinic acid m/z 353 [M-H+]-

Intens. x10 7 -MS2(352.9) 190.7 1.0

0.5 134.8 0.0 x104 -MS3(353.1->190.6) 6 85.1 172.8 4 126.7 2 0 2 -MS4(353.1->190.7->85.4) 1 0 -1 200 400 600 800 1000 m/z

5-caffeoylquinic acid m/z 353 [M-H+]-

Intens. x10 7 -MS2(352.9) 6 190.7 4 2 0 x10 5 3 -MS3(353.1->190.7) 126.8 2 93.0 172.7 1 0 x104 -MS4(353.1->190.8->126.9) 1.0 83.4 0.5

0.0 200 400 600 800 1000 m/z

Hande Karaköse Jacobs University Bremen 124

Phytochemical Characterization of Stevia rebaudiana

4-caffeoylquinic acid m/z 353 [M-H+]-

Intens. -MS2(352.9) x107 172.7

0.5

0.05 x10 -MS3(353.1->172.7) 93.0 1.0

0.5 154.7 71.3 0.0 2 -MS4(353.1->172.7->93.1) 1 0 -1 200 400 600 800 1000 m/z Cis-5-caffeoylquinic acid m/z 353 [M-H+]-

Intens. -MS2(352.9) x10 6 190.7 4 2 0 x10 4 -MS3(353.1->190.7) 4 126.8 85.0 172.7 2

0 -MS4(353.1->190.7->126.8) 1000 108.9

500

0 200 400 600 800 1000 m/z

3,5-dicaffeoylquinic acid m/z 515

Intens. -MS2(515.0) x10 8 352.9 0.50 0.25 190.7 0.007 x10 -MS3(515.3->352.9) 190.7 2 1 134.7 0 x10 5 -MS4(515.3->353.1->190.7) 2 126.8 1 85.1 172.7 0 100 200 300 400 500 600 700 m/z

Hande Karaköse Jacobs University Bremen 125

Phytochemical Characterization of Stevia rebaudiana

4,5-dicaffeoylquinic acid m/z 515

Intens. -MS2(515.0) x10 8 352.9 0.50 0.25 172.7 202.7 254.8 298.9 0.007 x10 -MS3(515.3->352.9) 1.5 172.7 1.0 0.5 134.8 0.0 x10 5 -MS4(515.3->353.1->172.8) 4 93.0 110.8 2 71.3 154.7 0 100 200 300 400 500 600 700 m/z

Cis-4,5-dicaffeoylquinic acid m/z 515

Intens. -MS2(515.0) x10 7 0.75 352.9 0.50 0.25 172.7 202.8 0.00 x10 6 2.0 -MS3(515.2->352.9) 172.7 1.5 1.0 0.5 134.8 0.04 x10 -MS4(515.2->353.0->172.8) 2 93.0

1 110.9 71.3 154.7 0 100 200 300 400 500 600 700 m/z

Cis-4,5-dicaffeoylquinic acid m/z 515

Intens. x107 -MS2(515.0) 352.9 1.0 0.5 172.7 0.06 x10 -MS3(515.3->353.0) 2 172.7

1 134.8 0 x10 4 -MS4(515.3->353.1->172.8) 1.0 93.1 154.7 71.3 0.5 110.9 0.0 100 200 300 400 500 600 700 m/z

Hande Karaköse Jacobs University Bremen 126

Phytochemical Characterization of Stevia rebaudiana

Rutin m/z 609 [M-H+]-

Intens. -MS2(609.0) x10 7 1.0 300.8

0.5 299.8 178.6 270.8 342.9 0.06 x10 -MS3(609.4->300.6) 1.0 270.7 178.7 0.5 106.9 0.0 x104 -MS4(609.4->300.6->271.1) 2 243.7 1 184.7 0 200 400 600 800 1000 m/z

Quercetin-galactoside m/z 463 [M-H+]-

Intens. -MS2(462.9) x10 7 300.8

0.5 299.8 178.6 0.0 x10 6 -MS3(463.2->302.6) 178.7 0.75 270.8 0.50 0.25 106.9 0.00 x10 4 -MS4(463.2->300.7->178.7)

4 150.7 2 0 200 400 600 800 1000 m/z

Hande Karaköse Jacobs University Bremen 127

Phytochemical Characterization of Stevia rebaudiana

Kaempferol-glucopyranoside; Quercetin-rhamnoside; Kaempferol-glucopyranoside; Quercetin-rhamnoside m/z 447

Intens. -MS2(446.9) x107 284.8

1

05 x10 -MS3(447.1->284.9) 1.5 198.7 242.7 1.0 0.5 673.9 0.0 2000 -MS4(447.1->284.9->198.8) 1500 170.8 1000 500 0 200 400 600 800 1000 m/z

Intens. -MS2(446.9) x10 6 283.8 1.0 0.5 254.7 326.9 0.05 x10 -MS3(447.1->285.8) 254.7 2 1 150.7 226.7 0 4000 -MS4(447.1->284.3->255.2) 3000 226.8 2000 1000 0 200 400 600 800 1000 m/z

Intens. -MS2(447.0) x10 7 0.75 300.8 0.50 299.9 0.25 178.7 0.00 x10 5 -MS3(447.2->300.7) 6 178.6 4 150.7 2 270.7 106.8 0 x10 4 -MS4(447.2->300.7->178.8)

4 150.7 2 106.8 0 200 400 600 800 1000 m/z

Hande Karaköse Jacobs University Bremen 128

Phytochemical Characterization of Stevia rebaudiana

Kaempferol-rhamnopyranosyl-glucopyranoside(rutinoside) isomers; Quercetin- dirhamnoside; Apigenin-diglucoside/galactoside m/z 593

Intens. -MS2(446.9) x10 6 6 284.8 4 2 0 x10 6 -MS3(593.3->284.5) 0.75 254.7 0.50 0.25 150.7 0.00 x10 4 -MS4(593.3->284.5->255.3) 1.5 210.7 1.0 162.6 0.5 0.0 200 400 600 800 1000 m/z

Quercetin pentoside m/z 433

Intens. -MS2(432.9) x10 7 300.8

0.5

0.0 x10 6 -MS3(433.1->300.7) 178.7 0.75 150.7 0.50 0.25 270.7 106.9 0.004 x10 -MS4(433.1->300.8->178.7) 4 150.7 2 107.2 0 200 400 600 800 1000 m/z

Intens. -MS2(432.9) x10 7 299.8 0.5

178.7 0.06 x10 -MS3(433.2->300.5) 270.7 2

1 178.7

04 x10 -MS4(433.2->300.5->271.0) 2 242.7

1 186.7 0 200 400 600 800 1000 m/z

Hande Karaköse Jacobs University Bremen 129

Phytochemical Characterization of Stevia rebaudiana

Kaempferol-xylosyl-glucoside; Naringin m/z 579

Intens. -MS2(579.0) x10 6 299.8 4

2 178.6 414.9 05 x10 -MS3(579.3->300.1) 8 270.7 6 4 2 178.7 0 1500 -MS4(579.3->300.1->270.9) 243.8 1000 500 0 200 400 600 800 1000 m/z

Intens. -MS2(579.0) x10 6 3 299.8 2 1 178.6 342.9 414.9 489.0 560.9 05 x10 -MS3(579.3->300.0) 270.7 6 4 2 150.7 04 x10 -MS4(579.3->300.0->270.9) 0.75 226.7 0.50 0.25 198.6 0.00 200 400 600 800 1000 m/z

Apigenin-galactoside m/z 431

Intens. -MS2(430.9) x10 7 268.7

0.5

0.0 x10 4 -MS3(431.2->267.8) 6 224.8 4 148.8 2 267.6 0 -MS4(431.2->268.8->225.3) 1500 196.7 1000 500 168.7 0 200 400 600 800 1000 m/z

Hande Karaköse Jacobs University Bremen 130

Phytochemical Characterization of Stevia rebaudiana

Kaempferol-glucosylrhamnosyl-glucoside/galactoside m/z 755

Intens. -MS2(755.2) x106 593.0 3 2 1 469.0 284.8 367.0 05 x10 -MS3(755.7->593.3) 6 4 283.8 2 254.7 326.9 04 x10 -MS4(755.7->593.3->284.5) 6 254.7 4 2 150.7 0 200 400 600 800 1000 m/z

Quercetin-trisaccharide m/z 741

Intens. -MS2(741.2) x10 5 579.0

1 461.9 299.8 0 x10 4 -MS3(741.5->578.5) 299.8 3 2 1 354.9 414.9 0 -MS4(741.5->579.2->300.0) 750 270.7 500 250 0 200 400 600 800 1000 m/z Kaempferol 3-rhamnopyranosyl-rhamnopyranosyl-glucopyranoside m/z 739

Intens. x104 -MS2(739.2) 575.1 4 283.8 2 254.7 393.0 326.9 443.0 473.0 642.0 692.0 0 -MS3(739.5->575.2) 3000 339.0 2000 308.9 428.9 256.7 393.0 1000 282.9 338.1 212.7 547.1 162.8 0 -MS4(739.5->575.3->339.0) 200 295.8 262.8 100 0 100 200 300 400 500 600 700 m/z

Hande Karaköse Jacobs University Bremen 131

Phytochemical Characterization of Stevia rebaudiana

Quercetin-diglucoside-rhamnoside m/z 771

Intens. -MS2(771.2) x10 6 609.1 4 2 300.8 469.0 06 x10 -MS3(771.6->608.6) 1.5 1.0 299.8 0.5 270.8 342.9 0.05 x10 -MS4(771.6->609.4->300.9) 1.0 150.7 270.7 0.5

106.9 210.7 0.0 200 400 600 800 1000 m/z

Hande Karaköse Jacobs University Bremen 132

Phytochemical Characterization of Stevia rebaudiana

D. Quantification data of polyhenols in stevia

Minimum, average and maximum amounts (g/100g leaves) of polyphenols in all stevia samples (average values taken within +/- 3 σ)

3CQA 5CQA 4CQA totalmono 3,5 4,5 totaldiCQA k7g q3g totalflavones

min 0.002 0.193 0.010 0.205 0.173 0.210 0.311 0.092 0.001 0.329

ave 0.310 2.481 0.124 2.915 1.203 1.241 2.442 2.854 0.084 6.993

max 2.828 4.986 0.249 5.608 2.476 2.609 4.575 6.662 0.611 16.415

stddev 0.243 0.855 0.043 1.010 0.511 0.487 0.938 1.259 0.086 3.102

18.000 16.000 14.000 12.000 10.000 8.000 min 6.000 ave 4.000 max 2.000 0.000

Bar plot showing minimum, maximum and average values obtained from all 166 stevia samples within +/- 3 σ

Hande Karaköse Jacobs University Bremen 133

Quantity of CQAs and Flavonoid glycosides

3cqa 5cqa 4cqa Totalmono 3,5diCQA 4,5diCQA Total diCQA k7g q3g Total flav. Sample No. Origin Variety Harvesting Year ( g/100g) ( g/100g) ( g/100g) ( g/100g) ( g/100g) ( g/100g) ( g/100g) ( g/100g) ( g/100g) ( g/100g) 1 TCV 3 II 10.09.2010 0.139 1.847 0.092 2.078 0.804 0.624 1.428 0.787 0.087 1.641 2 TCV 4 II 14.09.2010 0.260 2.680 0.134 3.074 1.281 1.052 2.333 1.967 0.093 3.518 3 TCV 3 I 04.08.2010 0.325 2.900 0.145 3.369 0.512 1.155 1.667 1.541 0.085 3.041 4 TCV 1 I 04.08.2010 0.187 2.336 0.117 2.640 0.954 0.871 1.825 1.006 0.120 2.326 5 TCV 1 I 04.08.2010 0.232 2.197 0.110 2.538 0.735 0.814 1.549 0.609 0.136 1.529 6 TCV 2 I 04.08.2010 0.235 1.806 0.090 2.132 0.592 1.061 1.653 0.899 0.074 2.130 7 TCV 2 I 04.08.2010 0.198 1.961 0.098 2.256 0.670 0.959 1.630 0.987 0.107 2.208 8 TCV 4 I 04.08.2010 0.324 3.306 0.165 3.796 0.938 1.030 1.968 2.255 0.089 3.914 9 TCV 4 II 14.09.2010 0.223 2.911 0.146 3.280 1.413 0.997 2.410 2.290 0.073 4.009 10 TCV 2 II 28.09.2010 0.089 1.102 0.055 1.246 1.178 0.814 1.992 1.575 0.099 3.302 11 TCV 2 II 28.09.2010 0.051 0.593 0.030 0.673 0.514 0.488 1.002 0.582 0.110 1.180 12 Pojava 4 II 14.09.2010 0.221 2.103 0.105 2.429 0.778 0.993 1.771 1.735 0.080 2.696 13 TCV 3 I 04.08.2010 0.319 2.508 0.125 2.952 0.562 0.840 1.403 1.135 0.044 2.207 14 TCV 2 I 04.08.2010 0.253 1.876 0.094 2.223 0.537 0.899 1.437 0.767 0.110 1.507 15 TCV 1 I 01.09.2010 0.217 1.772 0.089 2.077 0.630 0.849 1.480 0.878 0.073 2.015 16 TCV 2 I 04.08.2010 0.232 2.058 0.103 2.393 0.557 0.791 1.348 0.772 0.145 1.627 17 TCV 3 I 04.08.2010 0.290 2.361 0.118 2.769 0.574 0.884 1.459 0.998 0.047 2.246 18 TCV 4 I 04.08.2010 0.317 2.057 0.103 2.477 0.519 0.853 1.372 1.416 0.059 2.481 19 TCV 3 II 10.09.2010 0.233 2.197 0.110 2.541 0.853 0.802 1.655 1.080 0.044 2.299 20 TCV 2 II 28.09.2010 0.136 0.729 0.036 0.901 0.907 1.046 1.953 1.125 0.053 2.522 21 TCV 1 II 28.09.2010 0.041 0.684 0.034 0.759 0.247 0.237 0.484 0.225 0.134 0.626 22 TCV 3 II 10.09.2010 0.276 2.361 0.118 2.755 0.922 1.130 2.052 0.471 0.090 1.103 23 TCV,Pojana 4 I 04.08.2010 0.279 2.615 0.131 3.024 0.492 0.795 1.287 1.210 0.142 2.022 24 TCV,pojana 3 I 04.08.2010 0.257 2.355 0.118 2.730 0.504 0.786 1.290 0.421 0.112 0.958 25 TCV 1 I 01.09.2010 0.198 1.615 0.081 1.894 0.473 0.653 1.126 0.450 0.107 1.097 26 TCV 4 II 14.09.2010 0.199 1.651 0.083 1.933 0.694 0.912 1.606 1.518 0.008 2.679 27 TCV 4 I 04.08.2010 0.282 2.223 0.111 2.617 0.435 0.694 1.128 1.303 0.079 2.252 28 TCV 2 II 28.09.2010 0.220 1.833 0.092 2.145 1.331 1.476 2.807 2.254 0.044 5.026

Hande Karaköse Jacobs University Bremen 134

Quantity of CQAs and Flavonoid glycosides

Sample No. Origin Variety Harvesting Year 3cqa 5cqa 4cqa totalmono 3,5diCQA 4,5diCQA total diCQA k7g q3g Total flav. 29 TCV 3 II 10.09.2010 0.260 2.531 0.127 2.917 0.700 0.884 1.585 1.269 0.094 2.604 30 Uconor 4 II 07.05.2010 0.249 1.630 0.081 1.961 0.606 1.000 1.606 0.798 0.106 3.707 31 Uconor 3 I 11.08.2010 0.380 3.068 0.153 3.601 1.890 1.966 3.856 4.745 0.036 11.645 32 Uconor 2 I 11.08.2010 0.253 2.320 0.116 2.689 2.014 2.561 4.575 0.889 0.415 4.503 33 Uconor 2 II 07.05.2010 0.161 1.751 0.088 2.000 2.144 1.921 4.066 1.828 0.287 6.359 34 Uconor 4 II 07.09.2010 0.341 4.344 0.217 4.903 2.129 1.544 3.673 3.495 0.265 7.592 35 Uconor 4 I 11.08.2010 0.337 4.067 0.203 4.607 1.413 1.735 3.148 2.509 0.611 8.135 36 Uconor 4 I 11.08.2010 0.262 3.586 0.179 4.027 1.190 1.987 3.178 1.943 0.445 5.796 37 Uconor 4 I 11.08.2010 0.181 1.931 0.097 2.209 0.804 0.714 1.518 2.227 0.118 6.225 38 Uconor 3 I 11.08.2010 0.347 3.096 0.155 3.598 0.887 1.108 1.995 2.644 0.521 7.170 39 Uconor 4 I 11.08.2010 0.314 3.034 0.152 3.500 1.687 1.668 3.355 4.876 0.060 13.156 40 Uconor 2 I 11.08.2010 0.291 3.099 0.155 3.544 2.476 1.916 4.392 2.894 0.079 8.897 41 Uconor 3 I 11.08.2010 0.215 1.692 0.085 1.992 0.938 1.028 1.966 2.158 0.025 5.087 42 Uconor 3 II 07.09.2010 0.364 2.681 0.134 3.179 1.803 1.342 3.145 3.136 0.064 7.190 43 Agrinion 2 II 20.09.2010 0.386 2.634 0.132 3.152 1.695 1.681 3.376 3.204 0.067 6.682 44 Toumpa 1 I 30.07.2010 0.348 3.086 0.154 3.589 2.276 2.251 4.526 6.662 0.079 16.415 45 Portugal 1 I 07.07.2010 0.117 0.908 0.045 1.071 0.273 0.569 0.842 2.883 0.096 9.525 46 Amfilia 1 I 04.08.2010 0.719 4.135 0.207 5.061 1.333 1.391 2.725 3.026 0.047 7.179 47 Toumpa 3 II 10.09.2010 0.423 2.908 0.145 3.477 1.498 1.362 2.861 3.991 0.010 11.230 48 Agrinion 2 I 09.08.2010 0.456 2.808 0.140 3.404 0.406 1.328 1.734 2.551 0.025 6.536 49 Toumpa 3 I 30.07.2010 0.413 3.810 0.191 4.414 1.311 1.583 2.893 3.599 0.037 9.241 50 Agrinion 4 I 09.08.2010 0.542 3.346 0.167 4.056 1.122 1.551 2.673 4.589 0.019 8.169 51 Agrinion 4 II 20.09.2010 0.118 1.355 0.068 1.541 0.726 0.844 1.570 4.043 0.079 7.890 52 Portugal 4 I 26.06.2010 0.267 1.706 0.085 2.059 0.637 0.460 1.098 3.828 0.023 8.718 53 Amfilia 4 I 04.08.2010 0.284 3.049 0.152 3.485 0.452 0.704 1.157 3.423 0.146 5.716 54 Amfilia 4 II 15.09.2010 0.260 2.315 0.116 2.691 0.647 0.842 1.489 4.584 0.086 7.711 55 Argentinie 2009 - - 2009 0.284 2.403 0.120 2.808 1.442 1.641 3.083 3.250 0.008 7.656 8.672 56 Paragan2009 - - 2009 0.380 3.021 0.151 3.551 1.842 1.574 3.417 4.226 0.041

Hande Karaköse Jacobs University Bremen 135

Quantity of CQAs and Flavonoid glycosides

Sample No. Origin Variety Harvesting Year 3cqa 5cqa 4cqa totalmono 3,5diCQA 4,5diCQA total diCQA k7g q3g Total flav. 57 Argentinien2010 - - 2010 0.288 2.138 0.107 2.533 1.357 1.427 2.784 3.420 0.009 8.073 58 Amfilia 2 II 15.09.2010 0.439 2.440 0.122 3.001 0.655 0.755 1.410 4.145 0.040 8.847 59 Agrinion 1 II 20.09.2010 0.123 1.341 0.067 1.531 1.035 1.245 2.280 3.638 0.043 7.423 60 Amfilia 2 I 04.08.2010 0.619 4.740 0.237 5.597 1.505 1.493 2.997 3.839 0.012 9.389 61 Amfilia 3 II 15.09.2010 0.407 2.143 0.107 2.658 0.914 1.110 2.024 3.792 0.113 8.905 62 Toumpa 1 I 30.07.2010 0.444 3.439 0.172 4.055 1.262 1.781 3.043 4.508 0.008 9.830 63 Amfilia 1 II 15.09.2010 0.317 1.932 0.097 2.346 0.642 0.714 1.356 4.884 0.112 15.769 64 Portugal 3 I 26.06.2010 0.223 2.125 0.106 2.454 0.754 0.738 1.492 4.666 0.035 13.528 65 Toumpa 1 II 10.09.2010 0.289 2.705 0.135 3.129 1.443 1.019 2.461 3.851 0.002 9.324 66 Agrinion 1 I 09.08.2010 0.374 3.483 0.174 4.031 1.249 1.487 2.737 4.356 0.078 9.066 67 Amfilia 3 I 04.08.2010 0.659 3.931 0.197 4.787 1.037 1.337 2.374 4.316 0.027 9.249 68 Toumpa 2 II 10.09.2010 0.366 3.010 0.151 3.527 1.053 0.914 1.967 2.577 0.058 6.943 69 Agrinio 3 II 20.09.2010 0.186 1.465 0.073 1.724 0.821 1.047 1.868 2.284 0.001 6.544 70 Toumpa 4 II 10.09.2010 0.442 3.975 0.199 4.615 1.331 0.868 2.200 3.638 0.088 7.843 71 Agrinion 3 I 09.08.2010 0.413 3.266 0.163 3.843 1.008 1.282 2.290 3.586 0.011 9.980 72 Toumpa 4 I 30.07.2010 0.591 3.768 0.188 4.547 1.522 2.381 3.903 2.278 0.087 4.869 73 Granada 2 I 15.09.2010 0.154 1.350 0.068 1.572 0.692 0.630 1.322 2.190 0.034 7.131 74 Granada 4 I 15.09.2010 0.160 1.962 0.098 2.220 0.369 0.559 0.928 4.286 0.063 8.816 75 Granada 3 I 09.09.2010 0.214 1.671 0.084 1.969 0.173 0.511 0.684 4.275 0.068 9.516 76 Equador - - - 0.015 0.308 0.015 0.338 0.325 0.210 0.535 0.099 0.135 0.350 77 DZ - - - 0.002 0.193 0.010 0.205 0.344 0.263 0.607 1.352 0.140 3.531 78 Kenia 2010 - - 2010 0.261 2.436 0.122 2.818 1.509 1.114 2.622 1.625 0.082 6.527 79 Paraguay 2009 - - 2009 0.450 2.772 0.139 3.361 2.033 1.895 3.928 4.301 0.036 10.225 80 Argentinien 2010 - - 2010 0.209 2.762 0.138 3.109 1.889 2.059 3.948 4.459 0.096 10.975 81 Indien 2010 - - 2010 0.122 1.322 0.066 1.510 0.790 1.757 2.547 2.175 0.122 8.091 82 Argenitinien 2009 - - 2009 0.354 2.488 0.124 2.966 2.019 2.526 4.546 6.294 0.005 13.673

83 Hohenheim 2010 - - 2010 0.123 1.133 0.057 1.313 0.266 0.450 0.716 3.298 0.030 7.328

Hande Karaköse Jacobs University Bremen 136

Quantity of CQAs and Flavonoid glycosides

Sample No. Origin Variety Harvesting Year 3cqa 5cqa 4cqa totalmono 3,5diCQA 4,5diCQA total diCQA k7g q3g Total flav. 84 Krim - - - 0.033 1.183 0.059 1.275 0.549 0.238 0.311 2.969 0.191 7.441 85 Turkei PS1 2010 - - 2010 0.089 1.626 0.081 1.796 0.774 1.066 1.840 3.177 0.174 8.623 86 Turkei PS2 2010 - - 2010 0.120 1.917 0.096 2.132 0.496 0.900 1.395 2.246 0.294 5.762 87 Uconor 7 I 13.07.2011 0.365 3.645 0.182 4.193 1.786 2.045 3.831 5.159 0.026 9.928 88 Agrinon 4 I 2011 0.431 3.399 0.170 4.000 1.203 1.461 2.664 1.512 0.067 3.939 89 Uconor 4 I 2011 0.473 3.279 0.164 3.916 1.720 1.900 3.620 3.915 0.054 9.788 90 Agrinion 3 I 2011 0.398 3.040 0.152 3.590 1.245 1.436 2.681 3.913 0.056 10.088 91 Agrinion 5 I 2011 0.463 3.208 0.160 3.832 1.174 1.400 2.574 3.030 0.047 8.126 92 Uconor 6 I 13.07.2011 0.561 4.671 0.234 5.465 1.947 2.290 4.237 0.092 0.048 0.329 93 Uconor 3 I 13.07.2011 0.372 4.986 0.249 5.608 1.664 2.022 3.686 3.102 0.088 7.721 94 Uconor 5 I 13.07.2011 0.520 3.282 0.164 3.967 1.425 1.588 3.013 1.764 0.141 5.377 95 Agrinion 6 I 2011 2.828 1.225 0.061 4.114 1.592 1.956 3.548 2.220 0.111 6.236 96 Toumpa 1 I 2011 0.479 3.600 0.180 4.260 1.321 1.578 2.899 3.210 0.059 8.628 97 Toumpa 2 I 2011 0.495 3.203 0.160 3.858 0.909 1.137 2.046 3.659 0.061 10.518 98 Toumpa 3 I 2011 0.491 4.099 0.205 4.795 1.208 1.535 2.743 3.972 0.002 10.938 99 Toumpa 4 I 2011 0.461 2.879 0.144 3.484 1.883 2.052 3.935 3.709 0.091 7.981 100 Amiflikeia 4 II 2011 0.595 3.762 0.188 4.544 1.193 1.435 2.628 3.610 0.058 7.079 101 Amiflikeia 5 II 2011 0.434 2.756 0.138 3.328 0.968 1.410 2.378 3.248 0.163 7.859 102 Amiflikeia 4 II 2011 0.371 2.363 0.118 2.852 1.528 1.556 3.084 3.852 0.092 7.591 103 Amiflikeia 3 II 2011 0.479 2.814 0.141 3.433 1.623 1.790 3.413 3.248 0.022 7.671 104 Amiflikeia - II 2011 0.458 2.723 0.136 3.317 1.125 1.161 2.286 3.309 0.123 7.305 105 Amiflikeia 6 II 2011 0.451 2.574 0.129 3.154 1.099 1.166 2.264 2.884 0.073 7.167 106 Amiflikeia 1 II 2011 0.535 2.200 0.110 2.845 1.349 1.416 2.765 2.641 0.019 6.327 107 APTTB 3 I 2011 0.246 1.663 0.083 1.992 0.721 0.603 1.324 3.675 0.041 9.730 108 APTTB 5 I 2011 0.192 1.249 0.062 1.503 0.705 0.703 1.409 3.821 0.096 11.274 109 APTTB 6 I - 0.180 1.177 0.059 1.416 0.756 0.701 1.457 4.118 0.018 10.068 110 APTTB 4 I - 0.206 1.362 0.068 1.637 0.800 0.803 1.603 3.516 0.055 10.562 111 Uconor 5 II 2011 0.237 2.303 0.115 2.656 1.837 1.190 3.027 3.066 0.009 8.264

Hande Karaköse Jacobs University Bremen 137

Quantity of CQAs and Flavonoid glycosides

Sample No. Origin Variety Harvesting Year 3cqa 5cqa 4cqa totalmono 3,5diCQA 4,5diCQA total diCQA k7g q3g Total flav. 112 Uconor 4 II 2011 0.293 2.063 0.103 2.459 1.406 1.275 2.681 3.578 0.035 7.598 113 Uconor 3 II 2011 0.300 3.108 0.155 3.564 1.553 1.525 3.077 2.840 0.054 6.933 114 Uconor 6 II 2011 0.301 2.170 0.109 2.580 1.482 1.228 2.711 3.110 0.025 8.264 115 Conaga 3 I 2011 0.356 2.621 0.131 3.107 0.901 1.506 2.408 3.041 0.030 7.221 116 Conaga 5 I 2011 0.250 2.422 0.121 2.792 1.216 1.422 2.638 2.195 0.097 6.593 117 Conaga 4 I 2011 0.354 2.309 0.115 2.778 1.045 1.524 2.569 3.227 0.067 7.027 118 Agrinion 5 II 2011 0.368 2.410 0.121 2.899 1.495 1.379 2.874 3.227 0.051 9.412 119 Agrinion 4 II 2011 0.358 3.144 0.157 3.659 1.433 1.514 2.947 3.564 0.068 7.308 120 Conaga 6 I 2011 0.326 3.301 0.165 3.792 1.295 2.068 3.362 3.428 0.088 10.269 121 Conaga 7 I 2011 0.262 2.747 0.137 3.147 0.507 1.189 1.696 0.637 0.225 2.064 122 Agrinion 6 II 2011 0.296 2.461 0.123 2.880 1.434 1.364 2.798 2.788 0.009 7.335 123 Agrinion 3 II 2011 0.247 2.626 0.131 3.004 1.570 1.342 2.913 4.336 0.019 9.887 124 Toumpa 7 I 2011 0.061 2.240 0.112 2.414 1.528 0.575 2.103 1.873 0.140 5.930 125 Toumpa 4 III 2011 0.148 1.884 0.094 2.126 1.311 1.310 2.621 3.680 0.060 8.463 126 Toumpa 3 III 2011 0.187 1.961 0.098 2.246 1.684 1.567 3.251 3.510 0.018 9.388 127 Toumpa 3 II 2011 0.175 2.490 0.124 2.789 1.650 1.271 2.921 3.769 0.013 9.366 128 Agrinion 4 II 2011 0.246 3.222 0.161 3.630 1.500 1.460 2.960 3.386 0.005 6.759 129 Toumpa 4 II 2011 0.155 2.863 0.143 3.161 1.725 1.426 3.151 4.079 0.142 8.424 130 Toumpa 3 II 2011 0.140 2.078 0.104 2.322 1.573 1.464 3.037 3.937 0.025 10.501 131 Toumpa 4 II 2011 0.112 2.289 0.114 2.515 1.321 1.288 2.609 4.684 0.057 9.631 132 Agrinion 3 II 2011 0.258 2.593 0.130 2.980 1.925 1.922 3.846 3.585 0.019 8.969 133 Toumpa 5 II 2011 0.134 2.375 0.119 2.628 1.732 1.454 3.186 3.827 0.092 11.186 134 Toumpa 6 II 2011 0.130 2.536 0.127 2.792 1.445 0.953 2.398 3.424 0.008 9.629 135 Toumpa 7 II 2011 0.116 2.285 0.114 2.516 0.850 0.657 1.507 1.442 0.179 4.449 136 TCV 5 I 30.06.2011 0.531 3.343 0.167 4.042 1.328 1.185 2.513 2.618 0.076 8.134 137 TCV 3 I 30.06.2011 0.441 3.153 0.158 3.752 1.578 1.266 2.844 2.870 0.034 7.671 138 TCV 6 I 30.06.2011 0.526 4.032 0.202 4.760 1.807 1.394 3.201 3.718 0.031 10.049 139 TCV 3 II 11.08.2011 0.267 2.692 0.135 3.094 1.854 1.155 3.010 3.503 0.068 8.909

Hande Karaköse Jacobs University Bremen 138

Quantity of CQAs and Flavonoid glycosides

Sample No. Origin Variety Harvesting Year 3cqa 5cqa 4cqa totalmono 3,5diCQA 4,5diCQA total diCQA k7g q3g Total flav. 140 TCV 6 II 17.08.2011 0.289 2.399 0.120 2.808 1.679 1.520 3.200 3.105 0.062 8.506 141 TCV 4 II 24.08.2011 0.376 2.375 0.119 2.869 1.563 2.609 4.171 2.493 0.192 5.320 142 TCV 7 I 18.08.2011 0.084 1.777 0.089 1.950 1.715 1.412 3.127 1.711 0.208 4.790 143 TCV 4 I 07.07.2011 0.340 1.843 0.092 2.275 1.070 1.232 2.302 2.974 0.009 6.305 144 TCV 5 II 17.08.2011 0.385 2.323 0.116 2.824 1.933 2.474 4.407 2.781 0.031 7.833 145 Amiflikeia 3 I 2011 0.501 2.934 0.147 3.582 0.433 0.928 1.360 2.114 0.075 5.301 146 Amiflikeia 4 I 2011 0.554 3.595 0.180 4.329 1.182 0.989 2.170 2.013 0.128 4.423 147 Amiflikeia 5 I 2011 0.628 3.086 0.154 3.869 1.326 1.124 2.450 2.036 0.001 5.264 148 Amiflikeia 6 I 2011 0.524 3.398 0.170 4.092 1.353 0.735 2.089 2.035 0.044 5.307 149 Amiflikeia 1 I 2011 0.485 3.690 0.184 4.359 1.836 1.674 3.510 2.939 0.052 8.505 150 Amiflikeia 2 I 2011 0.399 2.798 0.140 3.337 2.183 1.078 3.262 2.966 0.048 7.316 151 Amiflikeia 4 I 2011 0.358 2.465 0.123 2.946 1.903 1.234 3.138 4.101 0.011 8.871 152 Amiflikeia 3 I 2011 0.335 2.154 0.108 2.596 1.842 1.119 2.961 3.351 0.086 8.778 153 Toumpa 3 I 2011 0.307 2.097 0.105 2.508 1.939 1.677 3.616 3.622 0.001 10.033 154 Toumpa 5 I 2011 0.242 3.266 0.163 3.672 1.694 1.281 2.975 2.500 0.099 6.702 155 Toumpa 6 I 2011 0.127 2.546 0.127 2.800 1.054 0.770 1.824 1.953 0.068 5.836 156 Toumpa 4 I 2011 0.242 2.434 0.122 2.797 2.178 1.794 3.972 4.547 0.092 9.592 157 Toumpa 4 I 2011 0.307 2.551 0.128 2.985 1.560 1.357 2.918 3.952 0.065 8.032 158 Toumpa 3 I 2011 0.171 2.763 0.138 3.072 1.485 0.871 2.356 3.479 0.095 8.160 159 TCV 6 III 2011 0.063 1.872 0.094 2.029 1.716 1.789 3.505 3.010 0.036 7.856 160 TCV 5 III 2011 0.147 1.552 0.078 1.777 1.622 1.176 2.798 2.464 0.091 6.536 161 TCV 3 III 2011 0.161 1.777 0.089 2.026 0.987 1.077 2.064 2.300 0.157 5.554 162 Agrinion 3 III 2011 0.168 1.372 0.069 1.608 0.974 0.938 1.912 3.132 0.041 8.298 163 TCV 4 III 2011 0.120 2.087 0.104 2.311 1.110 0.883 1.993 2.346 0.194 4.654 164 Agrinion 6 III 2011 0.148 2.041 0.102 2.291 1.489 1.147 2.636 4.011 0.006 10.699 165 Agrinion 5 III 2011 0.124 1.897 0.095 2.116 1.269 0.794 2.063 3.501 0.064 9.157 166 Agrinion 4 III 2011 0.179 1.837 0.092 2.108 0.982 0.865 1.847 4.156 0.118 8.261

Hande Karaköse Jacobs University Bremen 139

Quantity of Trans & Cis-CQAs

3cqa 5cqa 4cqa Cis-5CQA 3,5diCQA 4,5diCQA Cis-4,5diCQA Cis-4,5diCQA Sample No. Origin Variety Harvesting ( g/100g) ( g/100g) ( g/100g) (g/100g) ( g/100g) ( g/100g) (g/100g) (g/100g) 1 TCV 3 II 0.139 1.847 0.092 0.313 0.804 0.624 0.002 0.022 2 TCV 4 II 0.260 2.680 0.134 0.516 1.281 1.052 0.011 0.037 3 TCV 3 I 0.325 2.900 0.145 1.153 0.512 1.155 0.009 0.035 4 TCV 1 I 0.187 2.336 0.117 0.060 0.954 0.871 0.009 0.045 5 TCV 1 I 0.232 2.197 0.110 0.518 0.735 0.814 0.006 0.024 6 TCV 2 I 0.235 1.806 0.090 0.590 0.592 1.061 0.009 0.030 7 TCV 2 I 0.198 1.961 0.098 0.593 0.670 0.959 0.007 0.028 8 TCV 4 I 0.324 3.306 0.165 1.123 0.938 1.030 0.016 0.031 9 TCV 4 II 0.223 2.911 0.146 0.657 1.413 0.997 0.006 0.026 10 TCV 2 II 0.089 1.102 0.055 0.121 1.178 0.814 0.006 0.018 11 TCV 2 II 0.051 0.593 0.030 0.023 0.514 0.488 0.023 0.057 12 Pojava 4 II 0.221 2.103 0.105 0.112 0.778 0.993 0.006 0.009 13 TCV 3 I 0.319 2.508 0.125 0.102 0.562 0.840 0.035 0.061 14 TCV 2 I 0.253 1.876 0.094 0.110 0.537 0.899 0.010 0.013 15 TCV 1 I 0.217 1.772 0.089 0.081 0.630 0.849 0.009 0.013 16 TCV 2 I 0.232 2.058 0.103 0.076 0.557 0.791 0.059 0.089 17 TCV 3 I 0.290 2.361 0.118 0.126 0.574 0.884 0.033 0.043 18 TCV 4 I 0.317 2.057 0.103 0.134 0.519 0.853 0.008 0.018 19 TCV 3 II 0.233 2.197 0.110 0.056 0.853 0.802 0.034 0.048 20 TCV 2 II 0.136 0.729 0.036 0.042 0.907 1.046 0.005 0.012 21 TCV 1 II 0.041 0.684 0.034 0.001 0.247 0.237 0.020 0.028 22 TCV 3 II 0.276 2.361 0.118 0.110 0.922 1.130 0.010 0.005 23 TCV,Pojana 4 I 0.279 2.615 0.131 0.100 0.492 0.795 0.051 0.078 24 TCV,pojana 3 I 0.257 2.355 0.118 0.101 0.504 0.786 0.013 0.024 25 TCV 1 I 0.198 1.615 0.081 0.065 0.473 0.653 0.037 0.051 26 TCV 4 II 0.199 1.651 0.083 0.073 0.694 0.912 0.007 0.018 27 TCV 4 I 0.282 2.223 0.111 0.113 0.435 0.694 0.014 0.013 28 TCV 2 II 0.220 1.833 0.092 0.070 1.331 1.476 0.010 0.014

Hande Karaköse Jacobs University Bremen 140

Quantity of Trans & Cis-CQAs

Cis 4,5-diCQA Cis 4,5-diCQA Sample No. Origin Variety Harvesting 3cqa 5cqa 4cqa cis 5CQA 3,5diCQA 4,5diCQA (g/100g) (g/100g) 29 TCV 3 II 0.260 2.531 0.127 0.090 0.700 0.884 0.018 0.018 30 Uconor 4 II 0.249 1.630 0.081 0.075 0.606 1.000 0.004 0.004 31 Uconor 3 I 0.380 3.068 0.153 0.021 1.890 1.966 0.011 0.011 32 Uconor 2 I 0.253 2.320 0.116 0.037 2.014 2.561 0.023 0.023 33 Uconor 2 II 0.161 1.751 0.088 0.015 2.144 1.921 0.026 0.026 34 Uconor 4 II 0.341 4.344 0.217 0.043 2.129 1.544 0.035 0.035 35 Uconor 4 I 0.337 4.067 0.203 0.067 1.413 1.735 0.006 0.006 36 Uconor 4 I 0.262 3.586 0.179 0.019 1.190 1.987 0.008 0.008 37 Uconor 4 I 0.181 1.931 0.097 0.017 0.804 0.714 0.009 0.009 38 Uconor 3 I 0.347 3.096 0.155 0.023 0.887 1.108 0.016 0.016 39 Uconor 4 I 0.314 3.034 0.152 0.013 1.687 1.668 0.007 0.007 40 Uconor 2 I 0.291 3.099 0.155 0.018 2.476 1.916 0.016 0.016 41 Uconor 3 I 0.215 1.692 0.085 0.029 0.938 1.028 0.013 0.013 42 Uconor 3 II 0.364 2.681 0.134 0.198 1.803 1.342 0.017 0.017 43 Agrinion 2 II 0.386 2.634 0.132 0.138 1.695 1.681 0.045 0.045 44 Toumpa 1 I 0.348 3.086 0.154 0.006 2.276 2.251 0.026 0.026 45 Portugal 1 I 0.117 0.908 0.045 0.028 0.273 0.569 0.004 0.004 46 Amfilia 1 I 0.719 4.135 0.207 0.242 1.333 1.391 0.044 0.044 47 Toumpa 3 II 0.423 2.908 0.145 0.188 1.498 1.362 0.017 0.017 48 Agrinion 2 I 0.456 2.808 0.140 0.128 0.406 1.328 0.029 0.029 49 Toumpa 3 I 0.413 3.810 0.191 0.006 1.311 1.583 0.025 0.025 50 Agrinion 4 I 0.542 3.346 0.167 0.204 1.122 1.551 0.029 0.029 51 Agrinion 4 II 0.118 1.355 0.068 0.040 0.726 0.844 0.018 0.009 52 Portugal 4 I 0.267 1.706 0.085 0.048 0.637 0.460 0.009 0.008 53 Amfilia 4 I 0.284 3.049 0.152 0.113 0.452 0.704 0.009 0.009 54 Amfilia 4 II 0.260 2.315 0.116 0.085 0.647 0.842 0.040 0.018 55 Argentinie 2009 - - 0.284 2.403 0.120 0.038 1.442 1.641 0.005 0.017 56 Paragan2009 - - 0.380 3.021 0.151 0.108 1.842 1.574 0.022 0.024

Hande Karaköse Jacobs University Bremen 141

Quantity of Trans & Cis-CQAs

Cis 4,5-diCQA Cis 4,5-diCQA Sample No. Origin Variety Harvesting 3cqa 5cqa 4cqa cis 5CQA 3,5diCQA 4,5diCQA (g/100g) (g/100g) 57 Argentinien2010 - - 0.288 2.138 0.107 0.035 1.357 1.427 0.013 0.042 58 Amfilia 2 II 0.439 2.440 0.122 0.107 0.655 0.755 0.012 0.014 59 Agrinion 1 II 0.123 1.341 0.067 0.002 1.035 1.245 0.005 0.025 60 Amfilia 2 I 0.619 4.740 0.237 0.229 1.505 1.493 0.016 0.016 61 Amfilia 3 II 0.407 2.143 0.107 0.087 0.914 1.110 0.017 0.051 62 Toumpa 1 I 0.444 3.439 0.172 0.104 1.262 1.781 0.022 0.025 63 Amfilia 1 II 0.317 1.932 0.097 0.041 0.642 0.714 0.007 0.021 64 Portugal 3 I 0.223 2.125 0.106 0.042 0.754 0.738 0.008 0.012 65 Toumpa 1 II 0.289 2.705 0.135 0.127 1.443 1.019 0.006 0.012 66 Agrinion 1 I 0.374 3.483 0.174 0.155 1.249 1.487 0.038 0.032 67 Amfilia 3 I 0.659 3.931 0.197 0.078 1.037 1.337 0.013 0.004 68 Toumpa 2 II 0.366 3.010 0.151 0.093 1.053 0.914 0.010 0.003 69 Agrinio 3 II 0.186 1.465 0.073 0.036 0.821 1.047 0.008 0.020 70 Toumpa 4 II 0.442 3.975 0.199 0.008 1.331 0.868 0.015 0.016 71 Agrinion 3 I 0.413 3.266 0.163 0.174 1.008 1.282 0.021 0.056 72 Toumpa 4 I 0.591 3.768 0.188 0.093 1.522 2.381 0.042 0.114 73 Granada 2 I 0.154 1.350 0.068 0.043 0.692 0.630 0.009 0.017 74 Granada 4 I 0.160 1.962 0.098 0.058 0.369 0.559 0.001 0.004 75 Granada 3 I 0.214 1.671 0.084 0.024 0.173 0.511 0.007 0.013 76 Equador - - 0.015 0.308 0.015 0.005 0.325 0.210 #DIV/0! #DIV/0! 77 DZ - - 0.002 0.193 0.010 0.001 0.344 0.263 #DIV/0! #DIV/0! 78 Kenia 2010 - - 0.261 2.436 0.122 0.030 1.509 1.114 0.009 0.029 79 Paraguay 2009 - - 0.450 2.772 0.139 0.082 2.033 1.895 0.060 0.092 80 Argentinien 2010 - - 0.209 2.762 0.138 0.004 1.889 2.059 0.025 0.059 81 Indien 2010 - - 0.122 1.322 0.066 0.022 0.790 1.757 0.032 0.063 82 Argenitinien 2009 - - 0.354 2.488 0.124 0.022 2.019 2.526 0.011 0.030

83 Hohenheim 2010 - - 0.123 1.133 0.057 0.027 0.266 0.450 0.000 0.001

Hande Karaköse Jacobs University Bremen 142

Quantity of Trans & Cis-CQAs

Cis 4,5-diCQA Cis 4,5-diCQA Sample No. Origin Variety Harvesting 3cqa 5cqa 4cqa cis 5CQA 3,5diCQA 4,5diCQA (g/100g) (g/100g) 84 Krim - - 0.033 1.183 0.059 0.020 0.549 0.238 -0.015 0.000 85 Turkei PS1 2010 - - 0.089 1.626 0.081 0.023 0.774 1.066 0.008 0.040 86 Turkei PS2 2010 - - 0.120 1.917 0.096 0.054 0.496 0.900 0.019 0.018 87 Uconor 7 I 0.365 3.645 0.182 0.182 1.786 2.045 0.010 0.014 88 Agrinon 4 I 0.431 3.399 0.170 0.129 1.203 1.461 0.023 0.013 89 Uconor 4 I 0.473 3.279 0.164 0.177 1.720 1.900 0.011 0.018 90 Agrinion 3 I 0.398 3.040 0.152 0.009 1.245 1.436 0.007 0.012 91 Agrinion 5 I 0.463 3.208 0.160 0.158 1.174 1.400 0.015 0.017 92 Uconor 6 I 0.561 4.671 0.234 0.251 1.947 2.290 0.022 0.007 93 Uconor 3 I 0.372 4.986 0.249 0.208 1.664 2.022 0.038 0.013 94 Uconor 5 I 0.520 3.282 0.164 0.196 1.425 1.588 0.050 0.065 95 Agrinion 6 I 2.828 1.225 0.061 0.132 1.592 1.956 0.014 0.002 96 Toumpa 1 I 0.479 3.600 0.180 0.163 1.321 1.578 0.034 0.041 97 Toumpa 2 I 0.495 3.203 0.160 0.148 0.909 1.137 0.048 0.146 98 Toumpa 3 I 0.491 4.099 0.205 0.175 1.208 1.535 0.029 0.075 99 Toumpa 4 I 0.461 2.879 0.144 0.183 1.883 2.052 0.038 0.094 100 Amiflikeia 4 II 0.595 3.762 0.188 0.008 1.193 1.435 0.024 0.033 101 Amiflikeia 5 II 0.434 2.756 0.138 0.149 0.968 1.410 0.005 0.013 102 Amiflikeia 4 II 0.371 2.363 0.118 0.128 1.528 1.556 0.022 0.059 103 Amiflikeia 3 II 0.479 2.814 0.141 0.145 1.623 1.790 0.045 0.112 104 Amiflikeia - II 0.458 2.723 0.136 0.140 1.125 1.161 0.033 0.027 105 Amiflikeia 6 II 0.451 2.574 0.129 0.138 1.099 1.166 0.007 0.010 106 Amiflikeia 1 II 0.535 2.200 0.110 0.128 1.349 1.416 0.030 0.105 107 APTTB 3 I 0.246 1.663 0.083 0.064 0.721 0.603 0.003 0.001 108 APTTB 5 I 0.192 1.249 0.062 0.058 0.705 0.703 0.008 0.018 109 APTTB 6 I 0.180 1.177 0.059 0.053 0.756 0.701 0.010 0.034 110 APTTB 4 I 0.206 1.362 0.068 0.057 0.800 0.803 0.014 0.011 111 Uconor 5 II 0.237 2.303 0.115 0.122 1.837 1.190 0.022 0.040

Hande Karaköse Jacobs University Bremen 143

Quantity of Trans & Cis-CQAs

Cis 4,5-diCQA Cis 4,5-diCQA Sample No. Origin Variety Harvesting 3cqa 5cqa 4cqa cis 5CQA 3,5diCQA 4,5diCQA (g/100g) (g/100g) 112 Uconor 4 II 0.293 2.063 0.103 0.136 1.406 1.275 0.029 0.058 113 Uconor 3 II 0.300 3.108 0.155 0.146 1.553 1.525 0.029 0.026 114 Uconor 6 II 0.301 2.170 0.109 0.113 1.482 1.228 0.020 0.043 115 Conaga 3 I 0.356 2.621 0.131 0.138 0.901 1.506 0.028 0.018 116 Conaga 5 I 0.250 2.422 0.121 0.157 1.216 1.422 0.018 0.059 117 Conaga 4 I 0.354 2.309 0.115 0.088 1.045 1.524 0.043 0.042 118 Agrinion 5 II 0.368 2.410 0.121 0.135 1.495 1.379 0.021 0.043 119 Agrinion 4 II 0.358 3.144 0.157 0.125 1.433 1.514 0.033 0.074 120 Conaga 6 I 0.326 3.301 0.165 0.004 1.295 2.068 0.104 0.166 121 Conaga 7 I 0.262 2.747 0.137 0.177 0.507 1.189 0.023 0.021 122 Agrinion 6 II 0.296 2.461 0.123 0.103 1.434 1.364 0.025 0.032 123 Agrinion 3 II 0.247 2.626 0.131 0.126 1.570 1.342 0.012 0.011 124 Toumpa 7 I 0.061 2.240 0.112 0.099 1.528 0.575 0.013 0.007 125 Toumpa 4 III 0.148 1.884 0.094 0.094 1.311 1.310 0.055 0.048 126 Toumpa 3 III 0.187 1.961 0.098 0.082 1.684 1.567 0.019 0.064 127 Toumpa 3 II 0.175 2.490 0.124 0.108 1.650 1.271 0.018 0.007 128 Agrinion 4 II 0.246 3.222 0.161 0.197 1.500 1.460 0.048 0.095 129 Toumpa 4 II 0.155 2.863 0.143 0.113 1.725 1.426 0.028 0.024 130 Toumpa 3 II 0.140 2.078 0.104 0.112 1.573 1.464 0.036 0.033 131 Toumpa 4 II 0.112 2.289 0.114 0.100 1.321 1.288 0.014 0.022 132 Agrinion 3 II 0.258 2.593 0.130 0.106 1.925 1.922 0.032 0.080 133 Toumpa 5 II 0.134 2.375 0.119 0.006 1.732 1.454 0.011 0.009 134 Toumpa 6 II 0.130 2.536 0.127 0.145 1.445 0.953 0.025 0.015 135 Toumpa 7 II 0.116 2.285 0.114 0.155 0.850 0.657 0.020 0.004 136 TCV 5 I 0.531 3.343 0.167 0.121 1.328 1.185 0.027 0.027 137 TCV 3 I 0.441 3.153 0.158 0.135 1.578 1.266 0.033 0.029 138 TCV 6 I 0.526 4.032 0.202 0.164 1.807 1.394 0.033 0.029 139 TCV 3 II 0.267 2.692 0.135 0.147 1.854 1.155 0.008 0.015

Hande Karaköse Jacobs University Bremen 144

Quantity of Trans & Cis-CQAs

Cis 4,5-diCQA Cis 4,5-diCQA Sample No. Origin Variety Harvesting 3cqa 5cqa 4cqa cis 5CQA 3,5diCQA 4,5diCQA (g/100g) (g/100g) 140 TCV 6 II 0.289 2.399 0.120 0.131 1.679 1.520 0.029 0.091 141 TCV 4 II 0.376 2.375 0.119 0.084 1.563 2.609 0.048 0.116 142 TCV 7 I 0.084 1.777 0.089 0.112 1.715 1.412 0.023 0.038 143 TCV 4 I 0.340 1.843 0.092 0.120 1.070 1.232 0.022 0.028 144 TCV 5 II 0.385 2.323 0.116 0.004 1.933 2.474 0.024 0.039 145 Amiflikeia 3 I 0.501 2.934 0.147 0.192 0.433 0.928 0.001 0.024 146 Amiflikeia 4 I 0.554 3.595 0.180 0.206 1.182 0.989 0.021 0.034 147 Amiflikeia 5 I 0.628 3.086 0.154 0.199 1.326 1.124 0.017 0.007 148 Amiflikeia 6 I 0.524 3.398 0.170 0.189 1.353 0.735 0.029 0.012 149 Amiflikeia 1 I 0.485 3.690 0.184 0.093 1.836 1.674 0.032 0.040 150 Amiflikeia 2 I 0.399 2.798 0.140 0.115 2.183 1.078 0.012 0.000 151 Amiflikeia 4 I 0.358 2.465 0.123 0.174 1.903 1.234 0.008 0.017 152 Amiflikeia 3 I 0.335 2.154 0.108 0.115 1.842 1.119 0.014 0.014 153 Toumpa 3 I 0.307 2.097 0.105 0.094 1.939 1.677 0.027 0.034 154 Toumpa 5 I 0.242 3.266 0.163 0.138 1.694 1.281 0.007 0.007 155 Toumpa 6 I 0.127 2.546 0.127 0.127 1.054 0.770 0.008 0.002 156 Toumpa 4 I 0.242 2.434 0.122 0.122 2.178 1.794 0.008 0.008 157 Toumpa 4 I 0.307 2.551 0.128 0.166 1.560 1.357 0.021 0.021 158 Toumpa 3 I 0.171 2.763 0.138 0.117 1.485 0.871 0.007 0.007 159 TCV 6 III 0.063 1.872 0.094 0.006 1.716 1.789 0.030 0.030 160 TCV 5 III 0.147 1.552 0.078 0.008 1.622 1.176 0.024 0.024 161 TCV 3 III 0.161 1.777 0.089 0.084 0.987 1.077 0.008 0.008 162 Agrinion 3 III 0.168 1.372 0.069 0.099 0.974 0.938 0.017 0.017 163 TCV 4 III 0.120 2.087 0.104 0.084 1.110 0.883 0.011 0.011 164 Agrinion 6 III 0.148 2.041 0.102 0.099 1.489 1.147 0.028 0.028 165 Agrinion 5 III 0.124 1.897 0.095 0.107 1.269 0.794 0.015 0.015 166 Agrinion 4 III 0.179 1.837 0.092 0.093 0.982 0.865 0.011 0.011

Hande Karaköse Jacobs University Bremen 145

Phytochemical Characterization of Stevia rebaudiana

E. Correlation & data distribution of CQAs

Correlation table of cis and trans CQAs cis5CQA cis45diCQa1 cis45diCQa2 cqa5 diCQA45

Correlation Coefficient 1.000 0.265** 0.183* -0.111 -0.075 cis5CQA Sig. (2-tailed) . 0.003 0.041 0.217 0.407 N 126 126 126 126 126 Correlation Coefficient 0.265** 1.000 0.731** -0.012 -0.056 cis45diCQa1 Sig. (2-tailed) 0.003 . 0.000 0.892 0.532 N 126 126 126 126 126 * ** Spearman's Correlation Coefficient 0.183 0.731 1.000 0.035 -0.026 rho cis45diCQa2 Sig. (2-tailed) 0.041 0.000 . 0.694 0.771 N 126 126 126 126 126 Correlation Coefficient -0.111 -0.012 0.035 1.000 0.586** cqa5 Sig. (2-tailed) 0.217 0.892 0.694 . 0.000 N 126 126 126 126 126 Correlation Coefficient -0.075 -0.056 -0.026 0.586** 1.000 diCQA45 Sig. (2-tailed) 0.407 0.532 0.771 0.000 . N 126 126 126 126 126 **. Correlation is significant at the 0.01 level (2-tailed). *. Correlation is significant at the 0.05 level (2-tailed).

Hande Karaköse Jacobs University Bremen 146

Phytochemical Characterization of Stevia rebaudiana

F. Structures of literature reported terpenes in Stevia rebaudiana

OH

OH O

myrtenol myrtenal pinocarveol alpha pinene beta pinene

OH OH

sabinene terpinene terpinen-4-ol verbenol 3-carene

OH O OH

geraniol

cumin aldehyde cymene limonene linalool

OH O MeO OH O

anethole borneol camphor 1,8 cineol carvacrol myrcene

Hande Karaköse Jacobs University Bremen 147

Phytochemical Characterization of Stevia rebaudiana

OH

nerolidol beta-trans-farnesene selinene

cubebene copaene beta elemene gamma cadinene delta cadinene

bisabolene bergamotene germacrene D humulene (caryophyllene)

HO H

H

calacorene calamenene alpha cadinol bourbonene

Hande Karaköse Jacobs University Bremen 148

Phytochemical Characterization of Stevia rebaudiana

PUBLICATIONS

1. H. Karaköse, R. Jaiswal, N. Kuhnert “Characterisation and quantification of hydroxycinnamate derivatives in Stevia rebaudiana leaves by LC-MSn” J. Agric. Food Chem., 2011, 59 (18), 10143–10150.

2. N. Kuhnert, F. Dairpoosh, R. Jaiswal, M. Matei, S.Deshpande, A.Golon, H. Nour, H. Karaköse, N. Hourani “Hill coefficients of dietary polyphenolic enzyme inhibitiors: can beneficial health effects of dietary polyphenols be explained by allosteric enzyme denaturing?” J Chem Biol. 2011 July; 4(3): 109–116.

3. H. Karaköse, N. Kuhnert “Profiling the chlorogenic acids of Stevia rebaudiana by tandem LC- MS” Polyphenol Commun. 2010, Vol 1, 544-546. 71.

4. N. Kuhnert, H. Karaköse, R. Jaiswal, “Analysis, characterization and pharmacokinetics of dietary hydroxycinnamates” Invited review chapter in CRC Handbook of Food Analysis, manuscript in press

5. G. Mikutis, H. Karaköse, R. Jaiswal, A. Le Gresley, T. Islam, M. Fernandez-Lahore, N. Kuhnert “Phenolic promiscuity in the cell nucleus” Food and function, in Press, 2012

6. H. Karaköse, R. Jaiswal, S. Deshpande, N. Kuhnert “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” (Manuscript)

7. H. Karaköse, N. Kuhnert “Development of a LC-ESI MS method for the identification and quantification of steviol glycosides” (Manuscript)

8. H. Karaköse, A. Golon, N. Kuhnert “Gas chromatographic analysis of lipids and volatile terpenes in Stevia rebaudiana” (In preparation)

9. H. Karaköse, R. Shah, N. Kuhnert “Identification of proteins by MALDI-TOF MS in Stevia rebaudiana” (In preparation)

Hande Karaköse Jacobs University Bremen 149

HANDE KARAKÖSE

Address: Clamersdorfer str. 21

28757 - Bremen/Germany

Personal Details

Date of Birth 01.11.1985

Place of Birth Ankara, Turkey

Email [email protected]

Mobile +4915208494817

Education

2009 – 2012 PhD in Chemistry Jacobs University Bremen, Germany/Bremen Title: Chemical Profiling of Stevia Rebaudiana Bertoni • Projects: Identification and profiling of all primary and secondary metabolites of stevia using HPLC-MS, MALDI-TOF and GC-MS. Method development for analysis and quantification of selected compounds and the affect of the growth conditions to the metabolite profile. Extraction of secondary metabolites. Protein extraction and isolation of stevia and identification by MALDI-TOF Lipid and volatile terpene profiling by GC-MS. Statistical analysis (e.g. PCA, ANOVA) of the dataset obtained by LC-MS. Solid phase extraction.

2007 - 2009 Master in Nanomolecular Science Jacobs University Bremen, Germany/Bremen Title: Profiling and Characterisation of Chlorogenic Acids by LC-MSn • Projects: Optimization of the extraction technique and determination of the main chlorogenic acids in various coffee, plum and potato samples by using an HPLC-MSn method. Isolation of selected chlorogenic acids by preparative LC and identification of novel chlorogenic acids Modelling of chlorogenic acids by computational chemistry methods Comparison of experimental spectroscopic data (NMR chemical shifts, Raman spectra, IR) with the calculated spectrical data obtained by quantum mechanical calculations.

2003 - 2007 Bachelor in Chemistry University of Dokuz Eylül Faculty of Science & Arts, İzmir Title: Precontration and Solid Extraction of Uranium (VI) from various water samples using N,N-Dibutyl-N`-Benzoylthiourea

Achievements & Awards

2007 Graduation with distinction; DEÜ, Faculty of Science & Arts, Chemistry Department

2007 Full Scholarship for Master Education in Jacobs University Bremen

2009 Fellowship for PhD in Jacobs University Bremen

Articles H. Karaköse, R. Jaiswal, N. Kuhnert “Characterisation and quantification of hydroxycinnamate derivatives in Stevia Rebaudiana leaves by LC-MSn” J. Agric. Food Chem., 2011, 59 (18), 10143–10150.

N. Kuhnert, F. Dairpoosh, R. Jaiswal, M. Matei, S.Deshpande, A.Golon, H. Nour, H. Karaköse, N. Hourani “Hill coefficients of dietary polyphenolic enzyme inhibitiors: can beneficial health effects of dietary polyphenols be explained by allosteric enzyme denaturing?” J Chem Biol. 2011 July; 4(3): 109–116.

H. Karaköse, N. Kuhnert “Profiling the chlorogenic acids of Stevia Rebaudiana by tandem LC-MS” Polyphenol Commun. 2010, 1, 544-546. 71

G. Mikutis, H. Karaköse, R. Jaiswal, A. Le Gresley, T. Islam, M. Fernandez-Lahore, N. Kuhnert “Phenolic promiscuity in the cell nucleus” Food and function, in Press, 2012.

H. Karaköse, N. Kuhnert “Development of a LC-ESI MS method for the identification and quantification of steviol glycosides” (Manuscript)

Books N. Kuhnert, H. Karaköse, R. Jaiswal, “Analysis, characterization and pharmacokinetics of dietary hydroxycinnamates” Invited review chapter in CRC Handbook of Food Analysis.

Conferences & Seminars

Münster/Germany ISC 2008 – 27th International Symposium on Chromatography 21 - 25 September 2008

Münster/Germany HPLC Masterclass Advanced Method Development (LC certification) 6 - 7 Mai 2010

Montpellier/France 25. International Conference on Polyphenols 23 - 27 August 2010

Poster Presentation: Profiling of Isomers of Chlorogenic Acids by LC-MSn

H. Karaköse, N. Kuhnert

Istanbul/Turkey Terpenist 2010 26 - 29 September 2010 Poster Presentation: Sweet Terpenes in Stevia Rebaudiana; H. Karaköse, N. Kuhnert

Bremen/Germany GDCh – Wissenschaftsforum Chemie 4 - 7 September 2011 Poster Presentation: Profiling, PCA Analysis and Quantification of Chlorogenic acids in Stevia Rebaudiana; H. Karaköse, N. Kuhnert.

Sitges/Spain 5th International Conference on Polyphenol and Health 17-20 October 2011 Poster Presentation: Characterization of Chlorogenic Acids in Stevia Rebaudiana Leaves by LC-MSn ; H. Karaköse, N. Kuhnert.

Internships

03.07 - 28.07.2006 Petkim Petrokimya Holding A.Ş. (Petroleum chemicals), Izmir/Turkey

07.02 - 04.03.2006 University of Leipzig, Faculty of Chemistry and Mineralogy Institute of Analytical Chemistry. Leipzig/Germany

27.06 - 05.08.2005 Petkim Petrokimya Holding A.Ş. (Petroleum chemicals), Izmir/Turkey

Languages English (very good) German (good)

Skills

Computer Skills Microsoft Windows 7& XP Microsoft Office Programs Linux / Ubuntu Open Office

Teaching Skills During my PhD and Master education, I have led seminars, supervised undergraduates in the laboratory Regularly supervise practicals for undergraduate students and have supervised the undergraduate research projects of two final year students. I gave several seminars for undergraduates in School of Engineering and Science

Interests Swimming, travelling, reading, photography.

Reference Prof. Nikolai Kuhnert Email: [email protected] Telephone: +49 421 200-3120 Fax: +49 421 200-3229 ARTICLE

pubs.acs.org/JAFC

Characterization and Quantification of Hydroxycinnamate Derivatives in Stevia rebaudiana Leaves by LC-MSn† Hande Karak€ose, Rakesh Jaiswal, and Nikolai Kuhnert*

School of Engineering and Science, Chemistry, Jacobs University Bremen, 28759 Bremen, Germany bS Supporting Information

ABSTRACT: Stevia rebaudiana leaves are used as a zero-calorie natural sweetener in a variety of food products in Asian countries, especially in Japan. In this study, the hydroxycinnamate derivatives of S. rebaudiana have been investigated qualitatively and quantitatively by LC-MSn. Twenty-four hydroxycinnamic acid derivatives of quinic and shikimic acid were detected, and 19 of them were successfully characterized to regioisomeric levels; 23 are reported for the first time from this source. These comprise three ff ff monoca eoylquinic acids (Mr 354), seven dica eoylquinic acids (Mr 516), one p-coumaroylquinic acid (Mr 338), one feruloylquinic ff ff ff acid (Mr 368), two ca eoyl-feruloylquinic acids (Mr 530), three ca eoylshikimic acids (Mr 336), and two trica eoylquinic acids ff ff fi (Mr 678). Cis isomers of di- and trica eoylquinic acids were observed as well. Three trica eoylquinic acids identi ed in stevia leaves are reported for the first time in nature. These phenolic compounds identified in stevia might affect the organoleptic properties and add additional beneficial health effects to stevia-based products. KEYWORDS: Stevia rebaudiana, chlorogenic acids, hydroxycinnamic acids, caffeoylquinic acids, caffeoylshikimic acids, tandem mass spectrometry

’ INTRODUCTION has been shown to be highly adaptable to cultivation in many other parts of the world. S. rebaudiana occurs naturally on acid Stevia rebaudiana is a plant belonging to the Asteraceae family of plants, which is native to Brazil and Paraguay. Due to the soils of pH 4 5 but will also grow on soils with pH levels of 6.57.5, making it an interesting alternative to plants cultivated natural sweetness of its leaves, S. rebaudiana has caught attention 7 in scientific and industrial fields to act as a natural zero-calorie on poor soils such as tobacco. In addition to diterpene glycosides, a number of secondary sweetener in many applications in the food industry. The leaves plant metabolites have been identified from S. rebaudina including contain ent-kaurene glycosides, comprising stevioside, rebaudio- labdane-type diterpenes, triterpenoids and steroids, flavonoids, sides A, B, C, D, E, and F, and dulcoside A. All of these diterpene ff and oil components. From S. rebaudiana, 10 labdane-type diter- glycosides comprise a steviol backbone structure; they di er only penoids were identified, including austroinulin, isoaustroinulin,6 in the glucose moiety at positions C13 and C19 (Figure 1). and sterebins (AH).8,9 A triterpenoid, lupeol 3-palmitate, was Stevioside is the main sweet-tasting glycoside in stevia and was also separated from stevia.10 As plant sterols, β-sitosterol, stigmas- 1 reported to be 250 300 times sweeter than sucrose. Rebaudio- terol, and campesterol were identified from S. rebaudiana.11 side A is the second most abundant ent-kaurene and sweetest Plant phenols are a large and diverse group of compounds compound in stevia; its sweetness is 400 times greater than that including hydroxycinnamates, tannins, flavonoids, stilbenes, cou- of sucrose, and it has more pleasant taste and is more water- marins, lignans, and lignins.12 Chlorogenic acids (CGAs) are the 2 soluble than stevioside. The amounts of diterpene glycosides most common hydroxycinnamate derivatives observed in the plant may vary depending on the growth conditions of stevia; however, kingdom. By definition, they are a large family of esters formed stevioside accounts for 413% (w/w) and rebaudioside A between quinic acid and one to four residues of certain trans- accounts for 24% (w/w),3 the other glycosides being present hydroxycinnamic acids, most commonly caffeic, p-coumaric, and in lower concentrations. ferulic; sinapic and dimethoxycinnamic acids also occur, and in The principal advantage of stevia metabolites is that they are some plant species various aliphatic acids may replace one or more natural, nonsynthetic products. Stevia leaves can be used in their of the trans-cinnamic acid residues.13 CGAs are involved in natural state (fresh or dried form), due to their high sweetening biological functions in plants such as defense against pathogens intensity. Only small quantities are needed in comparison to white and resistance to diseases. CGAs also participate in enzyme- sugar to achieve comparable sweetness. The primary use of stevia catalyzed browning reactions that may adversely affect the color, is as a commercial sweetener; it is used in a wide range of products flavor, and nutritional quality of dietary sources.14 such as soft drinks, ice cream, chocolate, yogurt, and baked and Several pharmacological activities of CGAs including antiox- cooked foods. Stevia products also have beneficial uses in various idant activity, the ability to increase hepatic glucose utilization,15,16 consumer care products such as toothpaste or mouthwashes.4,5 Stevia may also be used for obesity, diabetics, dental caries, and Received: June 1, 2011 therapeutic effects such as hypoglycemic activity.6 Revised: July 15, 2011 The majority of the annual stevia production of an estimated Accepted: August 2, 2011 4000 t is produced in China and South America. The stevia crop Published: August 02, 2011

r 2011 American Chemical Society 10143 dx.doi.org/10.1021/jf202185m | J. Agric. Food Chem. 2011, 59, 10143–10150 Journal of Agricultural and Food Chemistry ARTICLE

Figure 1. Structures and numberings of caffeoylquinic acids.

Figure 2. Base peak chromatogram of Stevia rebaudiana extract using ion trap MS in negative ion mode. For numbering, see Table 1. inhibition of the HIV-1 integrase,17,18 antispasmodic activity,19 and studies so far. CGAs and their metabolites display additionally inhibition of the mutagenicity of carcinogenic compounds20 highly favorable pharmacokinetic properties.21 23 Because the have been revealed by in vitro, in vivo, and human intervention polyphenols in stevia might affect the organoleptic properties of

10144 dx.doi.org/10.1021/jf202185m |J. Agric. Food Chem. 2011, 59, 10143–10150 Journal of Agricultural and Food Chemistry ARTICLE

Figure 3. Extracted ion chromatograms (EIC) of m/z 515 in negative ion mode (A) before and (B) after UV irradiation.

Figure 4. Structures and numbering of tricaffeoylquinic acids. stevia-based product and could add additional health benefits to UV Irradiation. The prepared sample of stevia leaf extract (1 mL) the product, the objective of the present study was to profile the was placed in a photoreactor (LuzchemLZC -4 V, Ottawa, Canada) phenolic content of S. rebaudina leaves with a particular emphasis under a shortwave UV lamp and irradiated at 245 nm for 40 min. n on hydroxycinnamate derivatives. LC-MS . The LC equipment (Agilent 1100 series, Bremen, Germany) comprised a binary pump, an autosampler with a 100 μL ’ MATERIALS AND METHODS loop, and a diode array detector with a light-pipe flow cell (recording at 320 and 254 nm and scanning from 200 to 600 nm). This was interfaced The chlorogenic acids, 3-caffeoylquinic acid, 4-caffeoylquinic acid, ff ff with an ion-trap mass spectrometer fitted with an ESI source (Bruker 5-ca eoylquinic acid (chlorogenic acid), 3,4-dica eoylquinic acid, 3,5- n ff ff Daltonics HCT Ultra, Bremen, Germany) operating in Auto-MS mode dica eoylquinic acid, and 4,5-dica eoylquinic acid, were purchased from 2 3 4 PhytoLab (Vestenbergsgreuth, Germany). All other chemicals were to obtain fragment ions m/z. As necessary, MS ,MS, and MS purchased from Sigma-Aldrich (Bremen, Germany). Stevia leaves were fragment-targeted experiments were performed to focus only on com- purchased from a market in Bremen, Germany. pounds producing a parent ion at m/z 335.1, 337.1, 367.1, 529.2, or n Sample Preparation. Two grams of S. rebaudiana leaves was 677.3. Tandem mass spectra were acquired in Auto-MS mode (smart immersed in liquid nitrogen, ground in a hammer mill, and extracted first fragmentation) using a ramping of the collision energy. Maximum with 150 mL of chloroform in a Soxhlet apparatus (Buchi B-811 fragmentation amplitude was set to 1 V, starting at 30% and ending at extraction system) for 2 h and then with 150 mL of methanol for 200%. MS operating conditions (negative mode) had been optimized another 2 h. Solvents were removed from the methanolic extract in using 5-caffeoylquinic acid28 with a capillary temperature of 365 °C, a vacuo, and extracts were stored at 20 °C until required. dry gas flow rate of 10 L/min, and a nebulizer pressure of 50 psi.

10145 dx.doi.org/10.1021/jf202185m |J. Agric. Food Chem. 2011, 59, 10143–10150 Journal of Agricultural and Food Chemistry ARTICLE

Figure 5. Tandem mass spectra of 1,3,5-triCQA in negative ion mode.

Figure 6. Tandem mass spectra of 3,4,5-triCQA in negative ion mode.

High-resolution LC-MS was carried out using the same HPLC at room temperature. The reaction mixture was stirred for 6 h and equipped with a MicroTOF Focus mass spectrometer (Bruker acidified with 2 mol/L HCl (pH ≈1) and then stirred for an additional Daltonics) fitted with an ESI source, and internal calibration was 3 h to remove the acetyl protecting groups. The layers were separated, achieved with 10 mL of 0.1 mol/L sodium formate solution injected and the aqueous phase was re-extracted with CH2Cl2 (1 20 mL) and through a six-port valve prior to each chromatographic run. Calibration EtOAc (2 20 mL). The combined organic layers were dried over was carried out using the enhanced quadratic calibration mode. Na2SO4 and filtered, and the solvents were removed in vacuo. The HPLC. Separation was achieved on a 150 3 mm i.d. column resulting esters were analyzed by HPLC-MS. containing diphenyl 5 μm with a 4 3 mm i.d. guard column of the same material (Varian, Darmstadt, Germany). Solvent A was water/formic ’ RESULTS AND DISCUSSION acid (1000 + 0.05 v/v), and solvent B was methanol. Solvents were Methanol extracts of stevia dry leaves were directly used for delivered at a total flow rate of 0.5 mL/min. The gradient profile was ffi from 10 to 70% B linearly in 60 min followed by 10 min isocratic and a LC-MS analysis. E cient separation and resolution were return to 10% B at 80 and 10 min isocratic to re-equilibrate. achieved with diphenyl packing and acetonitrile/water as solvent Calibration Curve of Standard Compounds. Stock solutions in the HPLC method. Negative ion mode was used for all MS of the standard compounds were prepared in methanol. A series of measurements. The HPLC method used here constitutes a 24 standard solutions was injected (5 μL) into the LC-MS system. The variation of methods employed previously, with variations areas of the peaks of each standard from UV chromatograms were used required to achieve sufficient separation of triacyl chlorogenic to make the respective standard curves. acids and ent-kaurene glycosides. In comparison to isolation of Synthesis of the Mixture of Regioisomers of Tricaffeoyl- CGAs from green coffee beans, no removal of proteins/peptides 13,24 quinic Acids. To a solution of quinic acid (96 mg, 0.5 mmol) and by Carrez reagent was necessary. DMAP (16 mg, 0.12 mmol) in CH2Cl2 (10 mL) were added triethy- All data for chlorogenic acids and diterpene glycosides pre- lamine (4 mL) and 3,4-diacetylcaffeic acid chloride (423 mg, 1.5 mmol) sented in this paper use the IUPAC numbering system,32 and

10146 dx.doi.org/10.1021/jf202185m |J. Agric. Food Chem. 2011, 59, 10143–10150 Journal of Agricultural and Food Chemistry ARTICLE

Table 1. Tandem Mass Spectral Data of Hydroxycinnamates in Stevia rebaudiana Leaf Extract

MS2 MS3 MS4

secondary peaks secondary peaks secondary peaks

m/z base base base no. compd (neg) peak m/z int m/z int m/z int peak m/z int m/z int m/z int peak m/z int m/z int m/z int

1 3-CQA 353.0 190.7 178.8 49 134.9 8 126.8 172.8 37 85.2 55 110.8 65 188.8 174.6 64 134.4 84 2 5-CQA 353.0 190.7 126.8 172.7 49 85.1 61 110.8 21 108.8 3 4-CQA 353.0 172.7 178.8 60 190.6 14 134.8 8 93.0 110.8 62 154.7 24 4 3,5-diCQA 515.1 353.0 190.8 8 190.7 178.8 49 134.9 7 126.8 93.0 98 85.2 62 172.6 48 5 3,4-diCQA 515.1 353.0 335.0 12 172.8 17 172.8 178.8 67 190.8 57 134.8 10 93.0 110.8 41 83.2 6 6 4,5-diCQA 515.1 353.0 299.0 3 254.9 7 172.8 17 172.8 178.6 52 190.8 28 135.0 7 93.0 110.8 89 83.0 18 7 a cis-3,5-diCQA 515.1 353.0 190.8 10 190.7 178.8 50 172.8 11 134.9 10 85.0 126.8 85 93.0 53 172.7 36 8 a cis-4,5-diCQA 515.1 353.0 172.8 13 172.8 178.6 66 190.6 35 134.9 11 93.0 110.9 39 83.0 10 9 cis-4,5-diCQA 515.1 353.0 172.7 7 172.7 178.8 66 190.8 59 134.9 12 93.0 110.9 23 83.0 19 10 a cis-4,5-diCQA 515.1 353.0 172.8 12 172.7 178.6 76 190.6 70 134.8 18 93.0 110.8 39 83.0 6 11 5-p-CoQA 337.1 190.7 162.8 6 126.8 172.7 42 108.8 44 92.8 32 12 3F,5CQA 529.1 367.0 353.0 12 192.7 7 178.6 2 192.7 172.6 13 133.7 14 133.7 126.6 16 13 C,FQA 529.1 367.1 349.0 7 178.7 10 178.7 160.8 73 134.8 85 134.7 14 4C,5FQA 529.1 353.0 254.8 5 172.7 17 172.7 178.6 65 190.6 24 134.7 11 93.0 110.8 32 59.4 52 15 5-CSA 335.1 178.7 172.8 11 134.8 20 134.7 16 4-CSA 335.1 178.7 160.6 82 134.8 51 134.7 17 3-CSA 335.1 178.7 160.8 4 134.8 42 134.8 18 5-FQA 367.1 190.7 172.8 3 85.0 126.8 85 172.6 28 19 FQA 367.1 178.7 190.8 33 160.8 12 134.8 72 134.7 106.8 5 20 FQA 367.2 176.8 161.6 130.8 57 21 3,4,5-triCQA 677.1 515.1 353.0 20 353.0 335.0 16 299.0 4 172.8 30 172.7 178.6 58 190.6 42 134.8 16 22 1,3,5-triCQA 677.1 515.1 353.0 16 353.0 335.0 15 254.9 5 172.7 28 190.8 178.6 72 172.6 90 136.7 10 23 triCQA 677.1 515.0 353.0 15 353.0 172.7 38 254.8 4 172.7 178.8 60 190.6 33 134.7 14 24 triCQA 677.1 515.1 353.0 20 353.0 172.7 22 254.8 4 172.8 178.6 87 190.8 90 134.7 15 structures are presented in Figure 1. Peak assignments of CGAs have been made on the basis of structure diagnostic hierarchical keys previously developed,24 26 supported by means of their parent ion, UV spectra, and retention times relative to 5-CQA using validated methods in our laboratory.24,27 More sensitive and more selective fragment-targeted MSn experiments were used for quantitatively minor components. The base peak chromatogram of stevia extract is shown in Figure 2. Abbrevia- tions and numbering are given in Figure 1. Stevia extract was analyzed by LC-MSn in the negative ion mode using an ESI ion- trap mass spectrometer, allowing assignments of compounds to regioisomeric level, and also by high-resolution mass spectro- metry using ESI-TOF in negative ion mode connected to LC. With the guidance of previous studies from Clifford,24 29 three CQAs (13), seven di-CQAs (410), three FQAs (1820), one p-CoQA (11), three CFQAs (1214), three CSAs (1517), and four tri-CQAs (2124) were located in the chromatogram. For all compounds the high-resolution mass data were in good agreement with the theoretical molecular formulas, with a mass error of below 5 ppm, confirming the elemental Figure 7. Extracted ion chromatograms (EIC) of m/z 677 in negative compositions of all compounds investigated. ion mode (A) before and (B) after UV irradiation. Reliable characterization of diterpene glycosides content in stevia is crucial. Because the structures of single glycosides are S. rebaudiana is given. Characterization of the compounds was very similar, they have very similar retention times in LC and achieved by ion-trap mass spectrometry with SIM, and confirmation therefore result in overlapping of peaks in the chromatogram. of elemental composition was provided by ESI-TOF measure- In this paper, the general profile of diterpene glycosides in ments (see the Supporting Information).

10147 dx.doi.org/10.1021/jf202185m |J. Agric. Food Chem. 2011, 59, 10143–10150 Journal of Agricultural and Food Chemistry ARTICLE

Table 2. Quantities of Mono- and Di-CQAs in S. rebaudiana Leaves

compd concn range calibration curve correl coeff calcd amount (μg/g)

3-CQA 1 μg/mL1 mg/mL Y = 4.457x 460.04 0.99 35.5 5-CQA 1 μg/mL3 mg/mL Y = 17.719x 2361.90 0.99 44.3 4-CQA 0.07 μg/mL1 mg/mL Y = 13.288x 1223.00 0.99 70.3 3,5-diCQA 0.09 μg/mL1 mg/mL Y = 5.3176x 529.76 0.99 145.6 3,4-diCQA 0.07 μg/mL1 mg/mL Y = 14.789x 1401.40 0.99 28.6 4,5-diCQA 0.03 μg/mL04 mg/mL Y = 16.251x 697.77 0.99 37.2

2 Characterization of Caffeoylquinic Acids (Mr 354) and regioisomers show m/z 178 (caffeic acid fragment) in their MS Dicaffeoylquinic acids (Mr 516). Three peaks were detected spectra. 4-CQA shows an intense characteristic fragment ion at at m/z 353.1 and assigned using the hierarchial keys previously m/z 160, which is absent in the MS2 spectra of 3-CSA and 5-CSA. 24 developed as well-known 3-CQA, 5-CQA, and 4-CQA. Three Characterization of Tricaffeoylquinic Acid (Mr 678). Four dicaffeoylquinic acid isomers were identified by their parent ion triacyl CQA isomers (Figure 4) were detected in the stevia m/z 515.2 and were assigned as 3,5-diCQA, 3,4-diCQA, and 4,5- extract at 677 for tricaffeoyls in neg. mode and confirmed as diCQA using the hierarchial keys.24,26 Three further peaks tricaffeoyl derivatives by targeted MS4 experiments. Assignment present as minor components showed fragmentation patterns of regiochemistry was assisted by an independent synthesis of a similar to that of 4,5-diCQA. We have recently reported on cis mixture of all four possible regioisomers of tricaffeoylquinic isomers of chlorogenic acids present in plant tissue exposed to acids. The chromatogram of the mixture of all theoretically pos- UV light, which have formed in a photochemical transcis sible four regioisomers of tricaffeoylquinic acid obtained through isomerization reaction.30 To confirm if the remaining three peaks synthesis showed two well-resolved peaks with retention times correspond to cis isomers, the extract was irradiated with UV and MS data identical to those present in the stevia extract along light at 245 nm for 40 min. After irradiation, a significant increase with an intense broad peak in a retention time range where the in the intensities of two peaks (9 and 10 in Figure 3) was two remaining isomers in the stevia extract were observed (see observed, if compared to their corresponding trans isomers from the Supporting Information). Detailed studies of the tandem the original plant extract. In addition, a significant increase was mass spectra at various retention times within the broad peak observed in the intensity of cis-3,5-diCQA (7 in Figure 3) peak suggest that this broad peak must correspond to two distinct accompanied by a decrease of the 3,4-diCQA (5 in Figure 3) unresolved regioisomers of tricaffeoylquinic acid. Comparison peak. This finding suggests that under the chromatographic of the chromatogram of the synthetic mixture with the extract conditions employed the cis isomer is coeluting with 3,4-diCQA allowed unambiguous assignment of the two regioisomers in the (Figure 3). extract by identity of the fragmentation pattern compared to the On the basis of increased intensity after UV irradiation and synthetic mixture. Identification of 1,3,5-triCQA in the extract fragmentation pattern, three additional cis isomers were ob- was followed automatically due to the absence of an MS4 base served for 4,5-diCQA. One of these isomers was assigned as peak at m/z 173 corresponding to a dehydrated MS2 base peak of cis-4,5-diCQA (9), and two of them were assigned as cistrans the quinic moiety characteristic of 4-acylated isomers. The MS4 (a cis) isomer, but the distinction between 4-cis,5-trans-diCQA base peak at m/z ∼191 and a secondary peak at m/z 178 (72% of and 4-trans-5-cis-diCQA was not possible (8 and 10). base peak) suggest the 3,5-disubstitution pattern (Figure 5). Characterization of Feruloylquinic Acid (Mr 368), p-Cou- 3,4,5-triCQA was identified by comparison to material described 36,37 maroylquinic Acid (Mr 338), and Caffeoylferuloylquinic Acid previously. (Figure 6) (Mr 530). Only one peak was detected at m/z 337.1, which was The two remaining peaks might be cis isomers of 3,4,5-triCQA identified as 5-p-CoQA according to its fragmentation pattern. and 1,3,5-triCQA, or they can correspond to either 1,4,5-tri- Three peaks were detected at m/z 367, and one of them was CQA and 1,3,4-tri-CQA or any of their cis isomers (see Table 1). identified as 5-FQA; the other two peaks could not be assigned However, current information does not allow us to discriminate due to their uncommon fragmentation pattern. unambiguously between these regioisomers at the moment. To A targeted MS3 experiment at m/z 529.2 ([M H+] ) probe whether cis isomers were present, the extract was again applied to the extract located three peaks, and two of them were irradiated with UV light, and after chromatographic analysis, a identified as 3F,5CQA and 4C,5FQA on the basis of their significant increase in the intensity of the peaks of 4-acylated characteristic fragmentations in MS2 and MS3 spectra. The isomers was observed (Figure 7). Otherwise, the experiment was assigments are achieved using the hierarchial keys previously inconclusive. It is worth noting that in theory for each tricaffeoyl developed, and mass spectra published previously are not pre- derivative eight stereoisomers with various transcis stereoche- sented here.24,31 mistries are possible, thus increasing the total number of isomeric Characterization of Caffeoylshikimic Acids (Mr 336). Caf- tricaffeoylquinic acids to 32. Given the identity of MS data and feoylshikimic acids (CSA) have been reported in date palms, the absence of characteristic shoulders in the UV spectra sweet basil, and carrot,32 35 and they have been characterized to characteristic for cis-caffeoyl derivatives, we tentatively assign regioisomeric level in yerba mate leaves by tandem mass spectra the two remaining isomers as 1,4,5-tri-CQA and 1,3,4-tri- previously.36 This class of compounds is reported here for the CQA. Only 3,4,5-tri-CQA has been previously reported in first time from the Asteraceae family of plants. A targeted MS3 nature, whereas the remaining isomers are reported here for experiment at m/z 335.1 ([M H+] ) applied to the extract the first time. located three peaks, and they were identified by their fragmenta- Quantification of Caffeoylquinic Acids. Following the qua- tion patterns as 5-CSA, 4-CSA, and 3-CSA (1517).36 All three litative profiling of chlorogenic acids in S. rebaudiana, we decided

10148 dx.doi.org/10.1021/jf202185m |J. Agric. Food Chem. 2011, 59, 10143–10150 Journal of Agricultural and Food Chemistry ARTICLE to quantify the levels of selected compounds. Chlorogenic acid (4) Chatsudthipong, V.; Muanprasat, C. Stevioside and related standard solutions were analyzed by LC-MS using the same compounds: therapeutic benefits beyond sweetness. Pharmacol. Ther. chromatographic method as used for stevia leaf extracts. For six 2009, 121,41–54. selected monoacyl- and diacylquinic acids, calibration curves (5) Brandle, J. E.; Rosa, N. Heritability for yield, leaf-stem ratio and stevioside content estimated from a landrace cultivar of Stevia-rebaudi- were obtained using six-point calibration from the UV chroma- – togram recorded at 320 nm. The individual amounts calculated ana. Can. J. Plant Sci. 1992, 72, 1263 1266. (6) Ibrahim, N. A.; El-Gengaihi, S.; Motawe, H.; Riad, S. A. Phyto- for mono- and dicaffeoylquinic acids are listed in Table 2, which chemical and biological investigation of Stevia rebaudiana Bertoni; also lists the correlation coefficient of linear regression for each 1-labdane-type diterpene. Eur. Food Res. Technol. 2007, 224, 483–488. standard sample and the concentration range. (7) Shock, C. C. Rebaudi’s stevia: natural noncaloric sweeteners. Among the monocaffeoylquinic acids, 4-CQA was found to be Calif. Agric. 1982,4–5. the most abundant compound, and among all CQAs 3,5-diCQA (8) Oshima, Y.; Saito, J.; Hikino, H. Sterebins A, B, C and D, was found to be the most abundant compound. The total bisnorditerpenoids of Stevia-rebaudiana leaves. Tetrahedron 1986, 42, chlorogenic acid amount determined here is around 370 μg/g 6443–6446. of dry leaf. (9) Oshima, Y.; Saito, J.-I.; Hikino, H. Sterebins E, F, G and H, fi diterpenoids of Stevia rebaudiana leaves. Phytochemistry 1988, 27, In this study we pro led the chlorogenic acids in S. rebaudiana – employing LC-MSn and LC-TOF techniques. A total of 24 624 626. (10) Yasukawa, K.; Yamaguchi, A.; Arita, J.; Sakurai, S.; Ikeda, A.; chlorogenic acids were detected in S. rebaudiana leaves, with ff fi Takido, M. Inhibitory e ect of edible plant extracts on 12-O-tetradeca- 23 compounds described for the rst time from this source. Tri- noylphorbol-13-acetate-induced ear edema in mice. Phytother. Res. 1993, CQAs were reported for the first time from S. rebaudiana with 7, 185–189. three regioisomers found for the first time in nature. CSAs were (11) Kinghorn, A. D. Stevia The Genus Stevia; CRC Press: Boca characterized for the first time from a plant belonging to the Raton, FL, 2001. Astareceae family by using tandem mass spectrometry. Quanti- (12) Rice Evans, C. A.; Miller, N. J.; Paganga, G. Structure fication of selected mono- and di-CQAs was achieved by using antioxidant activity relationships of flavonoids and phenolic acids. Free Radical Biol. Med. 1996, 20, 933–956. the UV chromatogram with total chlorogenic acid levels found to ff be 370 μg/g of dry leaf. (13) Cli ord, M. N.; Marks, S.; Knight, S.; Kuhnert, N. Character- ization by LC-MSn of four new classes of p-coumaric acid-containing diacyl chlorogenic acids in green coffee beans. J. Agric. Food Chem. 2006, ’ ASSOCIATED CONTENT 54, 4095–4101. (14) Im, H. W.; Suh, B. S.; Lee, S. U.; Kozukue, N.; Ohnisi- bS Supporting Information. Additional EIC of triacyl Kameyama, M.; Levin, C. E.; Friedman, M. Analysis of phenolic CGAs, MS2 +MS3 data of all compounds mentioned in the text, compounds by high-performance liquid chromatography and liquid fl table of high-resolution MS-TOF data for compounds identified, chromatography/mass spectrometry in potato plant owers, leaves, and structures of ent-kaurene terpenes. This material is free of stems, and tubers and in home-processed potatoes. J. Agric. Food Chem. 2008, 56, 3341–3349. charge via the Internet at http://pubs.acs.org (15) Kono, Y.; Kobayashi, K.; Tagawa, S.; Adachi, K.; Ueda, A.; ’ AUTHOR INFORMATION Sawa, Y.; Shibata, H. Antioxidant activity of polyphenolics in diets rate constants of reactions of chlorogenic acid and caffeic acid with reactive Corresponding Author species of oxygen and nitrogen. Biochim. Biophys. Acta: Gen. Subj. 1997, 1335, 335–342. *Phone: 49 421 200 3120. Fax: 49 421 200 3229. E-mail: (16) Hemmerle, H.; Burger, H. J.; Below, P.; Schubert, G.; [email protected]. Rippel, R.; Schindler, P. W.; Paulus, E.; Herling, A. W. Chlorogenic Funding Sources acid and synthetic chlorogenic acid derivatives: novel inhibitors of hepatic glucose-6-phosphate translocase. J. Med. Chem. 1997, 40, Financial support from the European Union (project DIVAS) is 137–145. gratefully acknowledged. (17) Robinson, W. E.; Cordeiro, M.; AbdelMalek, S.; Jia, Q.; Chow, S. A.; Reinecke, M. G.; Mitchell, W. M. Dicaffeoylquinic acid inhibitors of human immunodeficiency virus integrase: Inhibition of the core ’ ACKNOWLEDGMENT catalytic domain of human immunodeficiency virus integrase. Mol. – € Pharmacol. 1996, 50, 846 855. We acknowledge the technical assistance of Anja Muller. (18) Robinson, W. E.; Reinecke, M. G.; AbdelMalek, S.; Jia, Q.; Chow, S. A. Inhibitors of HIV-1 replication that inhibit HPV integrase. ’ DEDICATION Proc. Natl. Acad. Sci. U.S.A. 1996, 93, 6326–6331. † (19) Trute, A.; Gross, J.; Mutschler, E.; Nahrstedt, A. In vitro This paper is dedicated to Prof. M. N. Clifford on the occasion of antispasmodic compounds of the dry extract obtained from Hedera his 65th birthday. helix. Planta Med. 1997, 63, 125–129. (20) Stich, H. F.; Rosin, M. P.; Bryson, L. Inhibition of mutagenicity ff ’ REFERENCES of a model nitrosation reaction by naturally-occuring phenolics, co ee and tea. Mutat. Res. 1982, 95, 119–128. (1) Kolb, N.; Herrera, J. L.; Ferreyra, D. J.; Uliana, R. F. Analysis of (21) Hanlon, N.; Coldham, N.; Gielbert, A.; Kuhnert, N.; Sauer, sweet diterpene glycosides from Stevia rebaudiana: improved HPLC M. J.; Kingi, L. J.; Ioannides, C. Absolute bioavailability and dose- method. J. Agric. Food Chem. 2001, 49, 4538–4541. dependent pharmacokinetic behaviour of dietary doses of the chemo- (2) Crammer, B.; Ikan, R. Sweet glycosides from the stevia plant. preventive isothiocyanate sulforaphane in rat. Br. J. Nutr. 2008, 99, Chem. Br. 1986, 22, 915. 559–564. (3) Pol, J.; Hohnova, B.; Hyotylainen, T. Characterisation of Stevia (22) Stalmach, A.; Mullen, W.; Barron, D.; Uchida, K.; Yokota, T.; rebaudiana by comprehensive two-dimensional liquid chromatography Cavin, C.; Steiling, H.; Williamson, G.; Crozier, A. Metabolite profiling time-of-flight mass spectrometry. J. Chromatogr., A 2007, 1150,85–92. of hydroxycinnamate derivatives in plasma and urine after the ingestion

10149 dx.doi.org/10.1021/jf202185m |J. Agric. Food Chem. 2011, 59, 10143–10150 Journal of Agricultural and Food Chemistry ARTICLE of coffee by humans: identification of biomarkers of coffee consumption. Drug Metab. Dispos. 2009, 37, 1749–1758. (23) Farah, A.; Monteiro, M.; Donangelo, C. M.; Lafay, S. Chloro- genic acids from green coffee extract are highly bioavailable in humans. J. Nutr. 2008, 138, 2309–2315. (24) 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. (25) IUPAC Commission on the Nomenclature of Organic Chem- istry (CNOC) and IUPAC-IUB Commission on Biochemical Nomen- clature (CBN). Nomenclature of cyclitols. Recommendations, 1973. Biochem. J. 1976, 153,2331. (26) 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. (27) Clifford, M. N. Chlorogenic acids and other cinnamates nature, occurrence, dietary burden, absorption and metabolism. J. Sci. Food Agric. 2000, 80, 1033–1043. (28) Clifford, M. N. Chlorogenic acids and other cinnamates nature, occurrence and dietary burden. J. Sci. Food Agric. 1999, 79, 362–372. (29) Clifford, M. N. The Analysis and Characterization of Chlorogenic Acids and Other Cinnamates; Royal Society of Chemistry: Cambridge, U.K., 2003. (30) 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. (31) Clifford, M. N.; Wu, W. G.; Kuhnert, N. The chlorogenic acids of Hemerocallis. Food Chem. 2006, 95, 574–578. (32) Mansouri, A.; Embarek, G.; Kokkalou, E.; Kefalas, P. Phenolic profile and antioxidant activity of the Algerian ripe date palm fruit (Phoenix dactylifera). Food Chem. 2005, 89, 411–420. (33) Ziouti, A.; ElModafar, C.; ElMandili, A.; ElBoustani, E.; Macheix, J. J. Identification of the caffeoylshikimic acids in the roots of the date palm, principle fungitoxic compounds vis-a-vis Fusarium- oxysporum fspalbedinis. J. Phytopathol. 1996, 144, 197–202. (34) Gang, D. R.; Beuerle, T.; Ullmann, P.; Werck-Reichhart, D.; Pichersky, E. Differential production of meta hydroxylated phenylpro- panoids in sweet basil peltate glandular trichomes and leaves is con- trolled by the activities of specific acyltransferases and hydroxylases. Plant Physiol. 2002, 130, 1536–1544. (35) Kuhnl, T.; Koch, U.; Heller, W.; Wellmann, E. Chlorogenic acid biosynthesis characterization of a light-induced microsomal 5-O- 0 (4-coumaroyl)-D-quinate shikimate 3 -hydroxylase from carrot (Daucus- carota L) cell suspension cultures. Arch. Biochem. Biophys. 1987, 258, 226–232. (36) Jaiswal, R.; Sovdat, T.; Vivan, F.; Kuhnert, N. Profiling and characterization by LC-MSn of the chlorogenic acids and hydroxycinna- moylshikimate esters in mate (Ilex paraguariensis). J. Agric. Food Chem. 2010, 58, 5471–5484. (37) Jaiswal, R.; Kuhnert, N. Hierarchical scheme for liquid chro- matography/multi-stage spectrometric identification of 3,4,5-triacyl chlorogenic acids in green Robusta coffee beans. Rapid Commun. Mass Spectrom. 2010, 24, 2283–2294.

10150 dx.doi.org/10.1021/jf202185m |J. Agric. Food Chem. 2011, 59, 10143–10150 Supplemantary Information

TRICQAs

EIC of targeted MS for m/z 677

Intens. STEVIA_CGA_TAR7.D: EIC 677.0 -All MS x107 21 1.25

1.00

0.75

0.50

0.25 glycoside

0.00 44 46 48 50 52 54 Time [min]

3,4,5‐triCQA (21)

Intens. -MS [%] 677.1 100 338.0 515.1 160.7 419.9 0 [%] -MS2(677.1) 515.1 100 353.0 298.9 [%]0 -MS3(677.5->515.1), 353.0 100 172.7 254.8 [%]0 -MS4(677.5->515.3->353.0) 172.7 100

134.8 0 100 200 300 400 500 600 700 m/z

1,3,5‐triCQA (22)

Intens. [%] -MS, 579.1 100

293.0 338.0365.1 677.1 228.6 448.8 515.0 754.9 [%]0 -MS2(677.1), 515.1 100

353.0 629.3 0 [%] -MS3(677.3->515.0) 353.0 100 172.7 254.9 [%]0 -MS4(677.3->515.3->352.8 190.8 100

136.7 0 100 200 300 400 500 600 700 m/z

`

Supplemantary Information

4‐acylated‐triCQA (23/24)

Intens. [%] -MS 561.3 100 677.1 284.9 515.1 658.7 160.7 238.6 390.8 466.1 793.4 0 [%] -MS2(677.1) 515.0 100 353.0 467.2 0 [%] -MS3(677.2->515.1), 353.0 100 172.7 254.8 [%]0 -MS4(677.2->515.2->353.0), 100 172.7 134.7 325.0 0 100 200 300 400 500 600 700 800 m/z

4‐acylated‐triCQA (23/24)

Intens. [%] 677.1 100 593.1 448.9 497.0 390.8 487.1 791.3 112.8 186.6 238.7 0 [%] -MS2(677.1) 515.1 100 353.0 0 [%] -MS3(677.5->515.1), 353.0 100 172.7 254.8 0 [%] -MS4(677.5->515.3->353.1) 172.8 100

134.7 0 100 200 300 400 500 600 700 800 m/z

`

Supplemantary Information

Synthesis:

EIC of 677

Intens . CQAMIXTURE_1112.D: EIC 677.0 -All MS x106

1.25

1.00

0.75

0.50

0.25

0.00 0 10 20 30 40 50 60 Time [min]

1,3,5‐triCQA (22)

Intens. -MS [%] 503.1 100 677.1 186.7 260.7 341.0 402.7 460.7 544.7 0 [%] -MS2(677.1) 515.1 100 353.0 [%]0 -MS3(677.3->514.9), 353.0 100 172.7 298.9 0 [%] -MS4(677.3->515.2->353.0) 190.7 100 134.8 0 100 200 300 400 500 600 m/z

4‐acyl‐triCQA

Intens. -MS [%] 341.1 100 677.1 260.7 402.8 515.1 178.7 460.7 544.7 602.6 642.8 0 [%] -MS2(677.1) 515.1 100 353.1 0 [%] -MS3(677.4->515.1) 353.0 100 172.7 0 [%] -MS4(677.4->515.3->352.9) 172.7 100 134.8 0 100 200 300 400 500 600 m/z

`

Supplemantary Information

4‐acyl‐triCQA

Intens. [%] 677.1 100 338.1 515.1 160.7 260.7 460.7 582.7 [%]0 -MS2(677.1) 515.1 100

353.1 [%]0 -MS3(677.4->515.1) 353.0 100 172.8 254.9 299.0 [%]0 -MS4(677.4->515.2->353.0) 172.7 100

134.8 0 100 200 300 400 500 600 m/z

m/z 353

Intens . STEVIA_CGA_DP01.D: EIC 353.0 -All MS x107

1.5

1.0

0.5

0.0 -5 0 5 10 15 20 25 30 Time [min]

`

Supplemantary Information

3‐CQA (1):

Intens. -MS [%] 353.0 100 375.0 50 529.4 [%]0 -MS2(353.0) 190.7 100 50 134.9 [%]0 -MS3(353.2->190.8), 100 126.8

50 172.7

0 200 400 600 800 1000 m/z

5‐CQA(2):

Intens. -MS [%] 353.0 100 50 190.8 [%]0 -MS2(353.0) 190.7 100 50

[%]0 -MS3(353.1->190.8) 126.8 100 85.1 50 172.7 0 200 400 600 800 1000 m/z

4‐CQA(3):

Intens. -MS [%] 353.0 100 50 375.0 438.0 529.4 [%]0 -MS2(353.0) 172.7 100 50 134.8 [%]0 -MS3(353.2->172.7), 93.0 100 50 154.7 0 200 400 600 800 1000 m/z

`

Supplemantary Information m/z 515

3,5‐diCQA(4)

Intens. -MS [%] 515.1 100 537.1 353.0 635.0 [%]0 -MS2(515.1) 353.0 100

190.8 0 [%] -MS3(515.3->353.0), 190.7 100 134.9 [%]0 -MS4(515.3->353.1->190.8), 93.0 100 172.7 0 100 200 300 400 500 600 700 m/z

3,4‐diCQA (5)

Intens. [%] -MS 515.1 100 593.1 256.9 635.0 [%]0 -MS2(515.1) 353.0 100 172.8 0 [%] -MS3(515.3->353.0 172.8 100

134.9 [%]0 -MS4(515.3->353.1->172.8) 93.0 100 154.7 0 100 200 300 400 500 600 700 m/z

4,5‐diCQA (6)

Intens. [%] -MS 515.1 100 447.0 353.0 [%]0 -MS2(515.1) 353.0 100

172.8 254.8 298.9 [%]0 -MS3(515.3->353.0) 172.7 100

134.8 [%]0 -MS4(515.3->353.2->172.8), 93.0 100 71.3 154.7 0 100 200 300 400 500 600 700 m/z

`

Supplemantary Information

A cis‐3,5‐diCQA (7)

Intens. -MS [%] 515.1 100 537.0 353.0 613.0 0 [%] -MS2(515.1) 353.0 100 190.8 [%]0 -MS3(515.2->353.0) 190.7 100 134.9 [%]0 -MS4(515.2->353.1->190.7) 85.1 100 126.8 172.7 0 100 200 300 400 500 600 700 m/z

cis 4,5‐diCQA (9)

Intens. [%] -MS 515.1 100 537.1 260.7 353.0 431.1 635.1 771.2 [%]0 -MS2(515.1) 353.0 100

172.7 [%]0 -MS3(515.1->353.0) 172.7 100

134.9 [%]0 -MS4(515.1->353.0->172.8) 92.9 100 71.3 154.7 0 100 200 300 400 500 600 700 m/z

A cis‐4,5‐diCQA (8)

Intens. -MS [%] 515.1 100 353.0 0 [%] -MS2(515.1) 353.0 100 172.8 0 [%] -MS3(515.2->353.0) 172.8 100

134.9 0 [%] -MS4(515.2->353.1->172.7),

100 93.0 59.4 154.7 0 100 200 300 400 500 600 700 m/z

`

Supplemantary Information

A cis‐4,5‐diCQA (10)

Intens. [%] -MS 515.1 100 549.2 0 353.0 [%] -MS2(515.1) 353.0 100 172.8 0 [%] -MS3(515.2->353.0) 172.7 100 134.8 0 [%] -MS4(515.2->353.0->172.8)

100 93.0 71.2 154.7 0 100 200 300 400 500 600 700 m/z

5‐p‐CoQA (11)

Intens. -MS [%] 337.1 100 529.5 50 260.7 380.8 725.2 92.9 [%]0 -MS2(337.1) 190.7 100

50

0 [%] -MS3(337.2->190.8)

100 126.8

50 172.7

0 200 400 600 800 1000 m/z

5‐FQA (18)

Intens. -MS [%] 367.1 100

50 260.7 528.6595.1 96.9 186.7 [%]0 -MS2(367.1) 190.7 100

50 296.7 [%]0 -MS3(367.2->190.9)

100 85.1

50 172.7 0 200 400 600 800 1000 m/z

`

Supplemantary Information

5‐CSA (15):

Intens. -MS [%] 399.1 100 335.1 50 529.5 260.7 447.1 96.9 186.7 625.1 771.2 [%]0 -MS2(335.1) 178.7 100 50 134.8 290.9 [%]0 -MS3(335.3->178.8) 134.7 100 50

0 200 400 600 800 1000 m/z

4‐CSA (16):

Intens. -MS [%] 353.1 100 335.1 50 260.7 529.5 96.9 [%]0 -MS2(335.1) 178.7 100

50 290.9 [%]0 -MS3(335.2->178.8), 134.7 100

50

0 200 400 600 800 1000 m/z

3‐CSA (17):

Intens. -MS [%] 335.1 100

50 260.7 380.8 528.7 96.9 733.1 0 [%] -MS2(335.1) 178.7 100

50 134.8 260.6 0 [%] -MS3(335.3->178.7)) 134.8 100

50

0 200 400 600 800 1000 m/z

`

Supplemantary Information m/z 529

3F,5CQA (12)

Intens. [%] -MS 100 771.1 529.1 385.1 260.6 316.8 431.1 577.2 [%]0 367.0 -MS2(529.1) 100 192.7 460.7 [%]0 192.7 -MS3(529.2->367.1) 100 133.7 [%]0 -MS4(529.2->367.1->192.5) 133.7 100 59.4 0 100 200 300 400 500 600 700 m/z

(13)

Intens. [%] -MS 100 755.2 529.1 377.0 445.1 508.7 186.5 254.6 322.8 657.3 0 [%] -MS2(529.1) 367.1 100 178.7 460.7 0 [%] -MS3(529.4->367.7), 178.7 100 134.7

0 [%] -MS4(529.4->367.3->178.5)

100 134.7

0 100 200 300 400 500 600 700 m/z

4C,5FQA (14):

Intens. [%] -MS 529.1 100 657.3 705.3 176.5 238.6 312.7 396.8 447.0 0 [%] -MS2(529.1) 353.0 100 172.7 254.8 460.7 0 657.2 [%] -MS3(529.2->352.1) 172.7 100 134.7 [%]0 -MS4(529.2->353.0->172.8) 100 93.0 59.4 0 100 200 300 400 500 600 700 m/z

`

Supplemantary Information

TOF data

Compound Compound Molecular Experimental Theoretical Relative Error (ppm) No Formula m/z (M-H+)- m/z (M-H+)-

1 3-CQA C16H17O9 353.0865 353.0878 3.6

2 5-CQA C16H17O9 353.0889 353.0878 3.2

3 4-CQA C16H17O9 353.0883 353.0878 1.3

4 3,5-diCQA C25H24O12 515.1199 515.1195 0.8

5 3,4-diCQA C25H24O12 515.1190 515.1195 0.9

6 4,5-diCQA C25H24O12 515.1183 515.1195 2.3

15 5-CSA C16H16O8 335.0777 335.0772 1.4

16 4-CSA C16H16O8 335.0777 335.0772 1.3

17 3-CSA C16H16O8 335.0778 335.0772 1.8

18 5-FQA C17H20O9 367.1051 367.1035 4.6

11 5-pCoQA C16H18O8 337.0940 337.0929 3.2

21 3,4,5-triCQA C34H30O15 677.1498 677.1512 2.1

- triCQA C34H30O15 677.1517 677.1512 0.8

- triCQA C34H30O15 677.1516 677.1512 0.6

22 1,3,5-triCQA C34H30O15 677.1531 677.1512 2.8

`

Supplemantary Information

OR1 20 11 13 17 9 16 1 10 3 5 7 15

18 CO2R 19

Compound R R1 Molecular Experimental Theoretical Relative Formula m/z (M-H+)- m/z (M-H+)- Error (ppm)

Steviol H H C20H30O3 317.2093 317.2122 9.4

2 1 Steviolbioside H glc - glc C32H50O13 641.3181 641.3179 0.4

Rubusoside glc glc C32H50O13 641.3166 641.3179 2.0

2 1 Stevioside glc glc - glc C38H60O18 803.3751 803.3707 5.5

2 1 Rebaudioside A glc glc3 - glc C44H70O23 965.425 965.4235 1.6

1glc 2 1 Rebaudioside B H glc3 - glc C38H60O18 803.368 803.3707 2.8

1glc 2 1 Rebaudioside C glc glc3 - rham C44H70O22 949.427 949.4286 1.7 (Dulcoside B) 1glc 2 1 2 1 Rebaudioside D glc - glc glc3 - rham C50H80O28 1127.4726 1127.4763 3.3

1glc 2 1 2 1 Rebaudioside E glc - glc glc - glc C44H70O23 965.4199 965.4235 3.7

2 1 Rebaudioside F glc glc3 - xyl C43H68O22 935.4097 935.4129 3.5

1glc 2 1 Dulcoside A glc glc - rham C38H60O17 787.3732 787.3758 3.3

`