DISSERTATION O Attribution

Home , Centaureidin, Chicoric acid

o Attribution — You must give appropriate credit, provide a link to the license, and indicate if changes were made. You may do so in any reasonable manner, but not in any way that suggests the licensor endorses you or your use.

o NonCommercial — You may not use the material for commercial purposes.

o ShareAlike — If you remix, transform, or build upon the material, you must distribute your contributions under the same license as the original.

How to cite this thesis

Surname, Initial(s). (2012). Title of the thesis or dissertation (Doctoral Thesis / Master’s Dissertation). Johannesburg: University of Johannesburg. Available from: http://hdl.handle.net/102000/0002 (Accessed: 22 August 2017). Phytochemical profiling of secondary metabolites from Bidens pilosa plants and cell cultures

by Anza-Tshilidzi Ramabulana (201404841)

Dissertation submitted in fulfilment of the requirements for the degree of Magister Scientiae (MSc) In Biochemistry In the Faculty of Science at the University of Johannesburg

South Africa

Supervisor: Prof. Ian A. Dubery (UJ) Co-supervisor: Dr. Ntakadzeni E. Madala (Univen)

January 2020

Table of Contents

Dedication…………………………………………………………………………………………………… i Acknowledgements………………………………………………………………………………………….. ii List of abbreviations………………………………………………………………………………………… iii List of units………………………………………………………………………………………………….. v Summary……………………………………………………………………………………………………. vi

Chapter One: General Introduction 1.1. Background…………………………………………………………………………………………... 1 1.2. Hypotheses…………………………………………………………………………………………… 2 1.3. Aim………………………………………………………………………………………………….. 3 1.4. Objectives of the study……………………………………………………………………………….. 3 1.5. Workplan……………………………………………………………………………………………… 3 1.6. Outline of the dissertation……………………………………………………………………………. 3

Chapter Two: Literature Review 2.1. Bidens pilosa Linn…………………………………………………………………………………… 5 2.1.1. Medicinal importance of Bidens pilosa ………………………………………………………... 6 2.1.2. Bidens pilosa as a food source …………………………………………………………………. 7 2.2. Plant metabolites: phenolics………………………………………………………………………... 7 2.2.1. Biosynthesis of phenolics………………………………………………………………………. 9 2.2.2. The general phenylpropanoid pathway: synthesis of hydroxycinnamic acids (HCAs)………… 10 2.2.3. Hydroxycinnamic acids ………………………………………………………………………... 11 2.2.3.1. Hydroxycinnamic acid derivatives of quinic acid: chlorogenic acids……………………. 12 2.2.3.1.1. Biosynthesis of chlorogenic acids…………………………………………………... 14 2.2.3.1.2. Biological importance of chlorogenic acids………………………………………… 15 2.2.3.2. Hydroxycinnamic acid-tartaric acid esters………………………………………………….. 15 2.2.3.2.1. Biosynthesis of chicoric acid and other tartaric acid esters……………………...... 16 2.2.3.2.2. Biological importance of chicoric acid………………………………………………... 17 2.2.4. Variations of plant metabolites in different plant tissues…………………………………………. 17 2.2.5. Complexity of chlorogenic acids and related compounds: Analytical methods for separation and identification of regio- and geometric isomers………………………………………………………...... 18 2.3. Plant cell culture and secondary metabolite production……………………………………………. 19 2.3.1. Advantages and disadvantages of plant cell/tissue culture systems………………………………. 19 2.3.2. Elicitation of plant secondary metabolites………………………………………………………... 20 2.4. Plant metabolomics……………………………………………………………………………………. 21 2.4.1. Metabolomic approaches………………………………………………………………………….. 22 2.4.2. General metabolomics workflow…………………………………………………………………. 22 2.4.2.1. Sample preparation………………………………………………………………………….. 24 2.4.2.2. Sample analysis: LC-MS……………………………………………………………………. 25 2.4.2.3. Data mining (data processing and multivariate statistical analysis)………………………… 25 2.4.2.4. Compound identification and biological interpretation……………………………………... 26

Chapter Three: Experimental Procedures 3.1. Sample preparation workflows……………………………………………………………………….. 29 3.1.1. Plant material……………………………………………………………………………………… 29 3.1.2. Callus initiation…………………………………………………………………………………… 30 3.1.3. Cell suspension culture……………………………………………………………………………. 31 3.1.4. Cultivation of callus on different ratios of plant growth regulators………………………………. 31 3.1.5. Treatment of plants with methyl jasmonate (MeJA) and methyl salicylate (MeSA)……………... 32 3.2. Metabolite extractions………………………………………………………………………………… 32 3.3. Sample analysis: data acquisition…………………………………………………………………….. 33

3.3.1. Ultra-high performance liquid chromatography-quadruple time-of-flight spectrometry (UHPLC- qTOF-MS)………………………………………………………………………………………………. 33 3.3.2. Targeted analysis of hydroxycinnamic acid derivatives by UHPLC-qTOF-MS/MS…………… 34 3.4. Multivariate statistical analysis………………………………………………………………………. 34 3.5. Metabolite annotation, identification and relative quantification………………………………….. 35

Chapter Four: Results and Discussion 4.1. Profiling of hydroxycinnamic acid derivatives in leaves, stems and roots of Bidens pilosa………. 36 4.1.1. Multivariate statistical analysis of the phytochemical profiles of extracts from different tissues... 37 4.1.2. Annotation of metabolites………………………………………………………………………… 38 4.1.2.1. Characterisation of feruloylquinic acids (FQAs) and ρ-coumaroylquinic acid (ρCoQA)….. 39 4.1.2.2. Characterisation of caffeoylgycoside………………………………………………………. 40 4.1.2.3. Characterisation of mono-caffeoylquinic acids (CQAs) …………………………………… 40 4.1.2.4. Characterisation of di-caffeoylquinic acids (diCQAs)……………………………………… 41 4.1.2.5. Characterisation of tri-caffeoylquinic acids (triCQAs) and di-caffeoylquinic acid glycosides……………………………………………………………………………………………. 42 4.1.2.6. Characterisation of ρ-coumaroyl-caffeoylquinic acids……………………………………… 43 4.1.2.7. Characterisation of feruloyl-caffeoylquinic acids…………………………………………... 44 4.1.2.8. Characterisation hydroxycinnamoyl-tartaric acid esters……………………………………. 45 4.1.3. Comparison of the tissue-specific distribution and relative abundance of HCA derivatives ……. 50 4.2. Metabolomic profiling of Bidens pilosa leaves with altered metabolomic states induced by 54 stress-related phytohormones……………………………………………………………………………... 4.2.1. Multivariate statistical analysis of treated plant leaves…………………………………………… 56 4.2.1.1. Unsupervised multivariate analysis…………………………………………………………. 56 4.2.1.2. Supervised multivariate analysis……………………………………………………………. 57 4.2.2. Comparative analysis of metabolites identified in leaves subjected to treatment with signal 61 molecules………………………………………………………………………………………………... 4.2.3. Relative quantification of selected secondary metabolites identified in leaves treated with MeSA and MeJA………………………………………………………………………………………………... 66 4.3. Profiling of hydroxycinnamic acid derivatives in cell cultures of Bidens pilosa…………………... 69 4.3.1. Multivariate statistical analysis of cell cultures………………...………………………………… 70 4.3.2. Annotation of metabolites ………………………………………………………………………... 72 4.3.2.1. Characterization of the mono-acyl chlorogenic acids (CGAs) 73 4.3.2.2. Characterization of di-caffeoylquinic acids (diCQAs) and tri-caffeoylquinic acid (triCQA)……………………………………………………………………………………………… 73 4.3.2.3. Characterisation of ρ-coumaroyl-caffeoylquinic acids…...…………….. 74 4.3.2.4. Characterisation of feruloyl-caffeoylquinic acid……………………………………………. 76 4.3.3. Comparison of distribution and relative abundance of HCAs in cell cultures……………………. 81 4.4. Manipulation of undifferentiated cells of Bidens pilosa with plant growth regulators (PGRs)….. 83 4.4.1. Analysis of altered callus metabolomes in response to different plant growth regulator combinations………………………………………………………………………………...... 86 4.4.2. Multivariate statistical analysis of phytochemical profiles / constituents of callus maintained on different plant growth regulator combinations…………………………………………………………... 88 4.4.3. Comparative analysis of metabolites identified in callus of maintained on media with different PGR ratios……………………………………………………………………………………………….. 90

Chapter Five: General Conclusion ………………………………………………………………… 94

Chapter Six: References…………..…………………………………………………………………... 97

Dedication

This dissertation is dedicated to my parents, Mr Simon Godfrey Ramabulana and Mrs Florence Thivhilaeli Ramabulana

Maipfi a ngasi ṱaluse vhudipfi hanga a zwi kona. Ndi livhuwa thikhedzo yavho yothe na lufuno lwe vha ntsumbedza misi yoṱhe. Ndi livhuwa u gudsiwa u funesa Mudzimu na pfunzo nnṱha ha zwithu zwoṱhe.

Acknowledgements

Firstly, I would like to extend my gratitude and appreciation supervisor Prof IA Dubery for giving me the opportunity to learn under your great guidance. Thank you for all the academic insights and motivation to always work hard. I appreciate that you never gave up on me, but taught me with patience. I am also deeply grateful to your wife Mrs Marina Dubery for kindly helping me with all plant cell culture work.

To my co-supervisor Dr Ntakadzeni Edwin Madala, thank you for affording me the opportunity to study and always motivating me to work harder. This project would have never existed without your help. Ndi livhuwa thikhedzo yoṱhe. I would also like to thank my co- supervisor Prof PA Steenkamp, thank you for analysing my extracts.

To Dr Fidele Tugizimana, Mr Msizi Mhlongo, Mrs Imah Mwaba, Prof Lizelle Piater and Mrs Nombuso Buthelezi, Thank you for the constant support and academic assistance. Your guidance and scientific inputs kept me sane. Thank you for all the stimulating conversations and assistance when I needed it the most.

A million thanks to my siblings, Mrs Rofhiwa Mashudu Morathi, Andani Tondani Ramabulana and Edzani Ramabulana. I appreciate the support and for listening to me when I vent. To my sister from another mother Mpho Eulenda Ramabulana, thank you for your constant motivation and always reminding me I could do it.

I would also like to extend my appreciation to my best friends, Tebogo Thapedi, Mamokete Bokhale, Lerato Nephali and Samu Thango. Thank you for being the strongest support system. To the metabolomics group at UJ and the greenlab, thank you for being great colleagues and for all constructive conversations.

I would also like to extend my gratitude to the University of Johannesburg (UJ) and the Department of Biochemistry for the support. This study was also made possible the National Research Foundation (NRF) by their great financial assistance.

Most importantly, I give my thanks and praise to God almighty (a father like no other).

~And he said unto me, My grace is sufficient for thee: for my strength is made perfect in weakness. Most gladly therefore will I rather glory in my infirmities, that the power of Christ may rest upon me~ 2 Corinthians 12:9

List of Abbreviations

NAA 1-Naphthaleneacetic acid 2,4-D 2,4-Dichlorophenoxyacetic acid DAHP 3-Deoxy-D-heptulosonate-7-phosphate DAHPS 3-Deoxy-D-arabino-heptulosonate-7-phosphate synthase 4CL 4-Coumarate: CoA ligase EPSP 5-Enolpyruvylshikimate-3-phosphate ANOVA Analysis of Variance API Atmospheric Pressure Ionisation BPI Base peak intensity BAP Benzylaminopurine B. pilosa Bidens pilosa L. CFA Caffeic acid CQA Caffeoylquinic acid CFQA/FCQA Caffeoylferuloylquinic acid/ feruloylcaffeoylquinic acid CTA Caftaric acid

CaNO3 Calcium nitrate CE-MS Capillary electrophoresis-mass spectrometry

CO2 Carbon dioxide C4H Cinnamic acid 4-hydroxylase IsoCA Citric acid/ Isocitric acid CA Chicoric acid CGAs Chlorogenic acids CoA Coenzyme A CID Collision induced dissociation ρCoQA Coumaroylquinic acid C3H p-Coumarate 3'-hydroxylase CV-ANOVA Cross-validated analysis of variance diCQA Dicaffeoylquinic acid

H2O Dihydrogen monoxide/ water DNA Deoxyribonucleic acid DTIMS-MS Drift tube ion mobility spectrometry-mass spectrometry E. purpurea Echinacea purpurea ESI Electrospray ionisation FA Ferulic acid FQA Feruloylquinic acid FAO Food and Agriculture Organization HIV Human Immunodeficiency Virus H Hydrogen HCA(s) Hydroxycinnamic acid(s) HBA(s) Hydroxybenzoic acid(s) HQT Hydroxycinnamoyl CoA quinate hydroxycinnamoyl transferase HCT Hydroxycinnamoyl CoA shikimate/quinate hydroxycinnamoyl transferase HTT Hydroxycinnamoyl-CoA/ tartaric acid hydroxycinnamoyl transferase HCGQT Hydroxycinnamoyl D-glucose: quinate hydroxycinnamoyl transferase ISCID In-source collision-induced dissociation IUPAC International Union of Pure and Applied Chemistry IAA Indole-3-acetic acid ISR Induced systemic resistance IT Ion Trap IM-MS Ion mobility mass spectrometry-MS K-3-G Kaempferol-3-O-glucoside K-3-AG Kaempferol-3-acetyl-glycoside C_LEAVES Leaf-derived callus CS_LEAVES Leaf-derived suspensions LC-MS Liquid chromatography-mass spectrometry

MgSO4 Magnesium sulfate MS Mass spectrometry MSI Metabolomics Standard Initiative

iii

MeJA Methyl jasmonate MeSA Methyl salicylate MAP mono-Ammonium phosphate MKP mono-Potassium phosphate MVDA Multivariate data analysis MS Murashige and Skoog NTC Non-treated control NMR Nuclear magnetic resonance spectroscopy OAG Okanin acetylglucoside O-diAG Okanin di-acetylglucoside Okanin tri-acetylglucoside Okanin tri-acetylglucoside OPLS Orthogonal projection to latent structures-discriminant analysis PLS Partial least squares PAL Phenylalanine ammonia-lyase PGR(s) Plant growth regulator(s)

KNO3 Potassium nitrate PCA Principal component(s) analysis PC(s) Principal component(s) qTOF Quadrupole Time-of-Flight Q-3-GA Quercetin-3-O-gluconoride Q-3-G Quercetin-3-glycoside QA Quinic acid SiQA Sinapoylquinic acid SIMCA Soft Independent Modeling of Class Analogy SAR Systemic acquired resistance C_STEMS Stem derived callus CS_STEMS Stem derived cell suspensions TFTG Tetrahydroxyflavanone triCQA Tricaffeoylquinic acid TAG Tuberonic acid TAL Tyrosine ammonia-lyase UHPLC Ultra-high performance liquid chromatography UHPLC-qTOF-MS/MS-ISCID Ultra-high performance liquid chromatography-quadrupole time-of-flight mass spectrometry based in-source collision- induced dissociation UGCT UDP glucose: cinnamate glucosyltransferase UN United Nations VIP Variable importance in projection Var. Variant

List of Units

H Hour mg/L Milligram per litre µL/L Microlitre per litre v/v Volume per volume µmol/ m2/ s Micromole per second and square meter ˚C Degree Celsius S Second(s) g/L Gram per litre Mm Millimetre Rt Retention time m/z Mass-to-charge ratio Rpm Revolutions per minute Cm Centimetres mM Millimolar mL Millilitre m/v Gram per volume µL Microlitre Psi Pounds per Square Inch Μm Micrometre Min Minute kV Kilovolts V Volts L/h Litre per hour Da Daltons eV Electron volts G Grams

Summary

Natural products extracted from plants have increasing significance in health as well as drug discovery and development. Plants and their natural products play a significant role in providing alternative ways to treat a variety of diseases. These natural products / ‘neutraceuticals’ comprise, amongst others, of important biologically-active phenolics and antioxidants. B. pilosa is known for its medicinal and nutritional value owning to its rich phytochemical content. Plant metabolomics offers an avenue of comprehensive analysis, elucidation and characterisation of natural products. Various technological advances that couple liquid chromatography and mass spectrometry have been developed which assist elucidation of these products.

B. pilosa has been shown to contain various hydroxycinnamic acids (HCAs) and associated derivatives, quinic acid esters such as chlorogenic acids (CGAs) and tartaric acid esters. In the current study, plant metabolomics techniques were applied to investigate the distribution of HCA derivatives across tissues. Various HCA derivatives, quinic acid esters and tartaric acid esters were identified in extracts of the plant, of which a total of 30 were identified. These were found to be differentially distributed throughout leaves, stems and roots with tartaric esters (chicoric acid and caftaric acid) only occurring in aerial parts and absent in the roots. This might be indicative that an enzyme responsible for the biosynthesis of tartaric acid esters in B. pilosa could be localised in the chloroplasts. HCA derivatives and CGAs were found to be structurally complex considering that they occur as regio- and geometric isomeric forms. In this study, a UHPLC-qTOF-MS/MS-based in-source collision-induced dissociation method was utilised to generate fragmentation data to assist in the differentiation of closely related isomeric forms of HCA derivatives. An ‘in-source collision-induced dissociation’ (ISCID) method was also applied to efficiently discriminate between the isomeric forms of these HCA derivatives.

In Planta, plant secondary metabolites are present in relatively minor concentrations under non- stressed conditions. An efficient strategy to enhance metabolite production is through elicitation. In this study, the effects of two elicitors/signal molecules (methyl salicylate, MeSA and methyl jasmonate, MeJA) were investigated with the aim to enhance secondary metabolites in leaf tissues of B. pilosa. Post multivariate statistical analysis and metabolite annotation, HCA derivatives, flavonoids, a signal molecule (tuberonic acid glucoside/ 12-hydroxyjasmonic acid glucoside) and a primary metabolite (citric acid/ isocitric acid) were identified in extracts from leaves subjected to treatment with MeJA and MeSA. Although these two elicitors function through different signalling pathways, both elicited similar responses which indicate some cross-communication between the signalling pathways. These hormones led to the accumulation of metabolites of the phenylpropanoid pathway for example, MeSA resulted in an enhanced response of trans-5-caffeoylquinic acid and MeJA resulted in an enhanced response of cis-5-caffeoylquinic acid. MeSA and MeJA also resulted in increases levels of

flavonoids such as quercetin-3-O-glucuronide and okanin di-acetylglucoside respectively. However, differential effects on the identified metabolites in response to these signal molecules were also observed.

Secondary metabolites can also be produced in in vitro plant cell culture which offers various advantages such as the ability to scale up cultivation of plant secondary metabolites. In this study, B. pilosa cell cultures (callus and cell suspensions) were established. The distribution of HCA derivatives in B. pilosa cell cultures derived from stems and leaves was studied. As observed in differentiated tissues, cell cultures derived from stem – and leaf explants showed some tissue-specific distribution of HCA derivatives that may be interpreted as a ‘retention’ of metabolomic memory in undifferentiated cells of B. pilosa. Cell suspensions of B. pilosa were shown to produce more HCA derivatives compared to the callus, of which 23 were identified, while in callus only 14 of these HCA derivatives were identified. Stem-derived callus was shown to have higher levels of most HCA derivatives (e.g. 4-caffeoylquinic acid, trans-4- caffeoylquinic acid and 3,5-di-caffeoylquinic acid) compared to leaf-derived callus. In contrast, leaf-derived cell suspensions were found to have higher levels of most HCA derivatives (e.g. tri-caffeoylquinic acid, trans-4-caffeoylquinic acid and 4-feruloyl-5-caffeoylquinic acid) compared to the stem-derived cell suspensions of B. pilosa. Although the number of different HCA derivatives identified in undifferentiated cells of B. pilosa, were less compared to that of differentiated leaf – and stem tissues, strategies to improve metabolite production in cell cultures of B. pilosa were suggested.

Post establishment of cell callus of B. pilosa, the effects of plant growth regulators (PGRs - auxin and cytokinin) were also investigated. In plant cell culture, PGRs can be used to induce regeneration of shoots and roots. In this study, six different PGR combination ratios of 2,4- dichlorophenoxyacetic acid (2,4-D) to benzylaminopurine (BAP) were investigated with the aim to induce organogenesis from B. pilosa callus. Although organogenesis was not achieved, the combination ratios of PGRs were shown to induce differential metabolic responses and thus distinct metabolomic profiles.

In conclusion, B. pilosa was found to be a phytochemically rich plant consisting of a vast number of HCA derivatives. The isomeric forms of these HCA derivatives could be efficiently discriminated, aided by minor fragmentation differences observed in their mass spectral profiles post UHPLC-qTOF-MS/MS-based ISCID. Potential elicitation of production of secondary metabolites in B. pilosa through exogenous treatment with signal molecules (MeJA and MeSA) was shown to be feasible. MeJA and MeSA induced metabolic reprogramming in the metabolome of B. pilosa that involved changes in similar metabolic pathways which possibly indicate interconnected nodes in metabolic networks. Undifferentiated cells derived from stems and leaf tissues retained genetic memory, producing secondary metabolites also identified in corresponding tissues of B. pilosa. Plant cell culture of B. pilosa was demonstrated to be a potential approach for in vitro production of biologically important secondary metabolites. This study also investigated effects due to the manipulation of PGRs by means of different auxin (2,4-D) to cytokinin (BAP) concentration ratios on callus. However, PGR manipulation did not result in redifferentiation of the undifferentiated cells to regenerate shoots

vii

and roots. Regardless, metabolomic profiling of B. pilosa callus demonstrated that the callus retained metabolic responsiveness to external PGRs. This demonstrated that PGRs can be used to manipulate the metabolome of B. pilosa for in vitro production of secondary metabolites.

The combined sets of results add valuable metabolomic insights and new information to the knowledge pool of hydroxycinnamic acids and associated chlorogenic acids of the phytochemically-rich B. pilosa. These results also add novel insights on the metabolomic responses induced by external stimuli on tissues and cell cultures of this plant.

viii

Chapter One: General Introduction

1.1. Background

Natural products research on traditional medicinal plants has gained interest worldwide. Bidens pilosa L (B. pilosa) has been known for its folkloric medicinal uses in various countries including South Africa. Furthermore, B. pilosa has been shown to be an important alternative food crop with high nutritional value as it contains fatty acids, proteins, fibre, carbohydrates, polyphenols and minerals (Adedapo et al., 2011). B. pilosa originated in South America but as a result of its fast reproduction and ability to propagate in most environments, it has since spread all over the world. In South Africa, it has been used to treat ailments such as menstrual and prostate disorders (Arthur, 2012; Bartolome et al., 2013; Semenya et al., 2013). Among others, B. pilosa has been shown to have medicinal properties against malaria (Cortés-Rojas et al., 2013), diabetes (Chien et al., 2009), hypertension (Bilanda et al., 2017), obesity (Liang et al., 2016) and syphilis (de Wet et al., 2012). The health properties of this plant are a result of its rich phytochemical composition consisting of over 300 compounds (e.g. phenylpropanoids, tannins, aromatics and alkaloids) (Bartolome et al., 2013; Xuan and Khanh, 2016).

Chlorogenic acids (CGAs) have been identified as a dominant bioactive group of compounds in B. pilosa (Chiang et al., 2004). CGAs are biologically important secondary metabolites that have been shown to have anti-viral activity by inhibiting HIV-1 integrase (Healy et al., 2009), anti-diabetic (Ferrare et al., 2018), anti-oxidant, anti-inflammatory (Liang and Kitts, 2016) and chemopreventive activities (Chiang et al., 2004). CGAs or hydroxycinnamic acid (HCA) derivatives present a structurally complex group of secondary metabolites. Classically CGAs are described as conjugates between HCAs with quinic acid (QA), but a broader description encompasses further conjugation with sugars and aliphatic side chains (Clifford, 1999; Jaiswal and Kuhnert, 2011; Plazas et al., 2013). In B. pilosa, other HCA derivatives of tartaric acid (chicoric, caftaric, coutaric and fertaric acid) have been identified which were also shown to have many health benefits (Khoza et al., 2016; Tsai et al., 2017; Ferrare et al., 2018). CGAs are also present in various isomeric forms, occurring as both regio-isomers and geometric isomers. The structural complexities of these biologically important compounds pose a challenge in structural elucidation and precise identification. Over the years, mass spectrometric (MS) techniques have been developed to assist in the characterisation and structural elucidation of CGAs (Clifford et al., 2003, 2017). In this study, a method based on ultra-high performance liquid chromatography-quadrupole time-of-flight mass spectrometry (UHPLC-qTOF-MS) linked to in-source collision induced dissociation (ISCID), previously described by Madala et al., (2014), was further optimised for characterisation of CGAs in B. pilosa. Furthermore, metabolomics tools and approaches were followed to allow for deeper insights to be derived from the complex data sets.

Metabolomics is a comprehensive quantitative and qualitative ‘omics’ approach assessing metabolites in a particular biological system under specific physiological conditions (Hall et al., 2002; Kouassi Nzoughet et al., 2017). Metabolomics studies can either be targeted or untargeted, where the latter is hypothesis-generating and aims at comprehensively assessing the metabolites of the defined biological system (Dunn et al., 2013; Ribbenstedt et al., 2018). In contrast, targeted metabolomics is hypothesis-driven and usually a smaller number of predefined metabolites are analysed (Roberts et al., 2012; Dunn et al., 2013; Kouassi et al., 2017). The general metabolomic workflow comprises of sample preparation, data acquisition, and – processing, followed by data mining (Kim and Verpoorte, 2010).

Plants contain a vast number of secondary metabolites, these are present in nature in minute concentrations (Hernández-Sotomayor et al., 2018). Various strategies have been developed that result in accumulation of plant secondary metabolites. Elicitation by exogenous treatments with signalling molecules has been shown to be one strategy that can increase the accumulation of secondary metabolites (Turner et al., 2002; Bi et al., 2007). Traditionally, phytochemicals are extracted from intact plant tissues, however plant tissue culture has gained interest as an efficient method for secondary metabolite production as most plant cells are totipotent in nature. Application of plant cell culture to produce secondary metabolites offers several advantages compared to acquiring secondary metabolites from whole plants; such as the ability to produce metabolites in aseptic environments independent of geographic and seasonal factors (Dorenenburg and Knorr, 1995; Sivanandhan et al., 2014). Plant cell culture is not only important for metabolite production but offers a platform for plant regeneration/organogenesis in culture. Plant growth regulators (PGRs, e.g. auxins and cytokinins) are important in development processes of the plant. Optimisation of concentration ratios of auxin and cytokinins in culture may result in shoot and root regeneration from callus cultures (Pernisova et al., 2009; Su et al., 2011; Schaller et al., 2015).

1.2. Hypotheses

 An optimized UHPLC-qTOF-MS/MS-based in-source collision-induced dissociation (ISCID) method will allow the characterisation of closely related isomers of CGAs from B. pilosa tissues and cultured cells.  Metabolomics will allow for the evaluation of the effects of exogenous treatment with signal molecules (MeJA and MeSA) on the metabolomic profiles of B. pilosa.  B. pilosa cells will undergo organogenesis when grown on media with different ratios of PGRs (auxins and cytokinins) accompanied by differential metabolite profiles in response to varying concentration combinations of PGRs.

1.3. Aim

 To acquire metabolomic insights regarding the presence of hydroxycinnamic acid derivatives and to gain information into the responses induced by external stimuli on tissues and cell cultures of B. pilosa.

1.4. Objectives

 To study the distribution patterns of HCA derivatives among different tissues of B. pilosa grown in controlled environments using LC-MS as bio-analytical technique.  Characterisation and differentiation of the isomeric forms of HCA derivatives in B. pilosa based on a UHPLC-qTOF-MS/MS-based ISCID method.  To assess the effects of exogenous treatments with signal molecules on the metabolomic profiles of B. pilosa leaves.  To compare the distribution patterns of HCA derivatives in plant cell cultures (callus and suspensions) of B. pilosa.  Investigation of the effects of PGRs on organogenesis from B. pilosa callus cultures and their metabolomic profiles

1.5. Workplan

 Grow B. pilosa plants under controlled environmental conditions for a period of two months.  Treat B. pilosa leaves by pressure infiltrating with signal molecules (MeJA and MeSA).  Establish callus of B. pilosa from sterile explant material and utilise friable white callus to establish cell suspensions of B. pilosa in liquid media.  Maintain B. pilosa callus on different concentration combinations of PGRs.  Prepare methanol extracts of cell suspensions, callus, callus grown on different ratios of PGRs and plant tissues (treated and untreated).  Separate and analyse extracts using UHPLC-qTOF-MS as an analytical platform.  Perform a targeted analysis of HCA derivatives using a UHPLC-qTOF-MS/MS ISCID method, to elucidate the isomeric forms of these derivatives.  Perform multivariate statistical analysis, annotate secondary metabolites and relatively quantify these in the different sample types analysed for B. pilosa.

1.6. Outline of the dissertation

This dissertation comprises of six chapters. Following the general introduction (Chapter 1) is the literature review (Chapter 2). The literature review summarizes the biological activities of B. pilosa and the phytochemicals and nutritional content of the plant. The plant metabolomics concepts, and workflow were also outlined. Chapter 3 provides an outline on the

methodologies used in this study, which follows the general metabolomic workflow. Chapter 4 (divided into 4 sections) presents the results obtained and the interpretation thereof. Section 4.1. is the results and discussion of the profiling of HCA derivatives in B. pilosa tissues (stem, leaves and roots), with the subsequent differentiation of the regio-isomers of HCA derivatives based on their varying fragmentation patterns generated by ISCID. Section 4.2 outlines the results and discussion of the metabolomic profiling of B. pilosa leaves with altered metabolomic states induced by stress-related phytohormones. Section 4.3 describes the metabolomic profiling of HCA derivatives in cell cultures of B. pilosa. Section 4.4 outlines the effects of different combinations of PGRs on the metabolite profiles and growth of B. pilosa callus. Lastly, Chapter 5 provides the general conclusions and future perspectives drawn from the study. Chapter 6 comprises of the references used in this dissertation.

Chapter Two: Literature Review

2.1. Bidens pilosa Linn.

Bidens pilosa L (B. pilosa) is an annual flowering, edible herb from the family Asteraceae (Figure 1) (Arthur et al., 2012; Bartolome et al., 2013; Xuan and Khanh, 2016). Three common variants of B. pilosa are B. pilosa var. radiata, B. pilosa var. pilosa, and B. pilosa var. minor (Chien et al., 2009). Common names of Bidens species are based on it bearing sticky seeds (Figure 1) or their rapid growth (Yang, 2014). In South Africa, it is known as ‘Blackjack’, and as ‘Beggar’s tick’ or ‘Needle grass’ in the United States of America and Barbados (Arthur et al.,2012). In South Africa, it has various vernacular names such as ‘Mushidzhi’ in Tshivenda, ‘Umhlabangulo’ in IsiXhosa, ‘Amalenjane’ or ‘Uqandolo’ in IsiZulu and ‘Maphodisa’ or ‘Makolonyane’ in Sepedi (Odhav et al., 2007; Maanda and Bhat, 2010; de Wet et al., 2012; Semenya et al., 2013; Bhat et al., 2016).

This invasive cosmopolitan weed originated from South America and was distributed to other parts of the world. It has been used traditionally as medicine in continents such as Africa, Asia and Oceania (Bartolome et al., 2013). It occurs mostly in tropical and hot areas such as Africa (Bilanda et al., 2017). Favourable conditions for B. pilosa’s growth are moderate dry soil and full exposure to the sun. Regardless of the favourable conditions, it can still grow in very dry soil (Bartolome et al., 2013). The plant can grow up to 1.5 m in height and is characterised by hairy stems, with simple ovate toothed leaves as shown in Figure 1 (Ashafa and Afolayan, 2009). Cultivation of B. pilosa was promoted by the Food and Agriculture Organization (FAO) of the United Nations (UN) because it easy to grow, edible and has a pleasant taste (Liang et al., 2016).

Figure 2.1: Physical appearance of the Bidens pilosa plant (a), its flowers and leaves (b) and sticky seeds (c), (taken from Yang, (2014)).

2.1.1. Medicinal importance of Bidens pilosa

B. pilosa has various bioactivities owing to its diverse phytochemicals, of which over 300 have been identified (Bartolome, et al., 2013; Xuan and Khanh, 2016). These include a variety of aliphatic compounds, flavonoids, terpenoids, phenylpropanoids, tannins, aromatics, alkaloids, cardiac glycosides and porphyrins, some classes of which are shown in Table 1 (Chang et al., 2007; Xuan and Khanh, 2016; Owoyemi and Oladunmoye, 2017). B. pilosa has been reported to treat over 40 diseases such as malaria (Cortés-Rojas et al., 2013), diabetes (Chien et al., 2009), hypertension (Bilanda et al., 2017), obesity (Liang et al., 2016), typhoid, helminthiases (Noumedem et al., 2017) and syphilis (de Wet et al., 2012). In various regions of the world, it has been used as an anti-microbial, anti-inflammatory agent, anti-bacterial agent and as an immunosuppressor attributed to its phytochemicals (Geissberger and Séquin, 1991; Liang et al., 2016). During treatment of disorders and diseases, the whole plant or plant parts of B. pilosa have been used (Owoyemi and Oladunmoye, 2017), generally as a potherb or a form of herbal medicine (Liang et al., 2016). For example, the roots of B. pilosa are used by Bapedi traditional healers to treat menstrual disorders and prostate disturbances (Semenya et al., 2013). B. pilosa extracts can be used in combination with extracts of other plants such as Sarcophyte sanguinea, Clematis brachiata and Rununculus multifidus for treatment of genital sores (de Wet et al., 2012).

2.1.2. Bidens pilosa as a food source

Besides its medicinal importance, B. pilosa is a nutritive leafy vegetable that is high in essential nutrient and vitamins as shown in Table 2.1 (Odhav et al., 2007). B. pilosa is an edible plant with a pleasant taste and has been used as a herb or an ingredient in teas and sauces (Bartolome et al., 2013). This plant is a wild plant, known to contain important fatty acids, proteins, fibre, carbohydrates, polyphenols and minerals but low in toxicants (Adedapo et al., 2011). Although underutilised, wild plant species have an advantage over cultivated plants as they might sometimes have higher nutritional value and are inexpensive to grow. In the Eastern Cape province of South Africa, Xhosa people cut up the leaves and mix them with maize meal while the Vhavenda people in Limpopo harvest the leaves and consume it as a wild vegetable (Maanda and Bhat, 2010; Bhat et al., 2016). Apart from South Africa it has been used as a substitute for tea or served as an addition to chamomile tea in Mexico, while in the Philippines it has been used to make Igorot rice wine by mixing the leaves and/or flowers with par-boiled rice (Morton, 1962).

Table 2.1: Nutritional content of dried and raw plant of Bidens pilosa (adapted from Bartolome, Villaseñor and Yang, (2013)).

2.2. Plant metabolites: phenolics

Plant metabolites (small molecular weight molecules that are intermediary products of metabolism) can be classified as either primary or secondary metabolites (Tiwari and Rana, 2015; Hong et al., 2016). Primary metabolites e.g. carbohydrates, lipids and amino acids are

responsible for growth development of the plant while secondary metabolites are produced for the plant’s survival under biotic or abiotic stress condition (Hong et al., 2016; Scossa et al., 2016; Hernández-Sotomayor et al., 2018). Main classes of secondary plant metabolites are terpenoids, sulphur-containing compounds (e.g. glucosinolates), nitrogen-containing alkaloids and phenylpropanoids and are responsible for plant diversity considering that over 250 000 of these have been identified (Scossa et al., 2016; Hernández-Sotomayor et al., 2018; Kundu and Vadassery, 2019).

Figure 2.2: Schematic representation of the different groups of the phenolic compounds in plants (Taken from Goleniowski et al., (2013)).

Plant secondary metabolites play several roles in survival and adaptation of the plant to its environment. These metabolites protect plants from microorganisms (bacteria, viruses, fungi and environmental conditions such as ultraviolet (UV) radiation (Bourgaud et al., 2001; Taofiq et al., 2017). Apart from the prior functions, plants may synthesize specialised secondary metabolites that have anti-nutritive effects towards herbivores and can also act as signalling molecules to attract pollinators and seed dispersal animals (Kundu and Vadassery, 2019). Secondary plant metabolites are also useful to humans as cosmetics, nutraceuticals, ethnomedicines, pharmaceuticals, bio-pesticides, flavours, fragrances, food additives and recreational drugs (Sivanandhan et al., 2014; Tiwari and Rana, 2015; Xuan and Khanh, 2016; Taofiq et al., 2017).

The main focus of this study will be on pharmacologically important phenolics (Figure 2.2), that are products of the shikimate -, pentose phosphate - and the phenylpropanoid pathways. They consist of a common carbon skeleton comprising of a hydroxylated benzene ring (Randhir et al., 2004; Mandal et al., 2010; Siqueira et al., 2012; Goleniowski et al., 2013). Their structures range from simple aromatic rings with one or two hydroxyl groups attached to complex polymerised compounds (Chen et al., 2016). Phenolic acids are widely distributed in plants and may be found as glycosides or as esters with steroids, alcohols or other natural compounds (Ghasemzadeh and Ghasemzadeh, 2011).

2.2.1. Biosynthesis of phenolics

The shikimate pathway (also known as chorismate biosynthetic pathway) shown in Figure 2.3A, is the pathway responsible for the formation of chorismate, a product used in the biosynthesis of aromatic amino acids; tyrosine, phenylalanine and tryptophan (Tzin and Galili, 2011). These aromatic amino acids are not only important in protein synthesis, but also serve as precursors of various secondary metabolites. Amongst these amino acids, phenylalanine is used as a precursor in the phenylpropanoid pathway shown in Figure 2.3 (Tzin and Galili, 2010, 2011; Fraser and Chapple, 2011). The first step of the shikimate pathway makes use of an intermediate of glycolysis (phosphoenolpyruvate) and an intermediate of the pentose phosphate pathway (erythrose-4-phosphate) in a condensation reaction, that produces 3-deoxy- D-heptulosonate-7-phosphate (DAHP), catalysed by 3-deoxy-D-arabino-heptulosonate-7- phosphate synthase (DAHPS) (Tzin and Galili, 2011). The formation of DAHP is followed by the downstream formation of shikimate which, is preceded by ring closure and a dehydration step. The substrate for 5-enolpyruvylshikimate-3-phosphate (EPSP) synthase is formed following the phosphorylation of shikimate. Chorismate is then formed through the removal of a phosphate group of EPSP (Herrmann, 1995; Fraser and Chapple, 2011). As shown in Figure 2.3B, phenylalanine is synthesised through the action of chorismate mutase (which uses chorismate as a substrate) yielding prephenate. Prephenate aminotransferase then catalyses the formation arogenate, an intermediate that is used by arogenate dehydratase in the final step yielding phenylalanine (Herrmann, 1995; Hoffmann et al., 2003; Hudson and Prabhu, 2010; Tzin and Galili, 2010). An alternative route (not shown in Figure 2.3) has been described, where phenylalanine is synthesised through a microbial-like phenylpyruvate pathway that utilises a cytosolic tyrosine: phenylpyruvate aminotransferase (Yoo et al., 2013).

Figure 2.3: Schematic representation of the shikimate pathway (A) that leads to the formation of chorismate, that is used as a common precursor in the biosynthesis of aromatic amino acids (taken from Tzin and Galili, (2010)). The end-product of the shikimate pathway is used to synthesise tyrosine and phenylalanine through a series of steps (B) (taken from Hudson and Prabhu, (2010)).

2.2.2. The general phenylpropanoid pathway: synthesis of hydroxycinnamic acids (HCAs)

The phenylpropanoid pathway is one of the most important metabolic pathways that use phenylalanine, which is synthesised from chorismate for the downstream biosynthesis of lignin, flavonoids, chlorogenic acids, lignans and anthocyanins (Nugroho et al., 2002; Fraser and Chapple, 2011). Phenylpropanoid metabolism is initialised in specific tissues in response to external environmental stimuli or stressors (Douglas, 1996). The general phenylpropanoid pathway is catalysed by three enzymes; phenylalanine ammonia-lyase (PAL), cinnamic acid 4- hydroxylase (C4H) and 4-coumarate: CoA ligase (4CL). The first three steps catalysed by these enzymes have been researched extensively and are illustrated in Figure 2.4 (Lepelley et al., 2007; Goleniowski et al., 2013). The first step in this pathway is the deamination of either phenylalanine or tyrosine by PAL or tyrosine ammonia-lyase (TAL), respectively forming trans-cinnamic acid/ p-OH-cinnamic acid and ammonia (Fraser and Chapple, 2011; Vargas- Tah and Gosset, 2015). Activities of TAL are often shown in fungi and monocotyledons. PAL catalyses the first most important regulatory steps in the biosynthesis of these phenolics in plants (Mandal et al., 2010). Further reactions are then carried out by hydroxylases and O- methyltransferases, resulting in hydroxycinnamic acids (HCAs) from which downstream phenylpropanoids are derived (Douglas, 1996). Cinnamic acid is hydroxylated by the action of C4H, a cytochrome P450-dependent monooxygenase, to form p-coumaric acid also known as 4-coumarate (Anterola et al., 2002; Fraser and Chapple, 2011). 4CL activity results in the formation of 4-coumaroyl CoA/ p-coumaroyl CoA in an ATP-dependent reaction (Comino et

al., 2007; Fraser and Chapple, 2011). p-Coumaroyl-CoA is used as a precursor in the formation of common HCAs like caffeic acid, ferulic acid and sinapic acid, which have a common carbon skeleton (C6-C3) as shown in Figure 2.5 (Fraser and Chapple, 2011; Goleniowski et al., 2013; Ozcan et al., 2014; Vargas-Tah and Gosset, 2015).

Figure 2.4: Three routes (1-3) proposed for the biosynthesis of hydroxycinnamic acid derivatives through the phenylpropanoid pathway (taken from Payyavula et al., (2015)). Key enzymes in the biosynthesis of the chlorogenic acid are PAL (phenylalanine ammonia-lyase), C4H (cinnamate 4- hydroxylase), 4CL (4-hydroxycinnamoyl CoA ligase), HCT (hydroxycinnamoyl CoA shikimate/quinate hydroxycinnamoyl transferase), C3H (p-coumarate 3'-hydroxylase), HQT (hydroxycinnamoyl CoA quinate hydroxycinnamoyl transferase), UGCT (UDP glucose: cinnamate glucosyltransferase), HCGQT, hydroxycinnamoyl D-glucose: quinate hydroxycinnamoyl transferase. (Payyavula et al., (2015)

2.2.3. Hydroxycinnamic acids

Phenolic acids are a group of polyphenols divided into two major groups namely are HCAs and hydroxybenzoic acids (HBAs), based on their common carbon structures; C6-C3 and C6- C1 respectively (Robbins, 2003; Călinoiu and Vodnar, 2018). The HCAs and HBAs are derived from cinnamic - and benzoic acid precursor molecules respectively, and have at least one substitution with a hydroxyl group (Heleno et al., 2015). The HCAs comprise of ferulic -, caffeic -, sinapic - and ρ-coumaric acid while the HBAs comprise ρ-hydroxybenzoic -, vanillic -, syringic - and gallic acids shown in Figure 2.2 (Călinoiu and Vodnar, 2018). HCAs are the larger group compared to the HBAs and are regarded as the biggest subgroup of phenolics. HCAs are commonly found in coffee, tea leaves, vegetables and fruits and they make up one- third of the phenolics in human dietary consumption (Teixeira et al., 2013; Vargas-Tah and

Gosset, 2015; Taofiq et al., 2017). HCAs are structurally diverse as they are characterised by a C6-C3, chemical backbone with a C3 side chain attached that may have either cis- or trans- configuration. Furthermore, they may also occur as esters with hydroxy acids (e.g. quinic acid), tartaric acids, sugar derivatives and glycosides or amides with polyamines, amino acids and peptides (Ozcan et al., 2014; Meinhart et al., 2017; Călinoiu and Vodnar, 2018). HCAs serve as precursor molecules for the biosynthesis of many other molecules such as chalcones, lignin, flavonols and anthocyanins as shown in Figure 2.4 (Alam et al., 2016).

2.2.3.1. Hydroxycinnamic acid derivatives of quinic acid: chlorogenic acids

Chlorogenic acids are classically described as a group of esters between HCAs (e.g. caffeic (CFA) -, ferulic (FA) -, р-coumaric (р-CoA) - and sinapic acid (SA)) and a quinic acid (1L- 1(OH),3,4/5-tetrahydroxycyclohexane carboxylic acid, QA) molecule as shown in Figure 2.5 (Clifford, 1999; Clifford et al., 2005; Jaiswal et al., 2010). Sometimes the quinic acid can form esters with trimethoxycinnamic - and dimethoxycinnamic acids (Jaiswal and Kuhnert, 2011). Chlorogenic acids are largely abundant in coffee in which more than 60 different CGA derivatives have been identified. These include caffeoylquinic, feruloylquinic, coumaroylquinic, caffeoyl-feruloylquinic, diferuloylquinic, dimethoxycinnamoylquinic and dicaffeoylquinic acids, some of which are indicated in Figure 2.5 (Clifford et al., 2006; Kaiser et al., 2013). In most plants, these derivatives are formed in response to abiotic stress to aid in the protection of the plant (Lallemand et al., 2012; Taofiq et al., 2017). They can also be produced in response to biotic stress as they have been shown to be effective against several insect herbivores that prey on plants (Kundu and Vadassery, 2019). Major groups of chlorogenic acids are caffeoylquinic acids (CQAs) and dicaffeoylquinic acids (1,3-diCQA, 1,5- diCQA, 3,4-diCQA, 3,5-diCQA and 4,5-di-CQA) (Zheng et al., 2017). The basic CGAs can be divided into several groups depending on the number and position of the acyl groups attached to the quinic acid. They can be divided into mono-esters, di-esters, tri-esters and tetra esters (Clifford, 1999). The complexity of these acids is vast, as mixed di-esters of caffeic acids with other HCAs may form, such as caffeoylferuloylquinic acid, caffeoylsinapoylquinic acid and caffeoylcoumaroylquinic acid (Jaiswal et al., 2010; Wianowska and Gil, 2019). Quinic acid may also bear aliphatic side chains such as tartaric -, malic -, succinic - and fumaric acid, making the CGAs even a more complex group of secondary metabolites (Clifford, 1999; Jaiswal and Kuhnert, 2011; Plazas et al., 2013).

Name Abbreviation R1 R2 R3 3-O-caffeoylquinic acid 3-CQA C H H 4-O-caffeoylquinic acid 4-CQA H C H 5-O-caffeoylquinic acid 5-CQA H H C 3-O-coumaroylquinic acid 3-ρCoQA ρCo H H 4-O-coumaroylquinic acid 4-ρCoQA H ρCo H 5-O-coumaroylquinic acid 5-ρCoQA H H ρCo 3-O-feruloylquinic acid 3-FQA F H H 4-O-feruloylquinic acid 4-FQA H F H 5-O-feruloylquinic acid 5-FQA H H F 3-O-sinapoylquinic acid 3-SiQA Si H H 4-O-sinapoylquinic acid 4-SiQA H Si H 5-O-sinapoylquinic acid 5-SiQA H H Si 3,4-di-O-caffeoylquinic acid 3,4-diCQA C C H 3,5-di-O-caffeoylquinic acid 3,5-diCQA C H C 4,5-di-O-caffeoylquinic acid 4,5-diCQA H C C 3,4-di-O-coumaroylquinic acid 3,4-diρCoQA ρCo ρCo H 3,5-di-O-coumaroylquinic acid 3,5-diρCoQA ρCo H ρCo 4,5-di-O-coumaroylquinic acid 4,5-diρCoQA H ρCo ρCo 3,4-di-O-feruloylquinic acid 3,4-diFQA F F H 3,5-di-O-feruloylquinic acid 3,5-diFQA F H F 4,5-di-O-feruloylquinic acid 4,5-diFQA H F F 3,4-di-O-sinapoylquinic acid 3,4-diSiQA Si Si H 3,5-di-O-sinapoylquinic acid 3,5-diSiQA Si H Si 4,5-di-O-sinapoylquinic acid 4,5-diSiQA H Si Si 3-O-caffeoyl-4-O-feruloylquinic acid 3C-4FQA C F H 3-O-caffeoyl-5-O-feruloylquinic acid 3C-5FQA C H F 4-O-caffeoyl-5-O-feruloylquinic acid 4C-5FQA H C F 3-O-feruloyl-4-O-caffeoylquinic acid 3F-4CQA F C H 3-O-feruloyl-5-O-caffeoylquinic acid 3F-5CQA F H C 4-O-feruloyl-5-O-caffeoylquinic acid 4F-5CQA H F C 3-O-coumaroyl-4-O-caffeoylquinic acid 3ρCo-4CQA ρCo C H 3-O-coumaroyl-5-O-caffeoylquinic acid 3ρCo-5CQA ρCo H C

4-O-coumaroyl-5-O-caffeoylquinic acid 4ρCo-5CQA C ρCo H 3-O-caffeoyl-4-O-coumaroylquinic acid 3C-4ρCoQA C ρCo H 3-O-caffeoyl-5-O-coumaroylquinic acid 3C-5ρCoQA C H ρCo 4-O-caffeoyl-5-O-coumaroylquinic acid 4C-5ρCoQA H C ρCo 3,4,5-tri-O-caffeoylquinic acid 3,4,5-triCQA C C C

Figure 2.5: Structures of some hydroxycinnamic acid derivates of quinic acid, including mono-, di-, tri- and hetero-chlorogenic acids.

2.2.3.1.1. Biosynthesis of chlorogenic acids

These complex secondary metabolites may be formed through three biosynthetic routes indicated in Figure 2.4 within the phenylpropanoid pathway, with the hydroxycinnamoyl-CoA quinate hydroxycinnamoyl transferase (HQT) mediated pathway as the main route (Zhang et al., 2017). As shown in route 1, Figure 2.4, HQT catalyses the reversible synthesis of GGAs by meditating an interchange of a CoA thioester and a quinate group, resulting in an esterification between caffeic acid coenzyme A (caffeoyl-CoA) and quinic acid, forming CGAs as products (Comino et al., 2007; Sonnante et al., 2010). Although synthesis of CGAs has been proven to be primarily through the route using HQT, other alternative routes have been described. These include the second route, which involves activities of hydroxycinnamoyl- CoA shikimate hydroxycinnamoyl transferase (HCT) and p-coumarate 3’ hydroxylase (C3H) in the synthesis of p-coumaroylquinate and hydroxylation, respectively, shown as route 2 in Figure 2.4. The presence of HCT and C3H in Arabidopsis with no accumulation of CGAs disputes the importance of this route (Chen et al., 2015; Payyavula et al., 2015). HCT also shows greater substrate specificity for shikimate compared to quinate, while HQT shows more affinity for quinate. In some plants, HCT could be supplemented by HQT (Sonnante et al., 2010; Lallemand et al., 2012), which has resulted in augmented levels of CGAs.

Lastly, CGAs may be synthesised in a pathway that utilizes caffeoyl-D-glucose as an activated intermediate and is converted to CGA by D-glucose: quinate hydroxycinnamoyl transferase as indicated as route 3 in Figure 2.4 (Lallemand et al., 2012; Chen et al., 2015). Overexpression of the HQT-gene in Coffea canephora has been proven to be directly related to high CGA accumulation (Lepelley et al., 2007). Plants normally synthesize these compounds as trans- isomers but, because of UV radiation, cis-isomers may also result (Clifford et al., 2008; Masike et al., 2018) Interestingly, the UV-induced cis-isomers of CGAs have been shown to have notable anti-tuberculosis activity significantly higher than that of the trans-counterparts, more importantly against the multiple-drug resistant Mycobacterium tuberculosis (Chen et al., 2011).

2.2.3.1.2. Biological importance of chlorogenic acids

Consumption of plants as food has been shown to be correlated to reduced human diseases or illnesses and this is attributed to the antioxidant content of the plants (Lepelley et al., 2007). CGAs are the major plant-derived antioxidants in the human diet and are found in beverages such as coffee, tea, fruit juice, wine and herbal concoctions (Liang and Kitts, 2016). These antioxidant metabolites are present in most plants consumed by humans such as artichoke, potato, tomato, apples, pear, eggplant and plums (Sonnante et al., 2010; Payyavula et al., 2015). CGAs have been shown to have many health benefits such as the inhibition of the mutagenicity of carcinogens and described as potential chemosensitizers leading to suppression of tumour cell growth (Kasai et al., 2000; Lukitasari et al.,2018). Moreover, the vicinal hydroxyl groups attached to the aromatic group of chlorogenic acids are responsible for this activity through scavenging of reactive oxygen species (Jaiswal and Kuhnert, 2011; Sato et al., 2011; Zhang et al., 2017). Interestingly, CGAs are reported to have anti-HIV properties, for instance, 3,5- dicaffeoylquinic acid binds the DNA integrase enzyme irreversibly, inhibiting the integration of the viral DNA into the host genome and consequently inhibiting HIV-1 replication (Robinson et al., 1996; Makola et al., 2016; Zheng et al., 2017). Other health benefits associated with CGAs include anti-hypersensitive - (Zhao et al., 2011) and anti-inflammatory activity (Zhao et al., 2011; Liang and Kitts, 2016). In studies performed in vivo and in vitro CGAs have been reported to be possibly related to prevention of chronic and cardiovascular diseases (Meinhart et al., 2017). Furthermore, these acids have been shown to regulate glucose and lipid metabolism (Meng et al., 2013). CGAs have also been shown to stimulate glucose uptake in skeletal muscles and to have anti-diabetic effects (Ong et al., 2013). They may also confer protection of neurons as CGAs prevent non-enzymatic glycosylation (Meinhart et al., 2017).

2.2.3.2. Hydroxycinnamic acid-tartaric acid esters

Similar to the esterification of HCAs to quinic acid, HCAs can also be esterified to tartaric acid as an acceptor substrate. A well-known example is chicoric acid (CA), an ester of tartaric acid and two caffeic acids, also known as dicaffeoyl-tartaric acid or 2,3-dicaffeoyl-L-tartaric acid. Chemically CA is described by its IUPAC name as ((2R,3R)-2,3-bis{[(2e)-3-(3,4- dihydroxyphenyl)oxy}succinic acid) (Lee and Scagel, 2013; Chhipa et al., 2014). This hydroxycinnamoyl-tartaric acid ester was first found in Cichorium intybus L. commonly known as chicory but has since been found in other plants such as iceberg lettuce, grapes, basil, spinach and in the leaves of other plants from the Asteraceae family such as B. pilosa and Echinacea purpurea (Nüsslein et al., 2000; Chkhikvishvili and Kharebava, 2001; Lee and Scagel, 2010; Khoza et al., 2016). CA is an important hydroxycinnamoyl-tartaric acid ester which is produced by plants to protect themselves from biotic stressors (fungi, viruses, bacteria and nematodes) as well as abiotic stressors (reactive oxygen species, UV radiation and ozone) and aids in repair after mechanical destruction of the plant (Lee and Scagel, 2013; Chhipa et al., 2014; Sullivan, 2014). In E. purpurea, chicoric acid is one of the major phenolics, but other

tartaric acid esters are present such as caftaric acid, feruloyltartaric acid, and caffeoyl- coumaroyltartaric acid, the mono-tartaric acid esters have also been found in grapes are also found in grapes and the leafy vegetable B. pilosa L. (Nüsslein et al., 2000; Vanzo et al., 2007; Khoza et al., 2016). These hydroxycinnamoyltartaric esters normally occur in the trans- configuration, but due to the double bond in their lateral sides, they may present cis-isomers (Vrhovšek, 1998; Vanzo et al., 2007).

2.2.3.2.1. Biosynthesis of chicoric acid and other tartaric acid esters

Chicoric acid is synthesized through the phenylpropanoid/ shikimate pathway and the synthesis of this acid is comparable to that of the CGAs. This polyphenol is stable in dry conditions but undergoes enzymatic breakdown in moist conditions (Lee and Scagel, 2013; Chhipa et al., 2014). The pathways involved in the synthesis of mono- and di-caffeoyltartaric acid esters are not supported by any molecular data but rather by biochemical experiments. These esters, just like CGAs, are synthesised by acyltransferase enzymes of the BAHD family (Legrand et al., 2016). The key enzyme in the synthesis of hydroxycinnamoyl-tartaric esters downstream the phenylpropanoid pathway has been hypothesised/ shown to be hydroxycinnamoyl-CoA/ tartaric acid hydroxycinnamoyl transferase (HTT) through esterification of caffeoyl-CoA or other hydroxycinnamoyl-CoAs and tartaric acid as shown in Figure 2.6. Genes encoding this acyl-transferase (HTT) still remain to be further characterised (Murthy et al., 2014; Sullivan, 2014).

Figure 2.6: A schematic representation showing the enzymatic reaction catalysed by hydroxycinnamoyl-CoA/ tartaric acid hydroxycinnamoyl transferase (HTT) in the synthesis of tartaric acid esters. (Taken from Sullivan, (2014)).

2.2.3.2.2. Biological importance of chicoric acid

Chicoric acid is a phenolic compound that has been recently reported in B. pilosa and has a wide range of biological activities such as antiviral, immunostimulatory and antioxidant properties (Khoza et al., 2016; Kuban-Jankowska et al., 2016). Similarly to CGA, one of the important antiviral activities of chicoric acid is the potential to inhibit HIV-integrase and inhibit replication of HIV (Healy et al., 2009; Chhipa et al., 2014). Chicoric acid has been shown to exhibit anti-diabetic properties as it improves uptake of glucose / increases glucose tolerance by up-regulating the secretion of insulin and antihyperglycemic effects (Lee and Scagel, 2010; Azay-Milhau et al., 2013; Ferrare et al., 2018). Other biological activities of CA include in vivo and in vitro stimulation of phagocytosis and the inhibition of hyaluronidase, an important enzyme in bacterial infections, thus inhibiting bacterial infections (Kuban-Jankowska et al., 2016). CA and its precursor metabolites (caffeic acid and caftaric acid) have been shown to be natural antioxidants by acting as scavengers of free radicals (Liu et al., 2017; Tsai et al., 2017). Elsewhere, CA has been experimentally proven to suppress the production of reactive oxygen species (Chhipa et al., 2014; Liu et al., 2017). Other biological activities of CA include anti- hepatitis B virus properties and have been shown experimentally using natural CA found in C. intybus (Zhang et al., 2014). CA shows a reduction of telomerase activity and induction of apoptosis, resulting in an inhibitory effect towards colon cancer cells (Tsai et al., 2012).

2.2.4. Variations of plant metabolites in different plant tissues

Differences in the amounts of secondary metabolites do not only vary from one plant to the next, but also among different tissues of the same plant (Achakzai et al., 2009). For instance, production of CA may vary among plant families, species and even different cultivars (Lee and Scagel, 2013). It may also vary within the different organs of the plant. For example, on a study done on different plant tissues of E. purpurea, it was found that the apical parts of the plants contained more CA than the lower parts of the plant (Lin et al., 2011). Another factor that affects the synthesis of CA is the age of the tissue. More mature tissues in some plants have been found to produce lower levels of CA compared to younger tissues (Lee and Scagel, 2013). Similarly, production of secondary metabolites such as CGAs can vary between plant species and between plant variants, and environmental factors and agricultural practices also influence the synthesis of these secondary metabolites (Farah and Donangelo, 2006). Environmental conditions such as salinity stress may result in an increase in CGA biosynthesis in plants through an increase of PAL activity in the phenylpropanoid pathway. (Yan et al., 2016). Post- harvest processing such as drying, washing, and brewing of coffee plant material influences the amount of secondary metabolite content such as CGA isomers (Farah and Donangelo, 2006; Lin et al., 2011; Liang and Kitts, 2016). Depending on the stimulus that triggers the production of a particular metabolite, its distribution will differ among different parts of the plant (Cortés- Rojas et al., 2013). In this study distribution of different HCA derivatives in different tissues of B. pilosa and in plants growing in different environments was investigated

2.2.5. Complexity of chlorogenic acids and related compounds: Analytical methods for separation and identification of regio- and geometric isomers

As briefly described in section 2.3.1., CGAs are a complex group of phenolics and these exhibit different physicochemical properties depending on the acyl moiety’s identity, number and its position when esterified to the quinic acid (Wianowska and Gil, 2019). Furthermore, CGAs occur in various isomeric forms, such as UV radiation-induced geometrical isomers. Although plants are thought to naturally synthesise trans-isomers more abundantly, evidence of the occurrence of cis-isomers has also been shown in plant cell cultures upon treatment with resistance-inducing elicitors (Mhlongo et al., 2015). As mentioned above, CGAs may be present as different regio-isomers, as the quinic acid has several positions where HCAs and other compounds may conjugate. Identification and quantification of structurally related compounds is challenging and may require the use of reliable advanced analytical techniques. In previous studies, geometric isomers have been differentiated using elution order but this has been shown to be unreliable (Zheng et al., 2017). Elution order of cis/ trans isomers has been shown to be dependent on chromatography parameters and the chemical nature of column stationary phases (Masike et al., 2018). Discrimination between geometric isomers has been recently achieved by liquid chromatography-mass spectrometry (LC-MS), where cis-isomers of diCQAs were shown to preferentially bind to alkali metals (sodium adduct-MS), making it possible to differentiate them from the trans-isomers (Makola et al., 2016). Development of LC-ion trap (IT)-MS-based hierarchical keys for the differentiation of acyl-quinic acids using mild fragmentation conditions has allowed for confidently assaying most CGA regio-isomers using their MSn fragmentation patterns (Clifford et al., 2003, 2017). Other advanced technologies have allowed for the differentiation of positional isomers of CGAs using ion mobility mass spectrometry-MS (IM-MS). More advanced analytical techniques such as drift tube ion mobility spectrometry-mass spectrometry (DTIMS-MS) are required to elucidate the cis/ trans isomers of diCQAs (Zheng et al., 2017). In recent years, major progress has been made for the structural elucidation of HCA derivatives. However, more innovations are required for the full characterisation of these metabolites and their isomeric forms. These isomeric forms of these metabolites have also not been fully characterised in B. pilosa. Previous studies by Madala et al., (2014), have shown possibilities of differentiating CGA regio-isomers based on an ultra-high performance liquid chromatography-quadrupole time-of-flight mass spectrometry (UHPLC-qTOF-MS) method linked to in-source collision induced dissociation (ISCID). In the current study this method was further investigated and optimised for the differentiation of both regio- and geometric isomers (including di-esters substituted with different acyl-moieties) present in B. pilosa extracts.

2.3. Plant cell culture and secondary metabolite production

Plants produce large numbers of secondary metabolites that are important in the production of flavours, fragrances, pharmaceuticals, food additives and many other applications (Rao and Ravishankar, 2002; Sivanandhan et al., 2014). Conventionally, the extraction of secondary metabolites has been performed on whole plants and is regarded as non-cost effective (Ling et al., 2011). Extractions from whole plants pose environmental threats such as reduction of local plant populations and loss of plant genetic diversity (Sivanandhan et al., 2014). Environmental instabilities may result in medicinal plants extinction making it difficult to acquire secondary metabolites from plants (Mulabagal and Tsay, 2004). Plant cell culture has become an attractive approach as it permits acquiring of secondary metabolites produced in parent plants independent of geographical and seasonal factors (Hellwig et al., 2004; Karuppusamy, 2009; Ling et al., 2011). Callus culture is described as culture that is formed from undifferentiated parenchyma cells on semi-solid media. Callus can be used to initiate cell suspension cultures and subsequent plant regeneration/ organogenesis. In-vivo, callus is initiated from wounding and is controlled by auxin and cytokinin (Brown, 1990; Kumlay and Ercisli, 2015). Plant suspension cultures are in vitro model systems that have been widely used to produce secondary metabolites in either fermenters or shaker flasks. They are mostly derived from friable callus (undifferentiated cells) that is initially grown on solid media, and transferred to shakers or fermenters with media, growth regulators and nutrients such as simple sugars resulting in single cell cultures or cell aggregates (Fischer et al., 1999; Rao and Ravishankar, 2002; Hellwig et al., 2004). In literature limited information is known about cell culture initiation and metabolite production from B. pilosa cell cultures, hence this study aims to study metabolite production across cell cultures of B. pilosa.

2.3.1. Advantages and disadvantages of plant cell/tissue culture systems

Cell culture offers an advantage over differentiated whole plants, such as faster metabolism and has condensed biosynthetic cycles (Dorenenburg and Knorr, 1995). Other benefits of using cell/organ cultures are that important phytochemicals can be produced throughout the year, independent of season and environmental conditions with short cultivation periods (2-4 weeks). Cell culture also allows for optimal growth conditions in an aseptic environment to be implemented (Dorenenburg and Knorr, 1995; Sivanandhan et al., 2014). Plant cell cultures are advantageous as they can be used to scale up cultivation of plant secondary metabolites, and these metabolites can be mass produced cost efficiently over continuous periods of time (Mulabagal and Tsay, 2004; Al-sane et al., 2005). In cell culture, defined metabolite production systems, with efficient downstream product recovery can be achieved. Novel compounds can also be produced independently of the metabolites produced in the parent plant (Rao and Ravishankar, 2002). Cell suspension cultures can be used to scale up the production of different metabolites in bioreactor systems (Ochoa-Villarreal et al., 2016).

Although plant cell/ tissue culture is a promising alternative to using whole plants for metabolite production, there are some limitations to this approach, such production of only certain/restricted metabolites in culture (Rao and Ravishankar, 2002; Qu et al., 2011). Little or no desired products may be produced that may also pose difficulties in scaling up the production of metabolites (Lee and Scagel, 2010; Hamany Djande et al., 2019). Biosynthetic instability of some cell lines are some of the negatives of using plant cultures (Rao and Ravishankar, 2002; Qu et al., 2011). However, different strategies have been proposed to counter the disadvantages of plant cultures, such as enhancing the genes responsible for the biosynthesis of desired metabolites. This is not always an easy approach as little information may be known about the genes (Hernández-Sotomayor et al., 2018). Other strategies that can increase yield of secondary metabolite production include; elicitation by stress inducers, plant cell immobilisation, precursor feeding, in situ product removal, use of good cell lines with high productivity, and scaling-up of production of metabolites in appropriate bioreactors (Rao and Ravishankar, 2002; Ling et al., 2011; Qu et al., 2011).

2.3.2. Elicitation of plant secondary metabolites

Plants use phytochemicals, particularly secondary metabolites they produce for resistance against pathogens and to tolerate abiotic stressors, as part of their innate immune systems (Iriti and Faoro, 2009). Secondary metabolites do not only play a role in the defence of plants but are also a good source of pharmaceuticals (Rao and Ravishankar, 2002). Many of the plants’ secondary metabolites are found in cells at low amounts, hence it is important to use in vitro methods that can lead to accumulation of medicinally-important secondary metabolites (Hernández-Sotomayor et al., 2018). Accumulation of secondary metabolites is favourable when plants are exposed to signal molecules and elicitors and these methods are also applied in plant cell culture (Zhao et al.,2005). Elicitation of cell suspension cultures (with biotic and abiotic stressors or elicitors) has been used effectively to trigger signal transduction pathways which in turn increases metabolite production (Zhang et al., 2004; Reddy et al., 2016). Plants growth regulators, which are hormonal molecules (phytohormones) and their synthetic counterparts, may be used in the elicitation of secondary metabolite production (Jamwal et al., 2018). Plant growth regulators such as 2,4-dichlorophenoxyacetic acid (2,4-D), indole-3-acetic acid (IAA), 1-naphthaleneacetic acid (NAA) have been shown to support the accumulation of certain secondary metabolites (Ramakrishna and Ravishankar, 2011). Plant defence is activated locally and systemically by signalling pathways involving jasmonates and salicylates. Molecules from these pathways can be used for effective elicitation for plant secondary metabolite production. Jasmonic acid (JA) and salicylic acid (SA) are required in the activation of many defence-related genes (Kalaivani et al., 2016). Exogenous treatments with signal molecules such as JA, SA, methyl jasmonate (MeJA) and methyl salicylate (MeSA) have been shown to increase production of metabolites such as phenolics (Turner et al., 2002; Bi et al., 2007). Current knowledge on responses of B. pilosa to external stimuli is limited, hence in the current study effects of growth regulators and signal molecules on metabolite production were investigated.

2.4. Plant metabolomics

Metabolomics is an ‘omics’ approach and is downstream of the genomic, epigenomic, transcriptomic and proteomic levels (Figure 2.7). Metabolomics is generally defined as the study of products from cellular metabolism, known as metabolites or low molecular weight molecules (<1500 Da) such as organic acids, amino acids, sugars and nucleotides (Sawada et al., 2009; Skibiński and Komsta, 2015; Tan et al., 2016; Kouassi et al., 2017). Metabolomics is termed a ‘global approach’ as it aims at a comprehensive, non-biased qualitative and quantitative analysis of metabolites in a biological system of interest under specific physiological conditions (Hall et al., 2002; Brown et al., 2005; Tugizimana et al., 2013; Gika et al., 2016). This approach gives a snapshot of the plant metabolome, which refers to the complete set of metabolites that corresponds to a set of genes in a biological system at a given time and physiological condition (Hegeman, 2010; Hong et al., 2016). Both endo- and exogenous metabolites, which are used or produced in chemical reactions, can be investigated during metabolomics studies (Liu and Locasale, 2017). Metabolites are involved in different cellular events and give information about upstream genes, transcripts, proteins and may give information about the absolute physiological state of the cell (Tautenhahn et al., 2012; Kumar et al., 2017).

Figure 2.7: Illustration indicating interconnections of the ‘omics’ approaches (Taken from Stringer et al. (2016)).

It is usually difficult to estimate or measure the complete metabolome of any biological system despite the simplest systems however, integrated techniques such as nuclear magnetic resonance spectroscopy (NMR) and mass spectrometry (MS) based techniques such as GC (gas chromatography)-MS and LC (liquid chromatography)-MS and CE (capillary electrophoresis)-

MS enable large scale analysis of highly complex mixtures of metabolites, thereby generating large multidimensional datasets (Hegeman, 2010; Hong et al., 2016; Vasilev et al., 2016).

2.4.1. Metabolomic approaches

Metabolomics strategies can be either untargeted (non-targeted) or targeted approaches (Gika et al., 2016; Kouassi et al., 2017). Ways to differentiate between these two approaches have been described and include the level of quantification achieved, number of metabolites determined and whether the study aims to generate a hypothesis or test a hypothesis. Another criterion in differentiating between these methods is the need for metabolite identification and structural elucidation (Dunn et al., 2013; Ribbenstedt et al., 2018). Untargeted metabolomics/ non-targeted approach is referred to as the ‘ultimate’ technique that aims to determine all measurable metabolites in the sample being analysed, by integrating various techniques. Conceptually, untargeted metabolomics can be described as a data-driven, hypothesis- generating science. This technique only allows for semi-quantification where differences between two populations or conditions are monitored or compared (Schrimpe-Rutledge et al., 2016; Li et al., 2018). Quantification in non-targeted approaches can also be termed as relative quantification/ semi-quantitative as relative changes in metabolite concentrations are monitored (Sawada et al., 2009; Wang et al., 2012; Liu and Locasale, 2017; Liang et al., 2018). This approach, that screens thousands of unknown features offers the advantage of identification of novel metabolites that can be used as biomarkers (Guijas et al., 2018). Given this approach’s advantages, challenges in identifying biomarkers may arise, as a sample may contain many metabolites with different physicochemical properties, concentrations and chemical classes (Gika et al., 2016). Untargeted metabolomics has been shown to be extremely difficult and time consuming (Sawada et al., 2009). However various recently developed software libraries, web-based tools and services are available to reduce the complexity of untargeted metabolomics (Li et al., 2018).

In contrast, targeted approaches aim to quantify and analyse a certain number of metabolites that are pre-selected (chemically and biochemically characterised) (Roberts et al., 2012; Dunn et al., 2013; Kouassi et al., 2017). Targeted metabolomics is a hypothesis-driven approach and aims to quantify a subset of pathway-specific metabolites (Kapoore and Vaidyanathan, 2016). Advantages of targeted LC-MS approaches include absolute quantification achieved by use of an internal standard, and quicker analysis as fewer resources are required to profile the specified metabolites (Tautenhahn, et al., 2012; Wang et al., 2012). However, targeted metabolomics is considered to have a low throughput, which does not support large scale network analysis (Li et al., 2013).

2.4.2. General metabolomics workflow

The general metabolomics workflow comprises of three interrelated consecutive experimental steps, which are sample preparation, data acquisition and data mining (Kim and Verpoorte,

2010; Lee et al., 2010; Kim et al., 2016). The analytical method of choice requires certain properties of the analyte and determines the sample preparation method (Kim and Verpoorte, 2010; Tugizimana et al., 2013). The schematic chart for the general metabolomics workflow is illustrated in Figure 2.8, and the steps mentioned above precede biochemical interpretation. After sample preparation has been completed, global metabolomic data is obtained from various metabolomic platforms that can be NMR- or MS-based (Zhang et al., 2012). Bio- informatic algorithms can then be used to perform quantitative and qualitative analysis to find peaks that are different across different samples and extract biologically meaningful information (Gardinassi et al., 2017). Considering that the metabolomic tool/ platform used in sample analysis shown in Figure 2.8 are high-throughput methods, generating large-scale data covering both known and unknown metabolites, it is, therefore, necessary to reduce the dimensionality of data by applying chemometric multivariate statistical methods. Post chemometric analysis, database searches and MS/MS analysis can be done on targeted metabolites (e.g. identified discriminatory markers) to assist in metabolite annotations and biochemical interpretations (Tautenhahn, et al., 2012; Tugizimana et al., 2013; Skibiński and Komsta, 2015; Vasilev et al., 2016; Kumar et al., 2017).

Figure 2.8: A schematic representation of the metabolomics workflow for plant metabolomics studies (Tugizimana et al., 2013; Cambiaghi et al., 2017).

2.4.2.1. Sample preparation

Sample preparation depends on the type of sample and the experimental design of the study being done or the research question. The steps in sample preparations include harvesting of samples, rapid metabolism quenching/ drying, sample storage and metabolite extraction as shown in Figure 2.8 (Kim and Verpoorte, 2010; Tugizimana et al., 2013; Andersson et al.,

2019). Sample preparation is a critical step (although not given much attention compared to other steps) in the metabolomics workflow and care must be taken not to introduce variability when preparing plant/cell samples. Enzymatic degradation and oxidation must also be taken into consideration. During harvesting, enzymatic reactions must be quenched immediately to stop metabolomic changes from occurring and this can be done by immediate freezing of samples in liquid nitrogen. Quenching of material allows for a snapshot of the real metabolomic state (Roessner and Bacic, 2009; Álvarez-Sánchez et al., 2010a; Kim and Verpoorte, 2010; Tugizimana et al., 2013). Quenching procedures should meet the following requirements: interruption of the metabolism should occur faster than the metabolomic changes in the sample, care should be taken to preserve the integrity of the samples (mostly for cells) and the quenching method should not induce physical or chemical variation within samples. When the quenching procedure is done, samples must be amenable to subsequent analytical methods (Álvarez-Sánchez et al., 2010b). As mentioned in section 2.3, metabolite concentrations or distribution in plants may differ because of environmental factors, hence environmental conditions should be considered as an important cause of variation within samples, unless plants are grown in controlled environments and plant tissues such as roots, stems and leaves should be separated during harvesting. Apart from the use of liquid nitrogen as a quenching procedure, various other methods are used such as drying (e.g. freeze drying, ambient air drying and oven drying) to remove water from samples that can still permit for enzymatic activity during sample preparation. Drying also inhibits any microbial growth and facilitates long term storage of samples (Kim and Verpoorte, 2010; Mushtaq et al., 2014).

Different extraction methods can be used, and in attempt to obtain a maximum amount, or ideally all metabolites in a sample, extraction methods can be combined. There is currently no method that ideally extracts all metabolites in a biological sample, as the sample may contain a vast array of metabolites with vast complexities ranging from different polarities and occurring at different levels in the samples (Kim and Verpoorte, 2010; Bijttebier et al., 2016). The extraction method is dependent on the nature of the sample of interest, where metabolites from solid samples are extracted through solid-liquid extractions such as supercritical fluid extraction, ultrasound-assisted extraction or microwave-assisted extraction and Soxhlet extraction (Kaufmann and Christen, 2002; Zhou et al., 2013). Liquid samples are extracted by using techniques such as liquid-liquid extraction and solid-phase extraction (Juhascik and Jenkins, 2009). The most common extraction method used is the solvent extraction method in combination with ultrasonication which increases extraction efficacy (Álvarez-Sánchez et al., 2010b; Kim and Verpoorte, 2010). During solvent extraction, it is important not to keep extracts in solvents and sunlight for too long as this could induce formation of structural artefacts, isomerisation and decomposition of sample constituents. Before sample analysis, extracts can be concentrated in a rotavapor at temperatures below 40℃ to avoid degradation of thermolabile metabolites (Jones and Kinghorn, 2006; Sauerschnig et al., 2018).

2.4.2.2. Sample analysis: LC-MS

Plants are complex organisms producing a wide range of metabolites that are diverse in physico-chemical properties, structures and in abundance, adding complexity to metabolomic studies. Two techniques are commonly used to acquire metabolomics data namely NMR- or MS-based techniques (Hong et al., 2016; Kumar et al., 2017). MS-based techniques are coupled with chromatographic techniques, to offer relatively fast, sensitive and selective metabolomic analysis done both quantitatively and qualitatively (Dunn and Ellis, 2005; Kumar et al., 2017). MS analysis can be enhanced significantly using liquid chromatography, gas chromatography and capillary electrophoresis techniques, which separate analytes, prior MS analysis based on their physicochemical properties (Tugizimana et al., 2013; Piasecka et al., 2019). Mass spectrometers generate ions and separate them based on their mass-to-charge (m/z) ratios and subsequently detect the ions post-separation (Dunn and Ellis, 2005; Balmer et al., 2013; Tugizimana et al., 2013; Kumar et al., 2017). LC-MS can be used to deteremine a wide range of mid-polar and non-polar organic compounds, except those on the extreme spectrum of volatility. Non-polar (also mildly polar metabolites) and polar metabolites can be chromatographically separated on reversed phase columns and on normal phase columns, respectively. Ionisation sources used in LC-MS analysis are usually Atmospheric Pressure Ionisation (API) and most often electrospray ionisation (ESI). Mass analysers such as quadrupole mass filters, quadrupole ion traps and time-of-flight are used (Dunn and Ellis, 2005; Tugizimana et al., 2013). In the current study, ultra-high performance liquid chromatography- quadrupole time-of-flight mass spectrometry (UHPLC-qTOF-MS) and UHPLC-qTOF- MS/MS were employed to characterise and structurally elucidate the secondary metabolites of interest, the hydroxycinnamic acids and derivatives.

2.4.2.3. Data mining (data processing and multivariate statistical analysis)

Metabolomics techniques (e.g. LC-MS) generate large metabolome data sets and a three-way data matrix is obtained that is comprised of retention time (Rt), peak intensity and mass spectra (m/z) directions (Tugizimana et al., 2016). The aim of data mining processes (data pre- processing and data pre-treatment) is to extract valuable information from raw data, eliminating noise, artefacts and adducts. Data processing involves the creation of a data table (numerical data matrix) from the MS raw data. Data mining can be performed through elimination of background noise, peak alignment, sample concentration normalisation, centering, transformation and data scaling (van der Berg et al., 2006; Godzien et al., 2013; Tugizimana et al., 2016; Piasecka et al., 2019). The above-mentioned data processing steps may be achieved by various commercialised software packages such as MarkerLynx (Waters Corporation), Progenesis QI (Waters Corporation), MetabScape (Burker Corp.), and Mass Hunter (Agilent). These softwares follow a similar pipeline of peak alignment, peak detection and metabolite annotation (Tugizimana et al., 2016; Piasecka, Kachlicki and Stobiecki, 2019). Open source tools are freely available for metabolomics data processing such as MetAlign,

MetaboAnalyst, OpenMS, MZmine, XC-MS and MVEN (Tomita et al., 2012; Gowda et al., 2014; Piasecka et al., 2019).

In metabolomics studies, an assumption is made, whereby it is assumed that data collected contains underlying information of biological significance to the biological phenomenon under investigation (Saccenti et al., 2014). Data generated is complex and difficult to interpret, but chemometric statistical tools assist with data reduction, visualisation and summarization (Wiklund et al., 2008; Godzien et al., 2013). Two statistical approaches can be applied to metabolomic data, which are univariate - or multivariate analysis. Univariate statistical methods analyse one variable (a metabolite that is increased or decreased between classes) at a time, and the significance of a single peak or variable is assessed. ANOVA (Analysis of Variance) and the Student’s t-test are used for univariate analysis. Multivariate data analysis (MVDA) takes into consideration two or more variables simultaneously (Goodacre et al., 2007; Godzien et al., 2013; Tugizimana et al., 2013; Saccenti et al., 2014). The two most popular methods of MVDA are principal component(s) analysis (PCA) and partial least squares (PLS) or orthogonal projection to latent structures-discriminant analysis (OPLS-DA). These statistical methods can be described as either unsupervised or supervised model respectively (Tomita et al., 2012; Tugizimana et al., 2013; Worley and Powers, 2015). PCA modelling aims to indicate systemic trends, outliers, pattern recognition and clustering in the dataset and is based on the linear transformation of metabolomic features. During OPLS-DA analysis, additional information on the relation between the dependent and independent variables are defined, thus ‘supervised’ (Wiklund et al., 2008; Tugizimana et al., 2013; Madala et al., 2014; Saccenti et al., 2014; Alonso et al., 2015; Worley and Powers, 2015; dos Santos et al., 2018). Open source software can be used to perform statistical analysis such as MetaboAnalyst and MetaGeneAlyse, also commercial chemometric tools are available such as SIMCA (Soft Independent Modeling of Class Analogy) (Umetrics) (Daub et al., 2003; Fukushima and Kusano, 2013; Tugizimana et al., 2013).

2.4.2.4. Compound identification and biological interpretation

Biological interpretation of metabolomic data, metabolite identification and functional analysis are interrelated steps. In metabolite identifications, molecular properties of metabolites are taken into consideration whereby experimentally determined accurate mass and mass spectral patterns are used to deduce empirical and molecular formulae of metabolites (Tugizimana et al., 2013; Watson, 2013). The Metabolomics Standard Initiative (MSI) has outlined four levels of which identified metabolites may be reported in literature. These are: Identified compounds (level 1), with chemical standards utilised and at least two orthogonal techniques; putatively annotated metabolites (level 2), which are based on spectral similarities and/ physicochemical properties; putatively characterized compound classes (level 3), which are based on similarities of physico-chemical properties or spectral properties with a known group of compounds and level 4, described as unknown metabolites which can still be quantified and differentiated based on their spectral data (Sumner et al., 2007; Viant et al., 2017). Metabolite identification that

entails the assigning of the correct chemical structure to each detected feature (unique combination of mass-to-charge ratio) is a challenging task as each feature may match to hundreds of chemical structures. Fragmentation can be performed and the resulting spectra can be compared to reference libraries (using computational tools) and experimental references (Erbilgin et al., 2019). With regards to the unique complexities of the chlorogenic acids (section 2.2.5), an LC-ion trap (IT)-MS-based hierarchical key (Figure 2.9) for the differentiation of acyl-quinic acids using mild fragmentation conditions has been developed to allow for the confident annotation of most CGA regio-isomers using their MSn fragmentation patterns (Clifford et al., 2003; 2017). Online databases that can assist with metabolite annotations (or at least the name of the identifier) include KNApSAcK (www.knapsackfamily.com), Chemspider (www.chemspider.com), Pubchem (pubchem.ncbi.nlm.nih.gov) and Dictionary of Natural Products (dnp.chemnetbase.com) (Matsuda et al., 2009; Afendi et al., 2012; Marco-Ramell et al., 2018). Following metabolite annotations, the metabolites can be mapped within the metabolic context of the organism (Rosato et al., 2018).

Figure 2.9: Hierarchical scheme for identification by LC-MSn of mono- and diacylchlorogenic acids not substituted at position 1 (Clifford et al., 2003).

Chapter Three: Experimental Procedures

3.1. Sample preparation workflows

In this experimental section, the general metabolomic approach was followed, that consists of sample preparation, sample analysis, data acquisition and data mining (Kim and Verpoorte, 2010). Sample preparation is a critical step in the metabolomic workflow as it influences subsequent metabolite extraction and analysis, data acquisition and biological interpretation. Caution in sample preparation is required as this can lead to incorrect biological interpretation from collected data (Chen et al., 2018). Sample preparation involves selection of starting material, harvesting, metabolic quenching leading towards metabolite extractions and reconstitution of samples that will be used for LC-MS analysis as outlined below (Figure 3.1) (Jorge, Mata and António, 2016).

Figure 3.1: Schematic representation of sample preparation for LC-MS analysis that was followed in this study (adapted from Jorge, Mata and António, (2016)).

3.1.1. Plant material

B. pilosa L. is a wild plant that is consumed as food (green leafy vegetable) or tea and herbal medicine. Its cultivation has been promoted by the Food and Agriculture Organization (FAO) of the United Nations (UN) as this plant is palatable and grows well in warm and moist environments. It is rich in phytochemicals of which over 200 have been identified and characterised (Arthur, 2012; Bartolome, Villaseñor and Yang, 2013; Liang et al., 2016; Singh et al., 2017).

B. pilosa seeds were collected from mature plants in the wild (Venda area of South Africa) and air-dried at room temperature. The seeds were cold shocked at 4˚C for 48 h; this was performed as a way of cold-wet stratification to break seed dormancy in summer perennials (Schütz and

Rave, 1999). The seeds were then sown in Culterra germination mix (Culterra, Muldersdrift, South Africa, http://culterra.co.za). Germinated plants were then grown under greenhouse conditions at 28˚C for a period of two months. The plants were watered twice a week and fertilised once every two weeks with a fertiliser containing 90 mg/L mono-potassium phosphate (MKP), 150 mg/L Soluptase, 20 mg/L Microples, 40 µL/L Kelp-P-Max, 650 mg/L CaNO3, 400 mg/L KNO3, 300 mg/L MgSO4, and 90 mg/L mono-ammonium phosphate (MAP). Plants stems, roots and leaves were harvested and immediately shock-frozen in liquid nitrogen to quench all metabolic reactions (Maier, Kuhn and Müller, 2010; Santos et al., 2016; Gong et al., 2017). The frozen plant tissues were stored at -80˚C, pending metabolite extractions.

3.1.2. Callus initiation

Production of plant secondary metabolites using in vitro cell culture has gained interest as it provides many advantages compared to plant systems such as no seasonal and environmental dependence for secondary metabolite production (Eibl et al., 2018). Cell callus is initiated from undifferentiated parenchyma cells (from wounded explant materials). Callus initiation is dependent on the ratios of auxins and cytokinins hormones. Usually, high auxin concentrations are used or a suitable ratio is chosen. Once initiation occurs, the cells can undergo somaclonal variation and, post several sub-culturing cycles, the callus can reach genetic stability. This most often coincides with stabilisation of growth, resulting in homogenous cell aggregates that produce metabolites uniformly (Brown, 1990; Filova, 2014; Bourgaud et al., 2001; Kumlay and Ercisli, 2015).

B. pilosa stem and leaf calli were initiated from sterilised explant material taken from plants grown under greenhouse conditions mentioned in section 3.1.1. Explant material was sterilised with 70% (v/v) ethanol for 10 s, then 1.5% (v/v) sodium hypochlorite solution for 20 min and rinsed with sterile distilled water. Cultures were initiated on Murashige and Skoog (MS) medium (Murashige and Skoog, 1962) supplemented with MS vitamins (0.5 mg/L nicotinic acid, 0.2 mg/L thiamine and 0.5 mg/L pyridoxine. Furthermore, to the MS medium 100 mg/L myo-inositol, 1 g/L casein hydrolysate, 30 g/L sucrose and 8 g/L phytoagar were added. In addition, the growth medium was supplemented with plant growth regulators: 0.45 mg/L 2,4- dichlorophenoxyacetic acid (2,4-D) and 1 mg/L 6-benzylaminopurine (BAP) at pH 5.8. Callus was grown in an incubator at 24 C and light intensity of 25 µmol/ m2/ s. Callus was sub- cultured onto fresh media every 14 d until callus growth stabilised and the callus was white and friable (approximately 3 months). The callus was harvested and stored at -80˚C, these cells were then used for subsequent metabolite extractions.

3.1.3. Cell suspension culture

Cell growth in suspension culture is performed in aqueous medium, under controlled sterile conditions and maintained with superior homogeneity of treatments compared to plant tissues. These are initiated using friable callus as starting material (Hellwig et al., 2004; Laura et al., 2018). Use of plant cell cultures offers many advantages such as the production of valuable secondary metabolites throughout the year, independent of the availability of plants, environmental conditions and seasonal changes and offers an environmentally friendly and renewable way of metabolite production (Dorenenburg and Knorr, 1995; Dörnenburg, 2008).

Cell suspension cultures of B. pilosa were initiated by sub-culturing friable callus that was initiated following the protocol explained in section 3.1.2. Friable callus was sub-cultured into MS medium supplemented with MS vitamins, with 0.45 mg/L 2,4-D and 1.0 mg/L BAP (pH 5.8) and grown in shaker flasks at room temperature at 120 rpm with a light/dark cycle of 12 h/ 12 h and maintained at low light intensity of 30 µmol/ m2/ s. The cultured cells were harvested using filter papers (70 mm diameter) on a vacuum filtration system and the cells were stored at -80˚C until metabolite extractions.

3.1.4. Cultivation of callus on different ratios of plant growth regulators

As mentioned in chapter 2, section 2.3.2 plant growth regulators (PGRs) can be used to elicit the accumulation of plant secondary metabolites in tissue culture. Furthermore, callus induction requires the use of appropriate concentration ratios of PGRs and an appropriate ratio may lead to plant regeneration due to totipotency of callus cells (Elaleem et al., 2009; Ikeuchi et al., 2013).

B. pilosa callus was initiated as explained in section 3.1.2. from explant material. The friable calli were subcultured on MS media with vitamins and different concentration ratios of auxin: cytokinin as listed in Table 3.1. in order to investigate the effect of combining auxin and cytokinins on undifferentiated B. pilosa cells.

Table 3.1: Concentration ratios of auxin (2,4-D) and cytokinin (BAP) used to grow B. pilosa callus on MS media. Condition number 2,4-D (mg/L) BAP (mg/L) Ratio (2,4-D: BAP) Condition-1 (1) 0.2 2.0 1:10 Condition-2 (2) 2.0 0.2 10:1 Condition-3 (3) 0.0 0.0 - Condition-4 (4) 0.45 1.0 1:2 Condition-5 (5) 0.3 4.0 1:20 Condition-6 (6) 0.2 8.0 1:40

3.1.5. Treatment of plants with methyl jasmonate (MeJA) and methyl salicylate (MeSA)

MeJA and MeSA are plant signalling molecules that are produced in plants for the development of resistance to insects and pathogens as described in chapter 2, section 2.3.2. (Repka et al., 2004). Exogenous applications of MeJA and MeSA to plants imitates plant responses similar to that when plants are attacked by pathogens or herbivores and can enhance the production of secondary plant metabolites (Turner et al., 2002; Bi et al., 2007; Stella de Freitas et al., 2019).

B. pilosa was grown under greenhouse conditions as described in section 3.1.1. for 8 weeks (~ 35 cm in height). Plants were removed from the greenhouse and three plants (biological replicates) were treated with 0.5 mM of MeJA and another three were treated with 0.5 mM MeSA and the last three replicates were non-treated controls (NTC). Solutions of MeJA and MeSA were prepared with 8 mM of magnesium sulphate. Plants for treatment were pressure infiltrated (with 1 mL blunt-end syringes) with a solution containing the MeJA or MeSA, while the NTC was pressure infiltrated with a solution containing 8 mM MgSO4, excluding the signalling molecules. Leaf material was harvested at 12 h and 24 h time-points for both treated and NTC plants. These leaf materials were immediately placed in liquid nitrogen and stored at -80℃ until metabolite extractions were performed.

3.2. Metabolite extractions

As outlined in chapter 2, section 2.4.2.1, a combination of extraction methods results in better yield of metabolites, hence in this study solvent extraction and ultra-sonication were used. Two grams (2 g) of the samples; cell suspensions, callus, callus grown on different ratios of 2,4-D: BAP, plant tissues (crushed frozen with liquid nitrogen using a mortar and pestle), treated plant material (crushed) were weighed and homogenised using a probe homogeniser at 100% intensity for 2 min in 20 mL (1:10 m/v) of 80% analytical grade methanol. Samples were sonicated for 30 min at 30°C in a bath sonicator. The crude extracts were centrifuged at 5100 rpm for 15 min and supernatants were evaporated under vacuum on a rotary evaporator at 55°C to approximately 1 mL. Samples were transferred to 2 mL Eppendorf microcentrifuge tubes and dried to completion overnight at 55°C in a heat block and fume hood (operating at negative vacuum conditions). The dried residues were then reconstituted with 500 µL of 50% methanol, sonicated for 30 min at 30°C followed by filtration using 0.22 µm nylon syringe filters into glass chromatography vials fitted with 500 µL inserts. Samples were capped and stored at 4°C until future analysis. To ensure experimental reproducibility, at least three independent biological replicates were prepared and three instrumental technical replicates were analysed (Ncube et al., 2014).

3.3. Sample analysis: data acquisition

Several approaches to elucidate plant metabolomes are available. MS-based techniques coupled with chromatographic separation offer sensitivity, speed, repeatability and selectivity (Piasecka et al., 2019). In this study ultra-high performance liquid chromatography-quadrupole time-of-flight mass spectrometry (UHPLC-qTOF-MS) was utilised with the parameters described below. Separation prior to MS allows for discrimination of isomers, which later reduces ion suppression whereby an ion would mask another ion which is not easily ionizable (Berg et al., 2013). UHPLC allows for increased resolution, speed and sensitivity based on the use of small porous particle sizes and high pressure (6000 psi) pumps (Chawla and Ranjan, 2016). UHPLC coupled with q-TOF-MS provides high resolution spectra resulting in accurate structural information of unknown compounds (Ma et al., 2019). However, unlike ion trap systems, q-TOF systems can only provide MS2 fragmentation data, hence the need of performing in-source fragmentation followed by conventional MS/MS fragmentation to obtain comprehensive fragmentation patterns (Baira et al., 2018). A novel method was previously optimised in our research group that indicated the use of UHPLC-qTOF-MS in-source collision-induced dissociation (ISCID) to differentiate between closely related isomers by mimicking MS2 and MS3 fragmentation in an ion trap-based MS (Madala et al., 2014). This method was also used to assist in accurate structural elucidation.

3.3.1. Ultra-high performance liquid chromatography-quadrupole time-of-flight mass spectrometry (UHPLC-qTOF-MS)

A UHPLC high-definition quadrupole time-of-flight MS instrument (UHPLC-qTOF SYNAPT G1 HDMS system, Waters Corporation, Manchester, UK) fitted with an Acquity HSS T3 C18 column (150 mm × 2.1 mm with particle size of 1.7 μm) (Waters, Milford, MA, USA) were used to chromatographically analyse the extracts. A sample volume of 3 µL was injected and the column was housed in a column oven thermo-stated at 60 C. A binary solvent system was used consisting of solvent A: 0.1% formic acid in Milli-Q water and solvent B: acetonitrile (UHPLC grade) with 0.1% formic acid. A binary solvent gradient (with solvent A and B) with a flow rate of 0.4 mL/min was used to separate analytes over 30 min. The separation conditions were: 2% B over 0.0-2.0 min, 2-60% B over 2.0-24 min, 60-95% B over 24–25 min, from 25- 27 min the conditions were maintained at 95% B and the column was washed 95-2% B over 27–28 min. The column was then allowed to re-equilibrate 2% B over a 2 min isocratic wash.

The chromatographic effluent was further analysed utilizing the SYNAPT G1 high definition mass spectrometer operating in both positive and negative electrospray ionisation (ESI) modes. The MS conditions were set as follows: capillary voltage of 2.5 kV, sampling cone voltage of 30 V, extraction cone of 4.0 V, source temperature of 120°C, cone gas flow of 50.0 L/h, desolvation gas flow of 550 L/h, m/z range of 100-1000, scan time of 0.2 s, interscan delay of 0.02 s, mode: centroid and lockmass: leucine-enkephalin (556.3 Da). For downstream structural elucidation, the MS analyses were set to result in both unfragmented and fragmented

analytes by ramping the collision energy from 15 to 60 eV in a series of for fragmentation experiments (MSE ).

3.3.2. Targeted analysis of hydroxycinnamic acid derivatives by UHPLC-qTOF-MS/MS

Prior to UHPLC–MS/MS analysis of extracts from B. pilosa cell suspensions and plant tissues, extracts were chromatographically separated as mentioned in section 3.3.1 above. A targeted approach was followed as fragmentation patterns of pre-selected hydroxycinnamic acid derivatives with parent ions [M-H]- at m/z 677 (tricaffeoylquinic acids), m/z 529 (feruloycaffeoylquinic acids), m/z 499 (coumaroylquinic acid), m/z 515 (dicaffeoylquinic acid), m/z 367 (feruloylquinic acid), m/z 353 (caffeoylquinic acids), m/z 337 (coumaroylquinic acid) were chosen for MS2analysis. Fragmentation was achieved by the ISCID method as explained where the collision energy (10-40 eV) and the cone voltage (10-100 V) were optimised to produce fragment ions of chlorogenic acids at m/z 191 [quinic acid-H]-, m/z 179 [caffeic acid- - - - H-H2O] , m/z 173 [quinic acid-H-H2O] and m/z 135 [caffeic acid-H-CO2] as described by Madala et al. (2014).

3.4. Multivariate statistical analysis

Prior to multivariate data analysis (MVDA), raw UHPLC-qTOF-MS data was processed using MarkerLynx XS™ software (Waters, Manchester, UK) utilising the following parameters: retention time (Rt) range of 0.50-22 min, mass range of 100-1000 Da, mass tolerance of 0.05 Da and a Rt window of 0.2 min. The two-way data matrix obtained from MassLynx was exported into ‘Soft Independent Modeling of Class Analogy’ software (SIMCA-15.0, Umetrics Corporation, Umea, Sweden) for statistical modelling. Statistical models computed were principal component analysis (PCA) and hierarchical cluster analysis (HCA), which are unsupervised models that show trends, clusters and relationships between samples.

In addition, orthogonal projection to latent structures-discriminant analysis (OPLS-DA), which gives a descriptive exploration and predictive analysis respectively of the data, where also computed. Unless stated otherwise, for all statistical models herein, Pareto scaling was applied (Bartel et al., 2013; Tugizimana et al., 2013; Worley and Powers, 2015). Model quality was evaluated by R2, which explains or reflects the goodness-of-fit of the model, and Q2, which reflects the model predictability (Godzien et al., 2013). Generated OPLS-DA models were also validated by a response permutations test and cross-validated analysis of variance (CV- ANOVA). Potential biomarkers of different sample groups were highlighted in the OPLS-DA S-plot, and only significant biomarkers with [P(corr)] of ≥ 0.5 and covariance of (p1) ≥ 0.5 were putatively identified (Eriksson, Trygg and Wold, 2008; Huan et al., 2017).

3.5. Metabolite annotation, identification and relative quantification

Accurate metabolite annotation is essential for accurate downstream metabolomic interpretation. Several criteria are defined for metabolite annotations for LC-MS such as the use of accurate mass, Rt and MS/MS fragmentation patterns. Accordingly, different levels of metabolomic annotation and identification have been defined (Sumner et al., 2007; Chaleckis et al., 2019). Metabolite annotation was executed based on mass spectral information from MSE and/or MS2 experiments, accurate mass information, elemental composition predictions and searches in various databases such as ChemSpider (www.chemspider.com) and Dictionary of Natural Products (www.dnp.chemnetbase.com). Mass spectral information was also compared to available literature and the metabolites were putatively identified due to the lack of available authentic standards. Metabolites were putatively identified to level 2 of the Metabolomics Standards Initiative (MSI) (Sumner et al., 2007).

MetaboAnalyst 3.0 (www.metaboanalyst.ca) was utilised to construct clustered color-coded heat-maps for visualisation and comparison of relative abundance/changing of metabolites identified (Xia et al., 2015). Peak intensities of identified metabolites were imported for exploratory statistical analysis into MetaboAnalyst. Data was normalised by median and log transformation was performed followed by Pareto-scaling of data sets, this was performed to reduce systematic variance within features. The computed HCA indicated samples with relatively similar abundances. The Pearson’s correlation was applied as a dissimilarity measure and Ward’s clustering algorithm was used. To simplify the visualisation of the changing patterns, group averages were used (Xia et al., 2009). Furthermore, partial least square- discriminant analysis (PLS-DA) was computed from MetaboAnalyst to mine data of B. pilosa leaves treated with MeJA and MeSA (section 4.2). This was performed to investigate the time- dependent responses of treated plants in comparison to NTCs. Box-and- whiskers plots were compiled to present the relative quantification of significant metabolites (VIP≥0.5) that were identified.

To assist in data interpretation (section 4.1), colour-coded PCA scores scatterplots were also constructed using the SIMCA-15.0 software to visualise the distribution of selected annotated metabolites in the different sample groups.

Chapter Four: Results and Discussion

4.1. Profiling of hydroxycinnamic acid derivatives in leaves, stems and roots of Bidens pilosa

As described in Chapter 2, B. pilosa contains a large contingency of natural products and has various biological activities attributed to its extracts, fractions and compounds (Silva et al., 2011; Owoyemi and Oladunmoye, 2017). Groups of metabolites that have been identified in B. pilosa include phenylpropanoids, flavonoids, terpenoids, porphyrins and aliphatic natural products. Among these, hydroxycinnamic acid (HCA) derivatives are a conspicuous group of metabolites with diverse biological activities (Ulbrich and Zenk, 1979; Robinson et al., 1996; El-Seedi et al., 2017). In this study, the distribution profile of HCA derivatives in stem, leaf and roots tissue of B. pilosa were studied with the aid of a high-throughput analytical method: UPLC-QTOF-MS. Structural elucidation and putative annotations were aided through the use of an in-source collision-induced dissociation MS-based method (Madala et al., 2014; Ncube et al., 2014). The HCA derivatives were shown to be a prominent group of metabolites in extracts from leaves, stems and roots tissue of B. pilosa as shown in the base peak intensity (BPI) chromatograms (Figure 4.1.1) framed with gold rectangles. Furthermore, the BPI chromatograms indicated differences in intensities of the HCA derivatives as shown in Figure 4.1.1. Within a specific plant species, the distribution of secondary metabolites is expected not to only vary as a function of developmental stage, but also among plant tissues, hence some tissue-specific metabolite variations are observed (Achakzai et al., 2009).

Figure 4.1.1: Representative UHPLC-QTOF-MS base peak intensity (BPI) chromatograms showing the separation of secondary metabolites in extracts of B. pilosa leaves (blue), stems (red) and root tissues. The gold rectangles indicate the chromatographic regions where hydroxycinnamic acid derivatives are present across the three tissues with some visible differences in intensities.

4.1.1. Multivariate statistical analysis of the phytochemical profiles of extracts from different tissues

To analyse the variability within and between the extracts from leaves, stems and roots, principal component analysis (PCA) was performed. PCA is an unsupervised, explorative chemometric tool for the reduction of dimensionality of complex datasets to provide insights into variations and systematic trends among sample groups (Wiklund et al., 2008; Tugizimana et al., 2016). In the current study, the computed model (score plot) of the PCA indicated that 41.1% and 24.3% of the variation was explained by PC1 and PC2, respectively (Figure 4.1.2- A). Statistical validation was described using R2 and Q2, which explain the goodness-of-fit of the model and model predictability, respectively. The model computed was acceptable for metabolomic analysis of the phytochemical data as the R2 > 0.7 and the Q2 > 0.4 (Godzien et al., 2013). For the computed model, R2= 0.890 and Q2= 0.874, respectively, were found to be statistically adequate to make relevant biological interpretations. The PCA model revealed an obvious separation among the three tissue types, as shown in Figure 4.1.2-A. This indicates that the metabolic constituents and their distribution in leaves, stems and roots tissue of B. pilosa varied significantly. This observation could explain the variation in intensities of metabolites observed in the BPI chromatograms shown in Figure 4.1.1. The m/z ions responsible for the clustering observed in the PCA scores scatterplot are indicated with a loading plot in Figure 4.1.2b, with the discriminatory ions furthest from the centre of the loadings plot. To determine correlations/ similarities among the different tissues, a hierarchical cluster analysis (HCA) was also computed, which applies an agglomerative (“bottom-up”) algorithm. Each observation is initially treated as an individual cluster; then after processing, groups are

merged to indicate group similarities (Rodriguez et al., 2019). These were represented by means of a dendrogram shown in Figure 4.1.1.b, which indicates differences in metabolomic profiles respective to the different tissues of B. pilosa. The dendrogram indicates that the metabolomic profiles of stems and roots of B. pilosa are closely related, contrasting to leaf tissues (Granato et al., 2018).

Figure 4.1.2: Principal component analysis (PCA) scores scatterplots, a corresponding loadings plot and hierarchical cluster analysis (HCA) dendrogram. (A) A PCA scores scatterplot of the Pareto-scaled data set obtained from LC-MS experiments. The model obtained was a six-component model (with PC 1 and PC 2 explaining 65.3% of the variation) that indicated the general clustering within the datasets of B. pilosa tissues (leaves (blue), stems (red) and roots (green) samples). The quality parameters of the model were: explained variation/ goodness-of-fit R2=0.890 and the predictive variance Q2=0.874. The ellipse in the PCA score scatterplot indicates the Hotellings T2 at 95% confidence interval. (B) Loading plot that indicates the ions responsible for the clustering observed in the scores scatterplot. (C) Hierarchical cluster analysis (HCA) plot that shows the hierarchical structure of the data in dendrogram format. The model computed shows tissue-specific clustering into two major groups, grouping roots and stems tissues together.

4.1.2. Annotation of metabolites

Mass spectral data were obtained in both positive and negative electrospray ionisation (ESI+/- ) modes. However, negative ionisation was preferred as the majority of metabolites were found to ionize better in this mode. In this study, a total of 30 chlorogenic acid (CGA) derivatives (both regio- and geometric isomers) were identified in B. pilosa plant tissues (listed in Table

4.1 and with the structures shown in Figure 4.1.9). CGAs are classically described as a subclass of phenylpropanoids formed where HCAs are esterified to quinic acid (commonly at positions C3, C4 and C5), while a wider scope of definition describes the CGAs as conjugates of many other compounds such as malic acid, succinic acid, fumaric acid and sugars (Xie et al., 2011; Clifford et al., 2017; Taofiq et al., 2017). Quinic acid has axial hydroxyl groups attached to carbon positions 1 and 3 and equatorial hydroxyl groups attached to carbons 4 and 5, where the HCAs (e.g. caffeic (CFA) -, ferulic (FA) -, р-coumaric (р-CoA) - and sinapic acid (SA)) attach to form CGAs (Clifford, 2000). HCAs are thought to be naturally synthesized with trans configurations, but it has been shown that the cis configuration can occur readily as a result of geometrical changes on the carbon-carbon double bond that can absorb UV radiation (Clifford et al., 2008; Masike et al., 2017; Mhlongo et al., 2015). CGAs are diverse and their regio - and geometrical isomers make it difficult to discriminate between them. Collision-induced dissociation (CID) in tandem mass spectrometry procedures coupled with ultra-high performance liquid chromatography (such as UHPLC-q-TOF-MS) can partially solve this problem and allows for efficient characterization of CGAs by use of fragmentation patterns (Xie et al., 2011; Madala et al., 2014; Liu et al., 2018).

4.1.2.1. Characterisation of feruloylquinic acids (FQAs) and ρ- coumaroylquinic acid (ρCoQA)

A metabolite (4) with a precursor ion, [M-H] -, at m/z 367 was identified based on the fragmentation patterns and retention time (Rt) shown in Table 4.1 and Figure 4.1.3-A. As described in the hierarchical scheme for LC-MSn identification of chlorogenic acids, proposed by Clifford et al. (2003), a base peak ion at m/z 193 [FA-H]- is a diagnostic peaks for 3-FQA, hence metabolite (4) was annotated as 3-FQA. One metabolite was also observed with molecular ion [M-H]-, at m/z 337 which was identified as 5-ρCoQA (2) acid as it produced a sole fragment ion at m/z 191 [QA-H]- showing the loss of a coumaroyl moiety as shown in Figure 4.1.3-B. Both FQAs and ρCoQA were found to be present in all tissues of B. pilosa.

Figure 4.1.3: Representative MS spectra showing the fragmentation pattern of 3-feruloylquinic acid (A) and 5-coumaroylquinic acid (B).

4.1.2.2. Characterisation of caffeoylgycoside

HCAs in nature may occur as soluble forms conjugated to organic acids and/ sugars (Linić et al., 2019). In this study, one caffeoylglycoside (3) was identified which, had a precursor ion, [M-H]-, at m/z 341 as shown in Figure 4.1.4. The molecular ion fragmented to give ions at m/z - - 179 [CFA-H] due to loss of a glycosyl residue and an ion at m/z 135[CFA-CO2] (Ncube et al., 2014). The caffeoylglycoside was found to be present in all the tissues of B. pilosa.

Figure 4.1.4: Representative MS spectrum showing the fragmentation pattern of caffeoylglycoside.

4.1.2.3. Characterisation of mono-caffeoylquinic acids (CQAs)

The hierarchical scheme for LC-MSn identification of mono-CGAs (Clifford et al., 2003) was used to assist in the identification of metabolites (5-7) with a precursor ion, [M-H]- at m/z 353 as listed in Table 4.1. Previous works by Madala et al., (2014); Ncube et al., (2014); Masike et al., (2017) were also used as a reference in putatively identifying these metabolites. Mass spectrometric and chromatographic elution order was also considered while determining the

regio - and geometric isomers of putatively identified metabolites. Notably, cis-isomers of 3- and 4-caffeoylquinic acids elute before their trans-isomers, while the trans-isomer of 5- caffeoylquinic acid elutes before its cis-isomer (Clifford et al., 2008). According to the hierarchical scheme Clifford et al. (2003), 4-CQA can be identified by the presence of an - intense m/z 173 [QA-H-H2O] base peak, while 3-CQA are known to produce base peak ions at m/z 191 [QA-H]- and a secondary ion at m/z 179 [CFA-H]- at a about 50% of the intensity of the m/z 191 base peak. 5-CQA fragmented to produce a sole ion at m/z 191 base peak, indicating the loss of a caffeoyl moeity. Hence metabolites 5-7 were identified as 3-CQA (5), 5-CQA (6) and 4-CQA (7) shown in Figure 4.1.5-A, -B and -C respectively (Madala et al. 2014). These mono-caffeoylquinic acids were observed in almost all B. pilosa tissues except 3-CQA, which was not present in roots.

Figure 4.1.5: Representative MS spectra showing the fragmentation patterns of 3-caffeoylquinic acid (A), 5-caffeoylquinic acid (B) and 4-caffeoylquinic acid (C).

4.1.2.4. Characterisation of di-caffeoylquinic acids (diCQAs)

In this study, six diCQAs shown in Table 4.1 that produced a molecular ion, [M-H]-, at m/z 515 were identified (metabolites 14-19). As shown in the hierarchical scheme for LC-MSn identification of CGAs, 3,5-diCQA (metabolites 16 & 17) can be differentiated from the other - present regio-isomers as no base peak of m/z 173 [QA-H2O-H] was observed. This represents the absence of a 4-acyl substitution as shown in Figure 4.1.5-B. To differentiate between 3,4- di-caffeoylquinic acid (14 & 15) and 4,5-diCQA(18 & 19) shown in Figure 4.1.5-A and - Figure 4.1.5-C, respectively, an intense ion at m/z of 335 [CQA-H2O-H] is noteworthy as it

represents 3,4-diCQA in an MS2 acquisition. Although it may be present in the fragmentation pattern of 4,5-diCQA, the intensity is comparatively lower. On a reversed-phase column the elution order of diCQA regio-isomers is expected to be as follows: 3,4-diCQA, 3,5-diCQA, followed by 4,5-diCQA eluting the latest, validating the annotation of di-CQAs shown in Table 4.1 (Clifford et al., 2005; Masike et al., 2018). Pairs of each di-CQA isomers were observed in leaves of B. pilosa as opposed to stems and roots where only one of the isomers were observed, which could suggest isomerisation due to exposure to UV radiation.

Figure 4.1.6: Representative MS spectra showing the fragmentation pattern of 3,4-di-caffeoylquinic acid (A), 3,5-di-caffeoylquinic acid (B) and 4,5-di-caffeoylquinic acid (C).

4.1.2.5. Characterisation of tri-caffeoylquinic acids (triCQAs) and di- caffeoylquinic acid glycosides

Four isobaric peaks with parent ions, [M-H]- at m/z 677 were observed. Metabolites 27 and 28 were annotated as di-caffeoylquinic acid glycosides as these metabolites fragmented to produce ions at m/z 515 [diCQA-H]- resulting from the loss of a glucosyl residue, m/z 353 [CQA-H]- resulting from loss of a caffeic acid and glucosyl residue, m/z 341 [CQA glycoside-H]- which indicated subsequent loss of the quinic acid residue. Other secondary ions observed were at - - m/z 179 [Caffeic acid-H] , and at m/z 173 [Quinic acid-H2O-H] as shown in Figure 4.1.7-A (Jaiswal et al., 2014).

Metabolites 29 and 30 were annotated as triCQAs as these produced fragment ions at m/z 515 - - - - [diCQA-H ], m/z 353 [CQA-H] , m/z 335 [caffeoylquinic acid-H2O-H] , m/z 191 [QA-H] , m/z - - 179 [CFA-H] and at m/z 173 [QA-H-H20] , also shown in Table 4.1 and Figure 4.1.7-B. The Rt was also considered as triCQAs are expected to elute later than 4,5-diCQA as these are more hydrophobic (Clifford et al., 2007). However, positions of acylation on tri-caffeoylquinic acids/ tri-acylated glycosides were not fully characterized as a description of tri- and tetra- acylation would require MS4 and/or MS5 spectra (Clifford et al., 2017). These tri-acylated HCA derivatives were observed to be only found in the leaves of B. pilosa.

Figure 4.1.7: Representative MS spectra showing the fragmentation patterns of di-caffeoylquinic acid glycoside (A) and tri-caffeoylquinic acid (B).

4.1.2.6. Characterisation of ρ-coumaroyl-caffeoylquinic acids

CGAs are a complex group of compounds that also encompasses hetero di-esters. Previously, the occurrence of ρ-coumaric acid-containing di-acyl-chlorogenic acids has been reported in green coffee beans (Clifford et al., 2006). To the best of our knowledge, these molecules have never been reported in B. pilosa. ρ-Coumaroyl-caffeoylquinic acids were identified by molecular ion, [M-H]-, at m/z 499 (Clifford et al., 2006; Jaiswal et al., 2010). Five of these isomers were observed, metabolites (9-13) in plant tissues of B. pilosa with the first two isomers occurring in pairs as listed in Table 4.1. Considering the elution order of these isomers, the ρ-coumaroyl-caffeoylquinic acids followed an elution order comparable to that of di-CQAs where, on a reversed phase column 3,4-di-esters eluted first, followed by the 3,5-di-esters and the 4,5-di-esters eluting last. The fragmentation spectra pattern generated using 20 eV collision energy level was considered when annotating these metabolites. The first pair was annotated as 3-coumaroyl-4-caffeoylquinic acid (9) and 3-caffeoyl-4-coumaroylquinic acid (10), shown in Table 4.1. 3-Coumaroyl-4-caffeoylquinic acid (9) fragmented to produce a base peak ion at - - - m/z 353 [CQA-H] and secondary ions at m/z 337 [ρCoQA-H] , m/z 335 [CQA-H2O-H] , m/z - - - 191 [QA -H] , m/z 173 [QA-H-H20] and m/z 163 [ρCoA-H] . The presence of a peak at m/z

173 was indicative of acylation at C4 of the quinic acid, the peak at m/z 335 indicated a dehydrated caffeoylquinic acid and its ratio of approximately 30% to the base peak ion was characteristic of a 3,4-CGA. An intense product ion at m/z 163 is characteristic for a coumaric residue at C3 of the quinic acid. The 3-caffeoyl-4-coumaroylquinic acid (9) fragmented to give a base peak m/z 337, which is indicative of a loss of a caffeoyl residue and secondary peak at m/z 173 which indicated that the coumaroyl residue was acylated at C4 of the quinic acid.

As previously mentioned, absence of m/z 173 demonstrates absence of acylation at position C4 of quinic acid. This resulted in the CGA being putatively annotated as a 3,5-di-ester due to its lack of product ion at m/z 173. This metabolite was identified as 3-caffeoyl-5-coumaroylquinic acid (11) and its fragmentation pattern produced a base peak at m/z 353 and secondary ions of m/z 337, 191, 179. The caffeoyl residue was assigned to position C3 as the secondary ions m/z 191 and 179 showed behaviour analogous to that of 3-caffeoylquinic acid where a base peak at m/z 191 is observed and an ion at m/z 179 is present at 50% intensity compared to the m/z 191 (Clifford et al., 2003).

Lastly, the last eluting pairs were annotated as 4,5-di-esters and are indicated in Table 4.1. The isomer that eluted first in this pair was annotated as 4-coumaroyl-5-caffeoylquinic acid (12) and the fragments observed were a base peak at m/z 337 and a secondary ion at m/z 173. Absence of the m/z 353 suggested that the caffeoyl residue was lost more extensively and most likely attached at position C5. According to Clifford et al., (2003) the acylation at position C5 is the easiest to remove followed by that at position C3, whilst the acylation at C4 is the most difficult to remove. The base peak at m/z 337 gave a fragment at m/z 173 suggesting that the coumaroyl residue was attached to position C4. The last isomer was annotated as 4-caffeoyl- 5-coumaroylquinic acid (13). The fragmentation pattern of this metabolite was shown to produce a base peak at m/z 353 and secondary ions at m/z 191, 179 and 173. These isomers were found to be mainly present in the aerial tissues of the plants. * Representative spectra of coumaroyl-caffeoylquinic acids shown under section 4.3 (Figure 4.3.4)

4.1.2.7. Characterisation of feruloyl-caffeoylquinic acids

The feruloyl-caffeoylquinic acids, metabolites (20-26) were identified by their parent ion, [M- H]-, at m/z of 529 (Clifford et al., 2003, 2006; Jaiswal et al., 2010), and chromatographically seven isomers were observed. The first two isomers were identified as 3-feruloyl-4- caffeoylquinic acid (20) and 3-caffeoyl-4-feruloylquinic acid (21). 3-Feruloyl-4-caffeoylquinic (20) was identified by its fragment ions, fragmenting to give ions at m/z 367 [FQA-H]-, m/z - - - - 353 [CQA-H] , m/z 335 [CQA-H2O-H] , m/z 193 [FA acid-H] , 179 [CFA-H] , m/z 173 [QA- - H-H20] and at m/z 134 [FA-H-CO2-CH3]. An intense product ion at m/z 335 indicated that this isomer was a CGA with the caffeoyl residue attached at position C4 of the quinic acid. The presence of an intense peak at m/z 193 indicated that the feruloyl residue was attached at position C3. The next isomer was identified as 3-caffeoyl-4-feruloylquinic acid (21) which

gave a base peak of m/z 367 and secondary ions at m/z 335 and m/z 173, indicating that the feruloyl residue is attached at position C4.

The next two isomers were annotated as 3,5-di-esters as they lacked a fragment ion m/z 173 which indicated no acylation at position C4. These were annotated as 3-feruloyl-5- caffeoylquinic acid (22) and 3-caffeoyl-5-feruloylquinic acid (23) respectively. The first eluting isomer was identified 3-feruloyl-5-cafffeoylquinic acid (22) that produced a base peak at m/z 367 which indicates extensive loss of the caffeoyl residue, suggesting acylation with a caffeoyl residue at position C5. The intense secondary ion m/z 193 indicated that the feruloyl residue was attached to position C3. 3-Caffeoyl-5-feruloylquinic acid (23) was identified by fragment ions at m/z 353, 367, 191 and 179. The caffeoyl residue was assigned to position C3 as ions m/z 191 and 179 showed behaviour similar to that of 3-caffeoylquinic acid where a base peak at m/z 191 is observed and a m/z at 179 is present at 50% intensity compared to the m/z 191 ions and the feruloyl moiety was assigned position C5.

The following isomers were annotated as 4-caffeoyl-5-feruloylquinic acid (24) and 4-feruloyl- 5-caffeoylquinic acid (25). 4-Caffeoyl-5-feruloylquinic acid (24) produced fragment ions at m/z 353, m/z 367, m/z 191 and m/z 173. 4-Feruloyl-5-caffeoylquinic acid (25) was identified by its fragment ions as it produces a base peak m/z 367 and secondary ions of m/z 193 and an intense ion at m/z 173, hence the feruloyl moiety was assigned to position C4. Absence of m/z 353 indicated acylation of the caffeoyl residues at position C5 of QA. Metabolite (26) produced a similar fragmentation pattern to metabolite (23), hence it was annotated as the second isomer of 3-caffeoyl-5-feruloylquinic acid. Feruloyl-caffeoylquinic acids were found mostly to be present in the aerial parts of the plant. * Representative spectra of feruloyl-caffeoylquinic acids shown under section 4.3 (Figure 4.3.5)

4.1.2.8. Characterisation hydroxycinnamoyl-tartaric acid esters

As detailed in Chapter 2, hydroxycinnamoyl-tartaric acid esters such as chicoric acid and caftaric acids, are biologically active compounds which are shown to have various health benefits and antioxidant properties (Healy et al., 2009; Kuban-Jankowska et al., 2016). These HCA derivatives are the main caffeic acid derivatives in E. purpurea but have been also identified in leaves of B. pilosa and in more than 60 plant genera (Lee and Scagel, 2010, 2013; Khoza et al., 2016). Caftaric acid (1, CTA) (Figure 4.1.8-A) was identified by its parent ion, [M-H]-, at m/z 311 which fragmented to produce ions at m/z 179 [CFA-H]- due to the loss of a - - tartaric acid (TA) residue, m/z 149 [TA-H] and m/z 135 [FA-CO2] which resulted from decarboxylation of the caffeic acid residue. Therefore, metabolite (1) was identified as a hydroxycinnamoyl-tartaric acid ester, caftaric acid (Table 4.1). Metabolite (8) (Figure 4.1.8- B) was identified as chicoric acid by its molecular ion [M-H]-, at m/z 473, which fragmented to give product ions at m/z 311 [CTA-H]- due to the loss of a second caffeic acid residue. Other - - daughter ions at m/z 149 [TA-H] and m/z 135 [CFA-CO2] were also observed (Khoza et al.,

2016). Profiling of the tartaric esters revealed that they are only present in the aerial parts of the plants and absent in the roots. This could suggest exclusivity in the biosynthesis of tartaric acid esters, suggesting localised biosynthesis of these esters in the aerial parts of the plant.

Figure 4.1.8B: Representative MS spectrum showing the fragmentation pattern of caftaric acid (A) and chicoric acid (B).

Table 4.1: Characterisation of chlorogenic acids (CGAs) consisting of hydroxycinnamic acid (HCA) derivatives of quinic acid (QA) and tartaric acid from tissues of Bidens pilosa

No. m/z Rt Fragment ions Molecular Metabolite Abbreviation Leaves Stems Roots (min) formulae 1 311.0392 4.5 179, 149, 135 C13H12O19 Caftaric acid CTA   x

2 337.0907 9.45 191 C16H18O8 5-Coumaroylquinic acid 5-ρCoQA   

3 341.0829 7.33 179, 135 C15H18O9 Caffeoylglycoside C-glycoside   

4 367.1273 6.69 193, 134 C17H20O9 3-Feruloylquinic acid 3-FQA   

5 353.0915 3.34 191, 179, 135 C16H18O9 trans-3-Caffeoylquinic acid trans-3-CQA   x

6 353.0769 6.54 191 C16H18O9 trans-5-Caffeoylquinic acid trans-5-CQA   

7 353.0833 7.18 191, 179, 173, 135 C16H18O9 trans-4-Caffeoylquinic acid trans-4-CQA   

8 473.0673 14.89 311, 179, 149, 135 C22H18O9 Chicoric acid CA   x

9 499.1187 16.08 353, 337, 335, 191, 173, 163 C25H24O11 3-Coumaroyl-4-caffeoylquinic acid 3ρCo-4CQA  x x

10 499.104 16.18 353, 191, 179, 173 C25H24O11 3-Caffeoyl-4-coumaroylquinic acid 3C-4ρCoQA  x x

11 499.1313 16.39 353, 337, 191, 179 C25H24O11 3-Caffeoyl-5-coumaroylquinic acid 3C-5ρCoQA  x x

12 499.1310 16.75 337, 173 C25H24O11 4-Coumaroyl-5-caffeoylquinic acid 4ρCo-5CQA  x x

13 499.1222 16.85 353, 191,179, 173 C25H24O11 4-Caffeoyl-5-coumaroylquinic acid 4C-5ρCoQA  x x

14 515.1099 14.32 353, 335, 191, 179, 173, 135 C25H24O12 3,4-di-Caffeoylquinic acid 3,4-diCQA-1   

15 515.1163 14.69 353, 335, 191, 179, 173, 135 C25H24O12 3,4-di-Caffeoylquinic acid 3,4-diCQA-2  x x

16 515.1140 14.96 353, 191, 179, 135 C25H24O12 3,5-di-Caffeoylqiunic acid 3,5-diCQA-1   

17 515.1170 15.03 353, 191, 179, 135 C25H24O12 3,5-di-Caffeoylqiunic acid 3,5-diCQA-2  x x

18 515.1139 15.53 353, 335, 191, 179, 173, 135 C25H24O12 4,5-di-Caffeoylquinic acid 4,5-diCQA-1   

19 515.1122 16.89 353, 191, 179, 173 C25H24O12 4,5-di-Caffeoylquinic acid 4,5-diCQA-2  x x

20 529.1188 16.04 353, 367, 335, 193, 179, 173 C26H26O12 3-Feruloyl-4-caffeoylquinic acid 3F-4CQA   x

21 529.1525 16.21 367, 335, 173 C26H26O12 3-Caffeoyl-4-feruloylquinic acid 3C-4FQA   x

22 529.1403 16.49 367, 193, 134 C26H26O12 3-Feruloyl-5-caffeoylquinic acid 3F-5CQA   x

23 529.1388 16.61 353, 191, 179 C26H26O12 3-Caffeoyl-5-feruloylquinic acid 3C-5FQA-1   x

24 529.1394 16.8 353, 367, 191, 173 C26H26O12 4-Caffeoyl-5-feruloylquinic acid 4C-5FQA   x

25 529.1378 16.97 367, 193, 173 C26H26O12 4-Feruloyl-5-caffeoylquinic acid 4F-5CQA   x   26 529.1395 17.09 353, 191, 179 C26H26O12 3-Caffeoyl-5-feruloylquinic acid 3C-5FQA-2 x 27 677.14 12.99 515, 353, 341, 353, 179, 173 C H O di-Caffeoylquinic acid glycoside diCQA-  31 33 17 x x glycoside-1 28 677.16 14.07 515, 353, 341 C H O di-Caffeoylquinic acid glycoside diCQA-  31 33 17 x x glycoside-2

29 677.16 17.93 515, 353, 191, 173 C34H30O15 tri-Caffeoylquinic acid triCQA-1  x x 30 677.15 18.21 515, 353,179, 173 C34H30O15 tri-Caffeoylquinic acid triCQA-2  x x present, x = absent / not detected.

Figure 4.1.9: Chemical structures of mono-, di- and tri-substituted hydroxycinnamic acid (HCA) derivatives of quinic acid (QA) and tartaric acid (metabolites 1-30) identified in tissues of Bidens pilosa.

4.1.3. Comparison of the tissue-specific distribution and relative abundance of HCA derivatives in Bidens pilosa

The occurrence or abundance of annotated HCA-derivatives in B. pilosa tissues was highlighted using a colour-coded heat-map (Figure 4.1.10), generated from MetaboAnalyst (www.metaboanalyst.ca), which is a comprehensive web-based tool that allows for metabolomic data interpretation and evaluation (Chong et al., 2018). The heatmap was used to further visually explore the dataset and indicate differential metabolite concentration patterns in tissues of B. pilosa and these were generated based on hierarchical clustering (Xia et al., 2015). In the constructed heat map, columns indicated the different plant tissues and the rows indicated the different HCA-derivatives identified and annotated. Significant differences in the distribution of HCAs in different tissues of B. pilosa were observed and are indicated by a colour gradient (dark blue to deep red) with deep red indicating the highest concentration (based on relative peak intensities) of HCA-derivatives in the plant tissues in contrast to dark blue representing the lowest concentrations.

As mentioned in section 4.1. tissue-specific variations in metabolite distributions were anticipated (Achakzai et al., 2009). Most annotated HCA-derivatives (specifically quinic acid esters) were found to be most abundant in stem (indicated by an orange rectangle (Figure 4.1.8)) and leaf tissues (indicated by pink rectangles (Figure 4.1.10). as opposed to the roots. As described in Mondolot et al. (2006) CGAs were found to be distributed mainly in chlorenchyma cells and appeared to be associated with chloroplast and were implicated to confer protection to chloroplast against light. These metabolites where also found to be localised in the vascular bundles and this could suggest that they are transported throughout plant organs. The distribution of CGAs in B. pilosa tissues could suggest that the synthesis of these metabolites could be localised in the aerial parts of the plant and possibly translocated throughout other plant tissues. Furthermore, this could suggest that in B. pilosa, hydroxycinnamoyl-CoA/ quinate hydroxycinnamoyl transferase (HQT) enzyme could be localised in or near the chloroplast, hence accumulation of HCA-derivatives in the aerial parts of the plant. Colour-coded scores scatterplots were also utilised to indicate differential clustering of some quinic acid esters that were abundant in either stems (Figure 4.1.11-A, 4ρCo-5CQA) or leaf tissues of B. pilosa (Figure 4.1.11-A, 3,5-diCQA-2).

In contrast, the tartaric acid esters were only abundant in the aerial parts of the plant and mostly in leaves, whilst absent in root tissues (indicated with yellow rectangles in (Figure 4.1.10). The adjacent dendrogram indicated relationships among variables and samples (Ivanisevic et al., 2015). To indicate the significant differences in the distribution of the tartaric acid and some quinic acid esters in the different tissues, colour coded PCA scores plots (Figure 4.1.11) were constructed. Caftaric acid (Figure 4.1.11-C, CTA) and chicoric acid (Figure 4.1.11-D, CA) were found to be abundant in the aerial parts of B. pilosa whilst relatively low in the roots. This corresponded to the analysis of different tissues of E. purpurea where chicoric acid was found to be present more in the apical parts of that plant (Lin et al., 2011). Absence of the tartaric acid esters (chicoric and caftaric acid) in the root tissues of B. pilosa could suggest a few

possibilities that would require further investigations. Among other effects mentioned in chapter 1, section 2.4. metabolite distribution patterns may differ, due to differences in expression of genes and localisation of enzymes responsible for their biosynthesis (Zhang et al., 2018). The hydroxycinnamoyl-CoA/ tartaric acid hydroxycinnamoyl transferase (HTT), an enzyme responsible for the biosynthesis of the tartaric esters could be localised in the aerial parts of the plant considering the apparent complete absence of these esters in roots (Sullivan, 2014), therefore suggesting its localisation in the chloroplast. However, further proteomics studies are required to validate the localisation of HTT in the chloroplast and measurements of posttranscriptional regulation would need to be conducted to connect the distribution of these esters to localisation of HTT. Deducing from the observations, HCA derivatives were generally abundant in the aerial parts of the plant, this could provide a chemical basis for the distinct usage of different tissues of B. pilosa, to maximally harness its bioactivities.

Figure 4.1.10: Heatmap generated from hierarchical clustering (indicating group similarities) illustrating the occurrence/ distribution of hydroxycinnamic acid derivatives in tissues of B. pilosa. Group averages were used to simplify the visualisation of the distribution of these HCA derivatives. Yellow rectangles indicate the distribution of CTA (caftaric acid) and CA (chicoric acid).

Figure 4.1.11: Colour-coded PCA scores plots indicating the occurrence and higher abundance of quinic acid esters (4ρCo-5CQA (A) and 3,5-diCQA-2 (B)) and tartaric acid esters (caftaric acid (C) and chicoric acid (D)) in the aerial tissues (leaves (L), stems (S)) of B. pilosa and absence/low intensity (indicated with dark blue colour) of these esters in roots (R) of B. pilosa.

4.2. Metabolomic profiling of Bidens pilosa leaves with altered metabolomic states induced by stress-related phytohormones.

In the current study, phytohormones (MeJA and MeSA) were applied onto B. pilosa leaf tissues, with the aim to investigate the effects of these defence mediating molecules on the metabolomes of B. pilosa. MeJA and MeSA are signalling molecules of plant defence that have an effect on the physiological and biochemical processes of the plant that may cause/ enhance the production of various secondary metabolites (Ghanati and Bakhtiarian, 2014). MeSA is a derivative of salicylic acid (SA) produced through the action of methyltransferases on SA (a key compound derived from the shikimic acid pathway) and is a significant constituent in defence signalling cascades related to systemic acquired resistance (SAR) (Dicke and Hilker, 2003). Similarly, MeJA is a volatile phytohormone (derived from jasmonic acid (JA)) which is also a key mediator in the induced systemic resistance (ISR) of a plant (Heuberger et al., 2014). Exogenous treatment of plants with phytohormones functions similarly to abiotic or biotic stressors as they induce defence-related metabolic pathways in plants, thereby altering the secondary metabolism of the plant (Kim et al., 2006). Although MeJA and MeSA have been shown to induce the production of secondary metabolites, such as phenolics in other plants (Bi et al., 2007), the effects of these plant hormonal signalling compounds have not been investigated in B. pilosa. In the current study, a metabolomic investigation was performed on leaf tissues (methanol extracts) of B. pilosa that were treated with 0.5 mM of MeJA and MeSA and harvested at 12 h and 24 h time intervals. Additionally, the comprehensive profiling of metabolites in B. pilosa leaf tissues under exogenous treatment will provide understanding into the interaction and regulation network of the altered metabolomes as regulated by these phytohormones.

The BPI chromatograms (Figure 4.2.1) of leaf tissues treated with MeJA and MeSA and harvested at 12 h (Figure 4.2.1.-A) and 24 h (Figure 4.2.1.-B) time intervals show minor differences compared to the control, with regard to the distribution of secondary metabolites as can be seen from the number of peaks and relative peak intensities. This required the employment of unsupervised exploratory multivariate statistical analysis (section 4.2.1.1) to visualise systematic trends within the data sets and the use of supervised multivariate statistical analysis (section 4.2.1.2) to identify discriminant features differentiating between the treatments with the signalling molecules and the controls.

Figure 4.2.1: UHPLC-qTOF-MS BPI chromatograms of methanol extracts of B. pilosa leaf tissues treated with MeSA and MeJA and harvested at 12 h (A) and 24 h (B) time intervals compared to the controls harvested at the respective times. This indicated that only minor differences were observed throughout the treatments in comparison to control samples.

4.2.1. Multivariate statistical analysis of treated plant leaves

4.2.1.1. Unsupervised multivariate analysis

To explore and visualise the data sets generated by the UHPLC-MS analyses of methanol leaf extracts of plants treated with MeJA and MeSA, unsupervised multivariate statistical analysis was performed. A PCA model (Figure 4.2.2.-A) was constructed which, reduced the dimensions of the data and explained the variance within the data set as mutually orthogonal PCs (Bartel et al., 2013). The model computed was a 13-component model which explained 33.8% of the variation accounted for by PC 1 and PC 2. The model was statistically adequate given the explained variation R2=0.785 and the predictive variance Q2=0.554. Regardless of the minor differences observed in the BPI chromatograms, the PCA scores plot indicated treatment-specific and time-dependent clustering (Worley and Powers, 2015). Furthermore, an HCA plot (Figure 4.2.2.-B) indicated intrinsic similarities and differences within the datasets was computed. As observed in the HCA plot (Figure 4.2.2.-B), two major nodes are observed, indicating differences between the leaf extracts treated and the respective control samples. The first major node (Figure 4.2.2.-B, left-side) indicated that leaves treated with MeJA and harvested at the 24 h time point (MeJA_24hr) are more closely related to control samples harvested at the 12 h time point (Control_24hr). This may indicate that leaves treated with MeJA begin to return to metabolic homeostasis, 24 h post treatment. Similarly, on the second major node some samples treated with MeSA harvested at 24 h (MeSA_24hr) clustered closely to control samples at the 24 h time point. This may also indicates that the leaves treated with MeSA eventually returns to metabolic homeostasis after 24 h. Treatment with MeSA and harvested at different time points (MeSA_12hr and MeSA_24hr) were shown to result in similar metabolomes as groups clustered closer to each other. These were also shown to have similar metabolome composition to leaf extracts treated with MeJA and harvested at 12 h, which could indicate that MeSA and MeJA treatments applied to B. pilosa leaves may have induced similar metabolomic pathways.

In order to annotate and identify the metabolites responsible for the differences observed in the computed PCA and HCA models, further investigations such as the application of supervised multivariate statistical analysis as described below were needed.

Figure 4.2.2: Unsupervised exploratory statistical analysis of B. pilosa leaf tissues treated (pressure infiltrated) with MeJA and MeSA and the non-treated control samples at 12 h and 24 h time points: (A) a PCA scores scatterplot of the Pareto-scaled data set obtained from LC-MS experiments of methanol extracts of treated and non-treated B. pilosa leaf tissues. The computed model was a 13-component model, with PC 1 and PC 2 explaining 33.8% of the variation. The quality parameters of the model were: explained variation/ goodness of fit R2=0.785 and the predictive variance Q2=0.554. The ellipse in the PCA score scatterplot indicates the Hotellings T2 at 95% confidence interval. (B) The hierarchical cluster analysis (HCA) plot shows the hierarchical structure of the data in a dendrogram format, showing treatment-specific clustering.

4.2.1.2. Supervised multivariate analysis

In this study, supervised OPLS-DA models were constructed that use prior class information. In this case, treated samples and control samples were designated as two different sample groups (Figure 4.2.3A). The sample groups (Control_12hr vs. MeSA_12hr) formed a clear separation in the scores space and this indicates the difference in the molecular composition of the samples (control vs. treated) and that these differences are responsible for the group separations observed in Figure 4.2.3A (Tugizimana et al., 2016). The significance of the OPLS-DA model was estimated using a response permutations test (Figure 4.2.3B) where 50 random permutations were performed and R2 and Q2 values were obtained. These values (R2 and Q2) were compared to those of the unperturbed OPLS-DA model. The permutated model resulted in the R2= (0.0, 0238) and Q2= (0.0, -0.450) which is lower than that of the originally computed OPLS-DA model (R2=0.598 and Q2=0.989) indicating that the original model is a statistically viable model. CV-ANOVA was also used to determine the significance of the model. A p-value of lower than 0.05 denotes a good model (Eriksson et al., 2008; Tugizimana et al., 2016) and the p-value of the computed OPLS-DA model was significant as the p-value = 7.92× 10-7. An OPLS-DA loadings S-plot (Figure 4.2.3C) was subsequently constructed to extract statistically significant and potentially important biomarkers positively related to the 0.5 mM MeSA treatment (harvested at the 12 h time point) were found in the upper right quadrant (indicated by yellow block) in contrast to those negatively related to the treatment indicated in the lower left quadrant (indicated by a black box) (Wiklund et al., 2008). Furthermore, variable importance in projection (VIP) scores (Figure 4.2.3D) were used to assess the importance of the biomarkers. Significant biomarkers were considered to have a VIP score of greater than 1.

Figure 4.2.3: An orthogonal projection to latent structures discriminant analysis (OPLS-DA) model computed of control (orange) and 0.5 mM MeSA (purple) treated leaf extracts at the 12 h time point post treatment. (A) A scores plot summarising the relationship between the two conditions. (B) A permutation test plot (n=50) in response to the OPLS-DA plot indicated in A, with the quality parameters indicated on the y-axis intercepts of the figure: R2= (0.0, 0238) and Q2= (0.0, -0.450). (C) A corresponding loadings S-plot, with statistically significant features indicated (either positively or negatively correlated to the treatment as indicated in yellow and black rectangles respectively). (D) A variable importance in projection (VIP) plot for the OPLS-DA model of samples from plants treated with 0.5 mM MeSA. Ions that are responsible for the significant separation observed between the two sample groups are indicated by a VIP score greater than 1.0. The computed OPLS-DA was significant, -7 validated by a p-value= 7.92× 10 .

A similar supervised statistical analysis was performed comparing control_12 h vs. MeJA_12 h (Figure 4.2.4), control_24 h vs. MeSA_24 h, (Figure 4.2.5) and control_24 h vs. MeJA_24 h (Figure 4.2.6) Supervised statistical analysis comparing control_12 h vs. MeJA_12 h (Figure 4.2.4), resulted in a unbiased binary statistically acceptable model (Figure 4.2.4-A) indicated by a R2= 0.991 and Q2= 0.980 compared to the permutated (Figure 4.2.4-B) model which gave R2= (0.0, 0.891) and Q2= (0.0, -0.502). CV-ANOVA indicated that the model was statistically significant shown by a p-value= 7.03 ×10-11. Statistically significant discriminant ions were chosen from the OPLS-DA S-plot as shown in (Figure 4.2.4-C). VIP scores (Figure 4.2.4D) were also used to assess the importance of the biomarkers, with significant biomarkers considered to have VIP score ≥ 1.

Figure 4.2.4: An orthogonal projection to latent structures discriminant analysis (OPLS-DA) model computed of control (orange) and 0.5 mM MeJA (grey) treated leaf extracts at the 12 h time point post treatment. (A) A score plot summarising the relationship between the two conditions. (B) A permutation test plot (n=50) in response to the OPLS-DA plot indicated in A, with the quality parameters indicated on the y-axis intercepts of the figure: R2= (0.0, 0.891) and Q2= (0.0, -0.502). (C) A corresponding loadings S-plot, with statistically significant features indicated in yellow and black. (D) A variable importance in projection (VIP) plot for the OPLS-DA model of samples from plants treated with 0.5 mM MeJA. Ions that are responsible for the significant separation observed between the two sample groups are indicated by a VIP score greater than 1.0. The computed OPLS-DA was -11 significant, validated by a p-value= 7.03 ×10 .

A binary statistical evaluation (Figure 4.2.5-A) was also performed comparing control samples and leaf extracts of B. pilosa treated with MeSA and harvested at after 24 h post treated (Figure 4.2.5). The computed model was statistically acceptable as described R2= 0.995 and Q2= 0.982 compared to the permutated model (Figure 4.2.5-A) which gave R2= (0.0, 0.651) and Q2= (0.0, -0.668). CV-ANOVA was used to cross-validate the model which, resulted in a p-value= 3.21 ×10-11. Similarly, as described discriminant ions positively and negatively correlated to the treatment (MeSA_24hr) were chosen from the OPLS-DA S-plot (Figure 4.2.5-C). A VIP scores plot (Figure 4.2.5-D) was also constructed to further validate significance of the chosen discriminant ions. Metabolites significantly affected by treatment were indicated by a VIP score of greater than 1.

Figure 4.2.5: An orthogonal projection to latent structures discriminant analysis (OPLS-DA) model computed of control (pink) and 0.5 mM MeSA (light blue) treated leaf extracts at the 24 h time point post treatment. (A) A score plot summarising the relationship between the two conditions. (B) A permutation test plot (n=50) in response to the OPLS-DA plot indicated in A, with the quality parameters indicated on the y-axis intercepts of the figure: R2= (0.0, 0.651) and Q2= (0.0, -0.668). (C) A corresponding loadings S-plot, with statistically significant features indicated in yellow and black. (D) A variable importance in projection (VIP) plot for the OPLS-DA model of samples from plants treated with 0.5 mM MeSA. Ions that are responsible for the significant separation observed between the two sample groups are indicated by a VIP score greater than 1.0. The computed OPLS-DA was significant, validated by a p-value= 3.21 ×10-11.

Supervised statistical analysis (Figure 4.2.6) was also performed comparing control samples and leaf extracts of B. pilosa treated with MeJA and harvested after 24 h post treated (Figure 4.2.6). The computed model was unbiased binary statistically acceptable model (Figure 4.2.6- A) indicated by a R2= 0.995 and Q2= 0.983 compared to the permutated (Figure 4.2.6-B) model which gave R2= (0.0, 0.609) and Q2= (0.0, -0.747). Cross validation was done based on CV- ANOVA in a p-value= 3.21 ×10-11. Similarly, as described discriminant ions positively and negatively correlated to the treatment (MeJA_24hr) were chosen from the OPLS-DA S-plot (Figure 4.2.6-C). A VIP scores plot (Figure 4.2.6-D) was used to further validate the significance of selected discriminant ions, described by a VIP score ≥ 1.

Figure 4.2.6: An orthogonal projection to latent structures discriminant analysis (OPLS-DA) model computed of control (pink) and 0.5 mM MeJA (yellow) treated leaf extracts at the 24 h time point post treatment. (A) A score plot summarising the relationship between the two conditions. (B) A permutation test plot (n=50) in response to the OPLS-DA plot indicated in A, with the quality parameters indicated on the y-axis intercepts of the figure: R2= (0.0, 0.609) and Q2= (0.0, -0.747). (C) A corresponding loadings S-plot, with statistically significant features indicated in yellow and black. (D) A variable importance in projection (VIP) plot for the OPLS-DA model of samples from plants treated with 0.5mM MeJA. Ions that are responsible for the significant separation observed between the two sample groups are indicated by a VIP score greater than 1.0. The computed OPLS-DA was significant, validated by a p-value= 2.63 ×10-11.

4.2.2. Comparative analysis of metabolites identified in leaves subjected to treatment with signal molecules

Following supervised multivariate statistical analysis, significant ions (with VIP scores ≥ 1) were putatively annotated based on accurate mass, Rt and MS fragmentation patterns and compared against literature as shown in Table 4.2 (abbreviations listed in the table). The signalling molecules (MeJA and MeSA) used in this study were shown to exert influence on similar metabolomic pathways (originating from the phenylpropanoid pathway, Figure 2.4) as most metabolites annotated were found to be chlorogenic acids or flavonoids, which were differentially induced. These results corresponded to results reported by Mhlongo et al., (2016) where phytohormones were shown to induce some metabolites of the phenylpropanoid pathway in Nicotiana tabacum cells. Previously, MeJA and MeSA have been reported to be important signal molecules in the plant’s systematic response to pathogens (Zhu et al., 2014).

B. pilosa is a plant rich in HCA derivatives of quinic acid and tartaric acid (section 4.1). The HCA derivatives are predominantly produced in plants as they confer protection against abiotic and biotic stressors (Kundu and Vadassery, 2019; Lallemand et al., 2012). Furthermore, the flavonoids also identified as significant ions differentially induced by MeJA and MeSA, are produced in plants and play a role in regulation of plant development and, similarly to HCA derivatives, also play a role in protection of plants against biotic and abiotic stressors (Mathesius, 2018; Treutter, 2005). Elsewhere, flavonoids were also shown to be a prominent group of metabolites in B. pilosa (Gbashi et al., 2017).

A time-course analysis of the significant metabolites (annotated as described above) in treated leaf extracts was also conducted and visualised by means of a VIP scores plot (Figure 4.2.7) derived from PLS-DA analysis performed in MetaboAnalyst, which summarises the weighted sum of squares of the PLS loading, factoring in the explained Y-variance of each component (Xia and Wishart, 2011). Metabolites with a VIP score of greater than 0.5 were considered to be significantly perturbated due to the treatments. Both MeJA and MeSA are known to regulate a number of physiological processes (i.e. plant growth and pant defence responses) in plants. Recently they have been used as elicitors to enhance the production of secondary metabolites in medicinal plants and to induce defence responses in crop plant. Previous studies have demonstrated that both hormones lead to the activation of the phenylpropanoid pathway in various plants and this is associated with enhanced activity of a key enzyme for the biosynthesis of phenolic compounds (PAL). Mhlongo et al., (2016) demonstrated that treatment of tobacco cells with different elicitors (including SA and MeJA) lead to activation of similar pathways but with different metabolic profiles. Among the differential pertubated metabolites were HCA derivatives. In the present study, HCA derivatives were identified as significant ions in response to treatment with either signal molecule. For example, trans-5-caffeoylquinic acid isomers 1 (trans-5-CQA-1) was upregulated in treatment with MeSA and harvested 12 h post- treatment. This contrasts with a previous study by Chang et al., (2019) where SA did not have an effect on the accumulation of phenolics in buckwheat, and results by (Ncube et al., 2012) where the biosynthesis of phenylpropanoids was not induced in Centella asiatica cells when treated with SA. Whilst other HCA derivatives such as cis-5-caffeoylquinic acid (cis-5-CQA) and chicoric acid isomer 1 (CA-1) accumulated in leaves treated with MeJA and harvested after 12 h of treatment. This was expected since in previous studies, MeJA was shown to increase the phenolic acid content (including cinnamic acid derivatives) by upregulating a phenylpropanoid pathway enzyme PAL in Brassica oleracea (Guan et al., 2019).

Flavonoids, also identified as significant ions in response to the treatments with the signal molecules, were also differentially regulated, as observed for tetrahydroxyflavanone triacetylglucoside isomer 1(TFTG-1), where it accumulated in control samples as opposed to treatment with the MeJA, where TFGTG-1 was reduced, but was also shown to reach homeostasis upon treatment with MeSA at both timepoints. This corresponds to treatment of Calendula officinalis, in which treatment with MeJA, which resulted in a decrease in flavonoid content (Ghanati and Bakhtiarian, 2014). In contrast, okanin di-acetylglucoside (O-diAG) was significantly increased in extracts from leaves treated with MeJA and harvested at 24 h, and reduced by the equivalent MeSA treatment. MeJA has been previously shown to be positively

correlated to the accumulation of flavonoids, like observations in B. pilosa where treatment with MeJA resulted in the accumulation of some flavonoids e.g. O-diAG, as stated above (Wang et al., 2015). Accumulation of flavonoids has been shown to be positively correlated with exogenous treatment of tea (Camellia sinensis) leaves with MeSA, where a time-course study showed an increase in flavonoid content from as early as 12 h after treatment. This corresponds to the observation in Figure 4.2.4. where accumulation of quercetin-3-O- glucuronide (Q-3-GA) was observed at the 12 h time interval. MeSA was shown to induce activity of PAL, a key enzyme in the phenylpropanoid pathway (Li et al., 2019). Besides flavonoids and HCAs, a signal molecule, tuberonic acid glucoside (12-hydroxyjasmonic acid glucoside or (TAG)), a glucosylated derivative of tuberonic acid was identified in leaf samples from plants treated with MeSA and harvested at 12 h. This could indicate a cross-talk of JA- and SA-dependent defence signalling pathways in B. pilosa (Koornneef et al., 2008), since TAG is important in the response of plants to wounding (Widemann et al., 2013). Most metabolites identified in leaf extracts treated with signal molecules play vital roles in the systemic response of the plant. Apart from the metabolites mentioned above, treatment with the phytohormones also affected the primary metabolism as citric acid/isocitric acid, a tricarboxylic acid (TCA), was also identified as a significant ion, which was upregulated in leaves, treated with MeJA and harvested after 12 h. This corresponds to a study performed on Brassica nigra where, in response to stress-induced by MeJA, it was found that regulation of the TCA cycle supported energy requirements for biosynthesis of defense molecules (Papazian et al., 2019).

Table 4.2: List of statistically significant and discriminatory ions and putative annotation of metabolites present in methanol leaf extracts of Bidens pilosa treated with 0.5 mM MeSA and MeJA and harvested at 12 and 24 h time points in comparison to the control samples. All ions had a VIP score > 1.0 and were annotated to MSI-level 2.

m/z Rt Fragme Molecular Metabolite Abbreviations Control MeSA MeJA Control MeSA MeJA Class References (min) nt ions formula 12 h 12 h 12 h 24 h 24 h 24 h

191.019 1.01 155, 111 C6H8O7 Citric acid/ Isocitric IsoCA x x  x x x Organic (Al Kadhi et al., acid acid 2017) 311.037 4.64 179, 146 C13H12O9 Caftaric acid isomer CTA-1 x x x  x x HCA (Khoza et al., 1 2016) 353.084 5.57 191 C16H18O9 cis-5- cis-5-CQA  x x  x x HCA (Madala et al., Caffeoylquinic acid 2014; Ncube et al., 2014; Masike et al., 2017) 353.085 6.07 191 C16H18O9 trans-5- trans-5-CQA-1 x  x x x x HCA (Madala et al., Caffeoylquinic acid 2014; Ncube et isomer 1 al., 2014; Masike et al., 2017) 311.074 7.96 179, 149 C13H12O9 Caftaric acid isomer CTA-2  x x  x  HCA (Khoza et al., 2 2016) 353.085 8.07 191 C16H18O9 trans-5- trans-5-CQA-2 x x x x x  HCA (Madala et al., Caffeoylquinic acid 2014; Ncube et isomer 2 al., 2014; Masike et al., 2017) 387.163 8.85 207, 163 C18H28O9 Tuberonic acid TAG x  x x x x Phyto (Marzouk et al., glucoside/ 12- hormon 2018) hydroxyjasmonic e acid glucoside 311.072 8.6 179, 146 C13H12O9 Caftaric acid isomer CTA-3 x x x  x  HCA (Khoza et al., 3 2016) 311.072 9.31 179,149 C13H12O9 Caftaric acid isomer CTA-4  x x  x  HCA (Khoza et al., 4 2016) 491.116 12.69 287, 151 C23H24012 Okanin OAG x x x x  x Flavono Okanin acetylglucoside id acetylglucoside

477.065 13.51 301 C21H18O13 Quercetin-3-O- Q-3-GA x    x x Flavono (Gbashi et al., gluconoride id 2017) 463.086 13.71 301 C21H19O12 Quercetin-3- Q-3-G  x x x x Flavono (Gbashi et al., glycoside id 2017) 515.120 14.10 353,335, C25H24O12 3,4-di- 3,4-diCQA  x x x x x HCA (Madala et al., 191, 173, Caffeoylquinic acid 2014; Masike et 135 al., 2017; Ncube et al., 2014) 447.089 14.49 285 C27H30O15 Kaempferol-3-O- K-3-G  x x  x x Flavono (Gbashi et al., glucoside id 2017) 515.116 14.62 353, 191, C25H24O12 3,5-di- 3,5-diCQA-2 x x x x  x HCA (Madala et al., 179, 135 Caffeoylquinic acid 2014; Masike et isomer 2 al., 2017; Ncube et al., 2014) 473.073 14.95 311, 179, C22H18O12 Chicoric acid CA-1  x x x x x HCA (Khoza et al., 149 isomer 1 2016) 515.117 15.22 353, C25H24O12 4,5-di- 4,5-diCQA x    x x HCA (Clifford et al., 191,179, Caffeoylquinic acid 2003) 173, 135 473.071 15.43 311, 179, C22H18O12 Chicoric acid CA-2 x  x x x x HCA (Khoza et al., 149 2016) 489.086 15.99 285 Kaempferol-3- K-3-AG x x x  x x Flavono (Khoza et al., C23H22O12 acetyl-glycoside id 2016) 533.128 17.6 287, 151, C25H26O13 Okanin di- O-diAG  x   x  Flavono (Khoza et al., 135 acetylglucoside id 2016) 575.139 17.94 287, 135 C27H28O14 Okanin tri- O-triAG x x x  x x Flavono (Khoza et al., acetylglucoside id 2016) 575.138 19.16 285, 135 C27H28O14 Tetrahydroxy- TFTG-1  x  x   Flavono (Khoza et al., flavanone id 2016) triacetylglucoside isomer 1 575.138 19.52 285, 135 C27H28O14 Tetrahydroxy- TFTG-2  x x  x  Flavono (Khoza et al., flavanone, id 2016) triacetylglucoside isomer 2 Indicate presence of ion / analyte, x indicates that ion / analyte is not present

Figure 4.2.7: VIP scores generated from PLS-DA analysis. Indicated are discriminating ions in MeSA and MeJA treated leaves of B. pilosa harvested at 12 h and 24 h time intervals in comparison to the corresponding control samples. Metabolites with a VIP ≥0.5 were considered to be affected by exogenous treatments with MeJA and MeSA. Abbreviations are as listed in Table 4.2.

4.2.3. Relative quantification of selected secondary metabolites identified in leaves treated with MeSA and MeJA

In the current study MeJA and MeSA were applied onto leaves of B. pilosa with an objective to induce production of secondary metabolites and elucidate the pathways affected by the exogenous treatments. As described above (section 4.2.2), in B. pilosa HCA derivatives,

flavonoids, a signal molecule and a primary metabolite were significantly affected by treatment with MeSA and MeJA. Selected metabolites were further relatively quantified based on their normalised concentrations and represented with box-and-whisker plots generated from MetaboAnalyst (Figure 4.2.7). HCA derivatives (Figure 4.2.7) were differentially regulated by treatments with MeSA and MeJA. MeSA was shown to relatively increase the accumulation of trans-5-CQA, while other HCAs (cis-5-CQA, CA and 4,5-diCQAs) accumulated in response to MeJA. Furthermore, exogenous application of MeJA and MeSA application on B. pilosa leaves lead to differential regulation of flavonoids (Figure 4.2.6). For example, in plants treated with MeJA, K-3-G was found to accumulate in higher concentration 12 h post-treatment and followed by a gradual decrease at 24 hrs. On the other hand, MeSA caused a higher accumulation and 12 h post-treatment and slight continuous increase was observed at 24 h post- treatment. In contrast, in MeJA and MeSA treated plants, O-triAG was maintained towards metabolomic homeostasis with, slightly decreasing in plants treated with MeSA at 12 h.

Various studies have reported that plant tissues/ cells treatments of MeJA and MeSA lead to differential regulation of secondary metabolites. Among these, are the phenolic such as HCAs and flavonoids (Yu et al., 2006). Treatment of Lactuca sativa (Kim et al., 2007) and Myrica rubra (chinese bayberry) with 0.1 and 1 mM MeJA lead to an increase of total phenolic content due to the enhanced activity of PAL (Wang et al., 2009). Kim et al., (2006) reported that 0.1 and 0.5 mM MeJA caused a significant increment of the total phenolic content in Ocimum basilicum. In the study rosmarinic, acid and caffeic acid were found to be the strong antioxidants. In tea leaves (Camellia sinensis), MeSA was found to enhance the biosynthesis of flavonoids by stimulating the phenylpropanoid pathway. In that study, time-cause accumulation of flavonoids was observed reaching a peak 2 days post-treatment with 1 mM MeSA and followed by gradual decrease thereafter, to a concentration lower than the control plants after 6 days (Li et al., 2019). Consistent with the time-course of flavonoid concentration, qRT-PCR results showed that MeSA activates PAL, a key enzyme for the biosynthesis of flavonoids as early as 12 h after the treatment, which peaked after 1 day and then gradually declined up to 6 days. Also, MeSA treatment leads to upregulation of genes (such as chalcone synthase, chalcone isomerase, flavanone 3-hydroxylase and dihydroflavonol-4-reductase involved in flavonoid biosynthesis (Li et al., 2019). Moreover, elicitors or plant treatment agents have been found to induce similar pathways, but the metabolic pool has been reported to be more specific to the stimuli. As observed in this study, both agents lead to the activation of similar pathways (Balmer et al., 2015; Mhlongo et al., 2016). The relative quantification performed also indicated a cross-talk between JA and SA signalling pathways as described in section 4.2.2, as observed (Figure 4.2.7) MeJA and MeSA both upregulated TAG at 12 h. IsoCA was shown to be upregulated in MeJA treated plants at 12 h and also slighty increased in MeSA treated plants. This indicated upregulation of a primary metabolite showing energy requirements in plants during biosynthesis of defence molecules as described in section 4.2.2. As observed in this study, the metabolome of B. pilosa is affected by exogenous treatment with signal molecules used as elicitors. Future studies on this plant can allow optimisation of elicitor concentrations, elicitation time and harvest time to permit accumulation of important bioactive secondary metabolites. These studies can be done in plant cell cultures which allow

homogenous distribution of signal molecules applied and scaling up of metabolites subsequently produced.

Figure 4.2.7: Box-and-whisker plots indicating the relative quantification of selected secondary metabolites in B. pilosa leaves treated with MeSA and MeJA compared to the relative controls. MeJA and MeSA were shown to have differential effects on the secondary metabolites and some time- dependent variations were observed.

4.3. Profiling of hydroxycinnamic acid derivatives in B. pilosa cell cultures

Plants are a source of many important natural products, also described as secondary metabolites, that have vast applications such as in the pharmaceutical -, cosmetic - food - and medicinal industries (Figueiredo et al., 2008). The commercial importance of secondary metabolites has generated an interest in finding ways to enhance their production by utilizing tissue/ cell cultures. Plant cells when cultured, display totipotency, and the individual cells possess the full set of genes necessary for all the functions of a plant, including secondary metabolism (Zhao et al., 2005; Jain et al., 2019). Production of secondary metabolites from plant cell cultures has become a popular approach as it offers several advantages (i.e. shorter cultivation time and simple systems for secondary metabolite production) when compared to the conventional plant growing approach (Srivastava et al., 2011). Furthermore, they have been used as a source of secondary metabolites, as they allow synthesis of secondary metabolites independent of geographical and seasonal variation, while also assisting in overcoming extraction of metabolites from limited resources. Additionally, cell cultures also permit controlled homogenous cultivation of commercially and biologically important metabolites (Fischer et al., 1999; Ochoa-Villarreal et al., 2016).

Cell cultures can be established from different explant material, such as leaves, stems and roots, for the production of various secondary metabolites. Callus culture is initiated on semi-solid media from undifferentiated cells and this is then used to subsequently initiate cell suspension in liquid media, with high rates of cell multiplication as indicated in chapter 3, sections 3.1.2 and 3.1.3 (Ramulifho et al., 2019). In this study, the distribution profile of HCA-derivatives in methanol extracts of stem- and leaf tissue-derived cell suspension and callus cultures were investigated by utilising UPLC-QTOF-MS as an analytical platform. A targeted approach was also followed where an optimised in-source collision-induced dissociation (ISCID) method assisted in the structural elucidation and putative annotations of the HCA derivatives identified (Madala et al., 2014; Ncube et al., 2014). Figure 4.3.1 indicates representative UHPLC-QTOF- MS BPI (base peak intensity) chromatograms of B. pilosa cell cultures with HCA-derivatives of quinic acids, indicated by means of yellow rectangles. Similarly to observations in section 4.1, HCA-derivatives were shown to be a prominent group of metabolites in callus (Figure 4.3.1-A) and cell suspensions (Figure 4.3.1-B) of B. pilosa.

Figure 4.3.1: (A) Representative UHPLC-QTOF-MS base peak intensity (BPI) chromatograms showing the separation of secondary metabolite in methanol extracts of callus cultures of B. pilosa initiated from leaves (blue) and stems (red) explant materials. (B) Representative BPI chromatograms indicating metabolites in methanol extracts of cell suspension cultures of B. pilosa initiated from friable callus obtained from leaves and stems as explant materials. Hydroxycinnamic acids derivatives that are present across the cell cultures are indicated with yellow rectangles.

4.3.1. Multivariate statistical analysis of cell culture data

PCA was employed to characterize tissue-specific variation or clustering between cell callus initiated from leaves (C_LEAVES) and stems (C_STEMS) of B. pilosa as shown in Figure 4.3.2-A. This unsupervised explorative chemometric tool reduces the dimensionality of the dataset, highlighting similarities and differences between the two groups (i.e. cell callus and suspensions initiated from stem and leaf tissue in this case) (van der Berg et al., 2006; Tugizimana et al., 2013, 2016). As seen in Figure 4.3.2-A, features from leaf cell callus (blue)

and stem cell callus (red) cluster separately from each other, indicating cell-line-specific metabolite differences between the two metabolomes. This may indicate that the cells, although undifferentiated, maintain a level of metabolite memory related to the original tissue-specific specialized metabolism of the stem and leaf explants. This might be due to differential genetic controls, superimposed on the totipotent nature of the cells in culture (Zhao et al., 2005; Jain et al., 2019). The PCA scores scatter plot (Figure 4.3.2-A) infographically depicts valuable information within the dataset, such as the variation observed. The computed model was a three-component model that explained 69.6% of the variation (Saccenti et al., 2014). Statistical validation was described using R2 and Q2 which describes the goodness-of-fit of the model and provide a measure of model predictability, respectively. The model computed was statistically significant as the R2 and Q2 values were greater than 0.7 and 0.5 respectively (Godzien et al., 2013). A corresponding hierarchical cluster analysis (HCA) was also computed to show similarities between the callus derived from the two tissues, this is represented by means of a dendrogram shown in Figure 4.3.2-B (Granato et al., 2018). In Figure 4.3.2-B, two major nodes are evident, which indicate that the metabolome compositions of sample extracts derived from stem - and leaf calli are different. Some variations within the groups are also observed as minor nodes that might be a reflection of biological variability, or to a certain extent, extraction efficiency and analytical reproducibility.

Similarly, MVDA was performed for cell suspension cultures initiated from stem-derived callus vs. leaf-derived callus. Tissue-specific variations were visualised by means of a PCA scores scatter plot. The PCA scores scatter plot Figure 4.3.2-C of cell suspensions indicated that the features clustered separately and the model constructed explained 72.4% of the variation. This indicated that the cell suspension cultures also retained metabolite memory related to the tissue-specific specialized metabolism. The goodness of fit and the predictive power of the model were explained as R2= 0.785 and Q2= 0.703 respectively. A corresponding HCA dendrogram was constructed to further explore the cell suspension dataset and this indicated tissue-specific differences within suspensions derived from B. pilosa leaf (CS_LEAVES) and stem (CS_STEMS) callus shown in Figure 4.3.2-D.

Figure 4.3.2: (A): Principal component analysis (PCA) scores scatterplots indicating tissue-specific differences within callus cultures of B. pilosa tissue initiated from leaves (blue) and stems (red) explant material. The model obtained was a three-component model and explains 69.87% of the variation. The quality parameters of the model were explained variation/ goodness-of-fit R2= 0.631 and the predictive variance Q2=0.558. The ellipse in the PCA score scatterplot indicates the Hotellings T2 with a 95% confidence interval. (B): A hierarchical cluster analysis (HCA) dendrogram showing that the extracts from stem and leaves callus cluster into different groups. (C): PCA scores plot of cell suspensions (initiated from callus culture of B. pilosa leaves and stems) data set indicating different clustering patterns. The model obtained was a three-component model and explains 72.4% of the variation. The goodness-of-fit and the predictive variance of the model were explained by R2= 0.785 and Q2= 0.703, respectively. The ellipse in the PCA scores scatterplot indicates the Hotelling's T2 with a 95% confidence interval. (D): HCA indicates differences between CS_Leaves and CS_Stems forming the two major nodes.

4.3.2. Annotation of metabolites

Chlorogenic acids (caffeoyl, feruloyl and coumaroyl quinic acid esters) were annotated as described in section 4.1.2., where mass spectral information obtained from the ESI negative ionization mode was preferred, as the majority of these metabolites were found to ionize better in this mode (Clifford et al., 2017). In the cell cultures of B. pilosa, 23 CGA derivatives (both regio- and geometric isomers) were identified for the first time. An UHPLC-qTOF-MS/MS- based in-source collision-induced dissociation method was utilised to generate fragmentation data of these metabolites, and the fragment ions are shown in Table 4.3. Published fragmentation data (Xie et al., 2011; Madala et al., 2014; Liu et al., 2018) and information in databases, such as Dictionary of Natural Products were used in the annotation of the identified features/markers (section 3.5). Fragmentation patterns of these identified markers were also

compared to surrogate standards from other plants analysed with the same analytical conditions (Clifford and Madala, 2017). These metabolites were putatively identified to level 2 of the Metabolomics Standards Initiative (MSI) (Sumner et al., 2007). Structures of the annotated metabolites (numbered as detailed below) are given in Figure 4.3.6.

4.3.2.1. Characterization of the mono-acyl chlorogenic acids (CGAs)

The CGAs were identified as described in section 4.1.2.3, metabolite (1) with a molecular ion [M-H]- at m/z 337, was identified as 5-coumaroylquinic acid (1, ρCoQA) as it produced a fragment ion at m/z 191 [QA-H]-. This metabolite was observed in extracts from suspensions of both cell lines but not in callus of B. pilosa. Metabolites (2-5) with a precursor ion [M-H]- at m/z 353 were annotated as caffeoylquinic acids as listed in Table 4.3. Based on the elution order and fragmentation patterns, metabolites (2-5) were identified as trans-3-CQA (2), cis-5- CQA (3), trans-5-CQA (4) and trans-4-CQA (5). The mono-CGAs were observed in both stem and leaf tissue suspensions of B. pilosa; besides, the trans-5-CQA (4) which was absent is stem suspensions. In cell callus of B. pilosa besides all mono-CGA were present besides cis-5-CQA (3).

A similar approach was followed in the identification two feruloylquinic acid (FQA) isomers, metabolites (6 & 7), which were identified by their precursor ion [M-H] - at m/z 367 and based on their fragmentation patterns and retention times (Rts) reported in Table 4.3. Although differing in intensities, the two FQA regio-isomers were identified in both stem and leaf callus and the cell suspensions derived therefrom. As described in the hierarchical scheme keys for n - - LC-MS identification of CGAs, base peaks at m/z 193 [FA-H] , at m/z 173 [QA-H-H20] and at m/z 191 [QA-H]-, were used as diagnostic peaks for 3-feruloylquinic acid (6) and 4- - feruloylquinic acid (7), respectively. FA also produces m/z 134 [FA-H-CO2-CH3] . Hence molecules 6 & 7 were annotated as 3-FQA (6) and 4-FQA (7), respectively

4.3.2.2. Characterization of di-caffeoylquinic acids (diCQAs) and tri- caffeoylquinic acid (triCQA)

In this study, three diCQAs (Figure 4.3.6), which produced a characteristic pseudo-molecular ion peaks [M-H]- at m/z 515, were identified (14, 15 & 16 - shown in Table 4.3). They were present in cell suspensions and callus originating from both leaves and stem tissues of B. pilosa. The regio-isomers were annotated similarly as described in section 4.1.2.4, based on the hierarchical scheme keys for LC-MSn identification of CGAs proposed by Clifford et al. (2003) and were identified as 3,4-diCQA (14), 3,5-diCQA (15) and 4,5-diCQA (16) as detailed in Table 4.3 (Clifford et al., 2005; Masike et al., 2018). One triCQA (23) was identified by its parent ion [M-H]- at m/z 677 and its fragment patterns. Fragment ions produced at m/z 515 - - - - [diCQA-H ], m/z 353 [CQA-H] , m/z 335 [CQA-H2O-H] m/z 191 [QA-H] , m/z 179 [CFA-H-

- - H20] and at m/z 173 [QA-H-H20] (Table 4.3) were used to putatively identify the tri-CQA. However, positions for acylation on tri-caffeoylquinic acid was not fully characterized, since that would require MS4 and/or MS5 spectra (Clifford et al., 2017), which is impossible to conduct with QTOF-MS utilised herein.

4.3.2.3. Characterisation of ρ-coumaroyl-caffeoylquinic acids

ρ-Coumaroyl-caffeoylquinic acids (Figure 4.3.6) were identified by a parent ion [M-H]- at m/z 499 (Clifford et al., 2006; Jaiswal et al., 2010) and this was done similarly to what is described in section 4.1.2.6. The elution order and fragmentation patterns of these metabolites were considered to assist in their annotation. Six of these isomers were observed in cell suspensions of B. pilosa, but were not detected in callus culture, metabolites (8-13, Figure 4.3.3-A) and they eluted in pairs as shown in Table 4.3, similarly to what was described by Clifford et al. (2003). These isomers were found to follow an elution order similar to that of diCQAs, where 3,4-di-esters eluted first, followed by the 3,5-di-esters and with the 4,5-di-esters eluting last (Clifford et al., 2005; Masike et al., 2018). As previously stated, in-source collision-induced dissociation was performed, and the 20 eVcollision energy level was found sufficient when annotating these metabolites.

Figure 4.3.3: Representative UHPLC-QTOF-MS/MS single ion chromatograms (SIC) showing the separation of coumaroyl-caffeoylquinic acids (A) and feruloyl-caffeoylquinic acids (B).

The first pair was annotated as 3-coumaroyl-4-caffeoylquinic acid (8) and 3-caffeoyl-4- coumaroylquinic acid (9), shown in Figure 4.3.4-A and B, respectively. 3-Coumaroyl-4- caffeoylquinic acid (8) fragments produced a base peak ion at m/z 353 [CQA-H]- and secondary - - - - ions at m/z 337 [pCoQA-H] , 335 [CQA-H2O-H] , m/z 191 [QA-H] , m/z 173 [QA-H-H20] , 163 [pCoQA -H]-. As described in section 4.1.2.6, a fragment ion at m/z 173 indicates acylation at C4 of the QA, which was also substantiated by a dehydrated caffeoylquinic acid (m/z 335), which is characteristic of a 3,4-di-chlorogenic acid. Acylation of the coumaric acid residue at C3 of the QA was indicated by an intense fragment ion at m/z 193. The associated isomer was identified as 3-caffeoyl-4-coumaroylquinic acid (9) (as described in section 4.1.2.6), which fragmented to yield a base peak m/z 337 and a secondary peak at m/z 173.

The next pair of isomers, metabolites (10-11) were identified as 3,5-di-esters due to their lack of a product ion at m/z 173. Figure 4.3.4-C indicates the fragmentation pattern of a metabolite annotated as 3-coumaroyl-5-caffeoylquinic acid (10) with a base peak at m/z 337 indicating that the caffeoyl residue was extensively lost and which suggests that the caffeoyl residue is attached at position C5 of QA. According to Clifford et al., (2003) the acylation at position C5 is the easiest to remove followed by that at position C3, whilst the one at C4 is the most difficult to remove. A product ion at m/z 163 was also observed which indicated that the coumaroyl residue was attached at position C3, hence this metabolite was annotated as 3-coumaroyl-5- caffeoylquinic acid (10). The other isomer of this pair was annotated as 3-caffeoyl-5- coumaroylquinic acid (11) as described in section 4.1.2.6. Its fragmentation pattern is indicated in Figure 4.3.4-D, showing a base peak at m/z 353 and secondary ions of m/z 337, 191 and 179. Also noteworthy, acylation at C5 of the quinic acid with the least polar residue (coumaroyl) resulted in 3-caffeoyl-5-coumaroylquinic acid (11, Rt 15.99 min) eluting later than 3-coumaroyl-5-caffeoylquinic acid (10, Rt 16.01 min) on a reverse column. This is also observed for the next to two 4,5-di-esters. These could suggest that when hetero-acyl CGAs are acylated with the least polar moiety on position C5, they are most likely to elute later

Lastly, the next pair was annotated as 4-coumaroyl-5-caffeoylquinic acid (12) and as 4- caffeoyl-5-coumaroylquinic acid (13) and the fragmentation patterns are indicated in Figure 4.3.4-E and 4.3.4-F respectively. 4-Coumaroyl-5-caffeoylquinic acid (12) fragmented to give a base peak at m/z 337 and secondary ions at m/z 173 and m/z 163, shown in Figure 4.3.4-E. The last isomer was annotated as 4-Caffeoyl-5-coumaroylquinic acid (13) that fragmented to give a base peak at m/z 353 and secondary ions at m/z 337, m/z 191, m/z 179 and m/z 173. These isomers were annotated as described in section 4.1.2.6.

Figure 4.3.4: Typical mass spectra of the fragmentation patterns of 3-coumaroyl-4-caffeoylquinic acid (A), 3-caffeoyl-4-coumaroylquinic acid (B), 3-coumaroyl-5-caffeoylquinic acid (C), 3-caffeoyl-5- coumaroylquinic acid (D), 4-coumaroyl-5-caffeoylquinic acid (E) and 4-caffeoyl-5-coumaroylquinic acid (F).

4.3.2.4. Characterisation of feruloyl-caffeoylquinic acid

The feruloyl-caffeoylquinic acids in cell cultures (callus and suspensions) of B. pilosa metabolites (17-22), were identified by their pseudo-molecular ion peak at m/z of 529. The chemical structures of the identified compounds are presented in Figure 4.3.6 (Clifford et al., 2003, 2006; Jaiswal et al., 2010). Chromatographically, six isomers were observed, which eluted in pairs as shown in Figure 4.3.3-B that were annotated similarly to as described as section 4.1.2.7. The first two isomers were identified as 3-feruloyl-4-caffeoylquinic acid (17) and 3-caffeoyl-4-feruloylquinic acid (18). A typical fragmentation pattern of 3-feruloyl-4- caffeoylquinic (17) shown in Figure 4.3.5-A indicated fragment ions at m/z 367 [FQA-H]-, m/z - - - - 353 [CQA-H] , 335 m/z [CQA-H2O-H] , m/z 193 [FA-H] , 179 [CFA-H-H20] , 173 [QA-H- - H20] and at m/z 134 [FA-H-CO2-CH3]. Figure 4.3.5-B shows the fragmentation of 3-caffeoyl- 4-feruloylquinic acid (18), which gave a base peak of m/z 367 and secondary ion at m/z 173, indicating that the feruloyl residue is attached at position C4 of QA.

The next eluting isomers were annotated as 3-feruloyl-5-caffeoylquinic acid (19) and 3- caffeoyl-5-feruloylquinic acid (20), which are 3,5-di-esters, indicated by lack of m/z 173. The fragmentation pattern of 3-feruloyl-5-caffeoylquinic acid (19 – Figure 4.3.5-C), shows a base peak at m/z 367 that indicates extensive loss of the caffeoyl residue. This suggests acylation with a caffeoyl residue at position C5, while the intense secondary ion at m/z 193 indicates that the feruloyl residue was attached to position 3. Figure 4.3.5-D shows the fragmentation pattern (annotated as explained in section 4.1.2.7) of 3-caffeoyl-5-feruloylquinic acid (20) and fragments observed were m/z 353, 367, 191 and 179.

The last two isomers were annotated as 4-feruloyl-5-caffeoylquinic acid (21) and 4-caffeoyl- 5-feruloylquinic acid (22). The first eluting isomer (4-feruloyl-5-caffeoylquinic acid (21)) showed a base peak m/z 367, secondary ions m/z 193 and an intense m/z 173 ion as shown in Figure 4.3.5-E, hence the feruloyl residue was attached at position C4 of QA. The absence of m/z 353 indicated acylation of the caffeoyl residue at position C5 of QA. The fragmentation pattern of 4-caffeoyl-5-feruloylquinic acid (22) is shown in Figure 4.3.5-F and indicated fragment ions at m/z 353, m/z 367, m/z 191 and m/z 173.

Figure 4.3.5: Typical mass spectra of the fragmentation patterns of 3-feruloyl-4-caffeoylquinic acid (A), 3-caffeoyl-4-feruloylquinic acid (B), 3-feruloyl-5-caffeoylquinic acid (C), 3-caffeoyl-5- feruloylquinic acid (D), 4-feruloyl-5-caffeoylquinic acid (E) and 4-caffeoyl-5-feruloylquinic acid (F).

Table 4.3: Characterisation of chlorogenic acids (CGAs) consisting of hydroxycinnamic acid (HCA) derivatives of quinic acid (QA) from two callus cell lines (C_L and C_S) and two cell suspension lines (CS_L and CS_S) of Bidens pilosa.

No. m/z Rt Fragment ions Molecular Metabolite Abbreviation C_L C_S CS_L CS_S (min) formula

1 337.0822 9.45 191 C16H18O8 5-Coumaroylquinic acid 5-ρCoQA x x  

2 353.084 3.2 191, 179, 135 C16H18O9 trans-3-Caffeoylquinic acid trans-3-CQA    

3 353.0821 6.27 191 C16H18O9 cis-5-Caffeoylquinic acid cis-5-CQA x x  

4 353.0884 6.31 191 C16H18O9 trans-5-Caffeoylquinic acid trans-5-CQA    x

5 353.0835 6.89 191, 179, 173, 135 C16H18O9 trans-4-Caffeoylquinic acid trans-4-CQA    

6 367.1003 6.52 193 C17H20O9 3-Feruloylqunic acid 3-FQA    

7 367.0986 10.91 191, 173 C17H20O9 4-Feruloylquinic acid 4-FQA    

8 499.1211 15.68 353, 337, 335, 191, 173, 163 C25H24O11 3-Coumaroyl-4-caffeoylquinic acid 3ρCo-4CQA x x  

9 499.1183 15.84 337, 335, 173, 164 C25H24O11 3-Caffeoyl-4-coumaroylquinic acid 3C-4ρCoQA x x  

10 499.1217 15.99 337, 163 C25H24O11 3-Coumaroyl-5-caffeoylquinic acid 3ρCo-5CQA x x  

11 499.12312 16.09 353, 337,191, 179 C25H24O11 3-Caffeoyl-5-coumaroylquinic acid 3C-5ρCoQA x x  

12 499.1352 16.65 337, 173, 163 C25H24O11 4-Coumaroyl-5-caffeoylquinic acid 4ρCo-5CQA x x  

13 499.1227 16.75 353, 337,191, 179, 173 C25H24O11 4-Caffeoyl-5-coumaroylquinic acid 4C-5ρCoQA x x  

14 515.1182 14.65 353, 335, 191, 179, 135 C25H24O12 3,4-di-Caffeoylquinic acid 3,4-diCQA    

15 515.121 14.93 353, 191, 179, 135 C25H24O12 3,5-di-Caffeoylquinic acid 3,5-diCQA    

16 515.1292 15.67 353, 335, 191, 179, 173, 135 C25H24O12 4,5-di-Caffeoylquinic acid 4,5-diCQA    

17 529.1315 15.92 367, 353, 335, 193, 179, 173, C26H26O12 3-Feruloyl-4-caffeoylquinic acid 3F-4CQA     134

18 529.1381 16.1 367, 335, 193, 173 C26H26O12 3-Caffeoyl-4-feruloylquinic acid 3C-4FQA    

19 529.1296 16.37 367, 193, 134 C26H26O12 3-Feruloyl-5-cafffeoylquinic acid 3F-5CQA    

20 529.1422 16.49 367, 353, 191, 179 C26H26O12 3-Caffeoyl-5-feruloylquinic acid 3C-5FQA    

21 529.1463 16.86 367, 193, 173 C26H26O12 4-Feruloyl-5-caffeoylquinic acid 4F-5CQA    

22 529.142 16.97 367, 353, 191, 179, 173, 135 C26H26O12 4-Caffeoyl-5-feruloylquinic acid 4C-5FQA    

23 677.1436 17.76 515, 353, 335, 191, 179, 173 C34H30O15 tri-Caffeoylquinic acid triCQA x x  x present, x = absent / not detected.

Figure 4.3.6: Chemical structures of mono-, di- and tri-substituted hydroxycinnamic acid (HCA) derivatives of quinic acid (QA) identified in cell cultures of Bidens pilosa.

4.3.3. Comparison of distribution and relative abundance of HCAs in cell cultures

Plant cell cultures have been shown to be an attractive approach for the controlled production of bioactive natural products such as phenylpropanoids, compared to the use of wild plants (Rao and Ravishankar, 2002). In cell suspensions of B. pilosa, 23 HCA-derivatives were identified, while only 14 of these were identified in the callus cultures as presented in Table 4.3. Although plant cell culture is a promising alternative for metabolite production and provides numerous advantages as mentioned in section 4.3., a significant limitation of using cell culture is that undifferentiated cells may accumulate secondary metabolites to a lesser extent compared to the parent plant (Qu et al., 2011; Nomura et al., 2018). As observed in this study, fewer HCA-derivatives were accumulated in cell cultures compared to plant tissues where 30 HCA-derivatives were identified (Table 4.1). However, various strategies can be employed to improve yield of secondary metabolites in plant cell cultures such as optimising cultural conditions (medium modifications) (AbouZid, 2014) precursor feeding (Hussain et al., 2012), immobilisation techniques (Rao and Ravishankar, 2002), elicitation (Wang et al., 2015) and screening for high-producing cell lines (Ochoa-Villarreal, 2016). As mentioned above, fewer HCA-derivatives were identified in cultured callus cells compared to cell suspensions. This could be because callus is maintained on semi-solid media, while cell suspensions are maintained in liquid medium, which is agitated to enhance oxygenation, promote better growth and transfer of nutrients (Murthy et al., 2014; Murthy et al., 2014; Jamil et al., 2018). Although cells in suspension cultures are undifferentiated, they may have reached genetic stability compared to callus cultures, considering the selection associated with multiple sub-culturing steps, thus accounting for better metabolite production in cell suspensions (Bourgaud et al., 2001).

Cell suspensions of B. pilosa were shown to be better starting material for in vitro cultivation for production of HCA-derivatives such as CGAs. Cultured cells of B. pilosa were noted to possess some form of a genetic memory and totipotency for the biosynthesis of some CGAs, similar to that of the parent plant. This observation may in future provide possibilities for bioreactor-based large-scale production of these biologically important secondary metabolites. Cultured cells are comparable to undifferentiated meristematic cells and lack chloroplasts. This could explain the apparent lack of accumulation of tartaric acid esters in B. pilosa cultures (Efferth, 2019). These esters were hypothesised (section 4.1.3) to be exclusively biosynthesised by the enzyme HTT, which possibly could be localised in the chloroplasts of B. pilosa. Although making photosynthetic cell cultures with functional chloroplasts has been deemed difficult and time-consuming, recent advances have been made where medium modification in Arabidopsis thaliana cultures resulted in chloroplast formation (Sello et al., 2017).

Following the identification of CGAs in the cell cultures of B. pilosa, the relative abundance of these metabolites based on their peak intensities were evaluated and compared among cell cultures derived from leaves and stem tissues of B. pilosa. Heat-maps were generated based on a hierarchical (agglomerative) clustering, indicating the individual values in the matrix as a

color gradient (dark blue-deep red) (Xia et al., 2009). The color-coded matrix and adjacent dendrograms were used to indicate relationships among variables and samples (Ivanisevic et al., 2015) as shown in Figure 4.3.7.

The relative abundance of HCA-derivatives identified in callus derived from stems and leaves of B. pilosa were compared. Observations drawn from the heatmap (Figure 4.3.7-A) indicate that a greater number of HCA-derivatives were found to be more abundant in callus derived from stem tissue (indicated by orange outline) in comparison to callus derived from leaf tissue. Even though most HCAs were found to be abundant in stem callus, some were also found to be abundant in leaf callus shown by a yellow outline in Figure 4.3.7-A. This indicated tissue- specific differences among the callus cultures, which indicate genetic memory in these undifferentiated cells. Tissue-specific differences were also observed for the plant tissues as indicated in section 4.1.3.

The relative abundance of the HCA-derivatives was also investigated in cell suspension cultures of B. pilosa represented by a color-coded heatmap shown in Figure 4.3.7-B. In contrast to the observations in Figure 4.3.7-A, HCA-derivatives were found to be more abundant in cell suspensions derived from leaf tissues (indicated by a yellow outline). The orange box in Figure 4.3.7-B indicated some HCA-derivatives that were abundant in cell suspensions derived from stem tissues. The cell suspension cultures also indicated some tissue- specific differences in distribution of these secondary metabolites, also indicating genetic memory of the undifferentiated cells.

Figure 4.3.7: Colour-coded heat-maps illustrating the relative occurrence/ distribution of hydroxycinnamic acid derivatives in callus (A) and cell suspensions (B) of B. pilosa. Group averages were used to simplify the visualisation of the distribution of these HCA-derivatives and a hierarchical cluster was used to indicate group similarities.

4.4. Manipulation of undifferentiated Bidens pilosa cells with plant growth regulators (PGRs)

As described in section 4.3.3, the use of plant cell culture has gained interest in the production of various bioactive plant secondary metabolites. Plant growth regulators (PGRs) such as auxins and cytokinins, are important plant hormones that act in controlling developmental processes of the plant such as forming meristems (Moubayidin et al., 2009). The meristems, containing a small group of pluripotent stem cells are responsible for formation of all tissues of a plant (Su et al., 2011; Lee et al., 2019). These hormones also act synergistically in controlling cell division in undifferentiated cells (Coenen and Lomax, 1997). In plant tissue culture, cells can be manipulated to regenerate the plant tissues from somatic differentiated cells under favourable conditions (organogenesis). Generally, an interaction between auxin and

cytokinins results in organogenesis as a low auxin: cytokinin ratio promotes shoot induction (caulogenesis), while high auxin: cytokinin ratio promotes root formation (rhizogenesis) (Pernisova et al., 2009; Su et al., 2011; Schaller et al., 2015; Hamany Djande et al., 2019). The mode of crosstalk between these PGRs varies with plant species and organs being studied (Moubayidin et al., 2009). Consequently, in this study the effects of different ratios of PGRs were investigated on B. pilosa callus derived from stems and leaf tissues in order to evaluate how PGR-mediated manipulation of the undifferentiated callus cells will affect the phytochemical profiles thereof.

In vitro organogenesis can be described as either direct or indirect. In the latter, PGRs stimulate totipotent cells of callus for organogenesis. (Malik et al., 2007). Organogenesis allows for control of plant development and production of specific tissues in vitro for secondary metabolite biosynthesis (Cardoso et al.,2019). B. pilosa stems and leaves explant materials were used to initiate callus on media with 0.45 mg/L 2,4-D (auxin) and 1.0 mg/L BAP (cytokinin). Calli that formed under these conditions were cut from the original explants and transferred to fresh medium of the same composition. Calli were cultured until cell growth stabilised post multiple subsequent sub-culturing steps (Figure 4.4.1.-A). The white friable callus was sub-cultured onto solid medium with different combinations of auxins and cytokinins (2,4-D: BAP) as shown in Table 3.1 to stimulate the undifferentiated cells towards root – or shoot organogenesis.

B. pilosa leaf callus was shown to grow well into friable white callus (Figure 4.1.1.-B1) in response to low auxin and high cytokinin concentration (0.2 mg/L 2,4-D and 2.0 mg/L BAP), respectively). Similarly, leaf callus grown on PGR ratios with very high cytokinin concentrations (0.3 mg/L 2,4-D: 4.0 mg/L BAP) (Figure 4.1.1.-B5) and (0.2 mg/L 2,4-D: 8.0 mg/L BAP) (Figure 4.1.1.-B6) still maintained growth with minor browning. Furthermore, stem callus maintained in media with low auxin and high cytokinin, also grew well as seen with leaf callus (Figure 4.1.1.-C1, C5, C6). In contrast, leaf callus maintained at high auxin and low cytokinin concentration (2.0 mg/L 2,4-D: 0.2 mg/L BAP) (Figure 4.1.1.-B2) showed comparably reduced growth and higher levels of callus browning, whilst stem callus grew considerably better under the same conditions (Figure 4.1.1.-C2). Similarly, callus browning and reduced growth (Figure 4.1.1.-B3) was also observed in callus maintained in the initiation media (0.45 mg/L 2,4-D:1.0 mg/L BAP), contrasting with stem callus that maintained growth and was not oxidised (Figure 4.1.1.-C3). Callus browning/ oxidation has been attributed to many factors that are correlated to an increase in phenolic content. The increase of phenolic content is relative to an increase of the activity of PAL, which converts phenylalanine to trans- cinnamic acid, subsequently leading to biosynthesis of cinnamates derivatives, which have been associated with browning (Alagarsamy et al., 2018). This results in elevated activity of enzymes such as polyphenol oxidase (Murata et al., 2014) and peroxidase (Wu and Lin, 2002), which convert phenolics to molecules that have a deleterious effect on callus cultures. Browning has also been correlated to carbohydrate metabolism in cell cultures. Callus browning has been described as a major problem that inhibits shoot formation and long-term maintenance of callus as observed in this study (Chaudhary and Dantu, 2015). Partial callus habituation was observed (Figure 4.4.1.-B2), where leaf callus grown on medium, without

added PGRs still maintained growth. Habituation is an occurrence in which division and growth of cells in culture become independent of added PGRs (Pischke et al., 2006; Hamany Djande et al., 2019). This was, however, not observed on stem callus maintained on PGR free media as reduced callus growth was observed (Figure 4.4.1.-C2).

In this study, B. pilosa cell culture did not show signs of possible shoot and root regeneration under the conditions investigated. The different media conditions resulted in either callus browning or white friable cells. Various factors such as multiple sub-culturing are known to result in loss of the totipotency of cells in plant culture as cells. Epigenetic changes can occur in cultured cells where genes responsible for organogenesis are modified (hypo- and hypermethylated). Other factors contributing to loss of totipotency and loss of regenerative abilities include genetic variation and selection pressure, resulting in loss of cells with genetic memory for totipotency (Gaspar et al., 2000). The age of plants used also affects the regenerative capacity of ex-plants, such that explants from younger plants have been found to possess better regenerative capacity compared to matured plants (Zhang et al., 2015). In future, de novo organogenesis of B. pilosa could be studied at different developmental stages of the plant. Culture conditions also play a role in regeneration of the plant, such as media nutrient content, pH, light and temperature. In consideration, further optimisation of media content and maintenance conditions of B. pilosa callus can be investigated to achieve shoot and root regeneration (Ikeuchi et al., 2016).

Figure 4.4.1: Representative images of B. pilosa friable stem - and leaf callus initiated on 0.45 mg/L 2,4-D and 1.0 mg/L BAP (A), which was sub-cultured onto media with different combinations of concentrations/ ratios of 2,4-D and BAP (mg/L). (B) indicates leaf callus of B. pilosa maintained on media containing PGRs (2,4-D to BAP) at the following ratios [1:10] (1), [10:1] (2), [0:0] (3), [1:2] (4), [1:20] (5) and [1:40] (6). (C) indicates B. pilosa stem callus also maintained on media with different 2,4-D: BAP ratios as described for the leaf callus. The concentrations were as follows: (1) 0.2 mg/L 2,4-D and 2.0 mg/L BAP, (2) 2.0 mg/L 2,4-D and 0.2 mg/L BAP, (3) 0 mg/L 2,4-D and 0 mg/L BAP (4) 0.45 mg/L 2,4-D and 1.0 mg/L BAP, (5) 0.3 mg/L 2,4-D and 4.0 mg/L BAP and (6) 0.2 mg/L 2,4-D and 8.0 mg/L BAP.

4.4.1. Analysis of altered callus metabolomes in response to different plant growth regulator combinations.

The manipulation of PGRs ratios in plant cell cultures does not only affect growth and developmental processes, but also regulates different pathways of secondary metabolism such as the phenylpropanoid pathway (Luczkiewicz et al., 2014). In plant cell culture, a variety of secondary metabolites accumulates differentially depending on the concentrations of PGRs (Lee et al., 2011). In this study, the effects of PGRs (2,4-D and BAP) on the metabolomes of B. pilosa callus were investigated. The following six different conditions of 2,4-D to BAP were investigated, condition 1 [1:10], condition 2 [10:1], condition 3 [0:0] condition 4 [1:2], condition 5 [1:20] and condition 6 [1:40]. The actual concentrations involved are reported in the legend to Figure 4.4.1. Changes in the metabolome of B. pilosa callus were studied with a high-throughput analytical method: UHPLC-QTOF-MS. Through visual inspection of BPI chromatograms (Figure 4.4.2.-A and Figure 4.4.2.-B) HCA derivatives were a notable group of metabolites in B. pilosa leaf - and stem callus as observed previously in section 4.3. The HCA derivatives are shown to have minor intensity differences throughout callus grown on

media with various ratios of PGRs as shown in gold rectangles in Figure 4.4.2. Although the combined effects of auxins and cytokinins are not fully understood, addition of PGRs in culture has been shown to generally have positive effects on the accumulation of secondary metabolites, including phenolics (Jamwal et al., 2018). To further visualise the systematic trends in response to the different PGR ratios, the dataset were subjected to multivariate statistical analysis (section 4.4.1).

Figure 4.4.2: Representative UHPLC-QTOF-MS base peak intensity (BPI) chromatograms showing the separation of secondary metabolites in methanol extracts of callus cultures of B. pilosa leaves (A) and stems (B) maintained on different ratios of 2,4-D to BAP ([1:10], [10:1], [0:0], [1:2], [1:20] and [1:40]). Hydroxycinnamic acid derivatives were found to be a prominent group of metabolites in these cultures, albeit with differential intensities.

4.4.2. Multivariate statistical analysis of phytochemical profiles/constituents of callus maintained on different plant growth regulator combinations.

Unsupervised multivariate statistical analysis was employed to explore datasets generated by the UHPLC-qTOF-MS analysis of methanol extracts of B. pilosa stem - and leaf callus maintained on media with different ratios of 2,4-D to BAP ([1:10], [10:1], [0:0], [1:2], [1:20] and [1:40]). To analyse the variability within and between the data sets, PCA score scatterplots models were constructed for leaf callus (Figure 4.2.3.-A) and stem callus (Figure 4.2.3.-C), which reduced the dimensionality of the data sets (Wiklund et al., 2008; Bartel et al., 2013; Tugizimana et al., 2016). The PCA scores scatterplot (Figure 4.2.3.-A) computed for leaf callus was an 11-component model of which PC1 and PC2 explained 30.6% and 14.5% of the variation within the dataset. Visually, the PCA scores scatterplot indicated differential metabolic profiles within the dataset, as condition-specific clustering was observed. Leaf callus maintained on media with higher cytokinin ([1:10], [1:20] and [1:40] 2,4-D to BAP] were found to be more closely related. The statistical validation of the model was described by the explained variation/ goodness of fit R2=0.876 and the predictive variance Q2=0.704. The statistical validation performed indicated that the model computed was fit, as acceptable models for biological data are described by R2>0.7 and Q2>0.4 (Godzien et al., 2013).

The PCA scores scatterplot (Figure 4.2.3.-C) computed for stem callus maintained on media with different combination ratios of PGRs indicated differential clustering of sample groups. Visually stem callus grown on media with [1:2], [1:40] and [10:1] 2,4-D to BAP clustered together and separate from the other conditions. This model was also an 11-component model, where PC1 and PC2 described 30.4% and 19.2% of the variation within the dataset, respectively. The model was found to be adequate to draw relevant biological interpretation described by R2=0.9 and the predictive variance Q2=0.78. Interestingly, for both stem - and leaf callus maintained on medium with no PGRs ([0:0] 2,4-D to BAP) a grouping separate from sample groups grown on media with PGRs was observed. This could indicate that callus grown on media without PGRs (i.e. habituated callus) is significantly different from that grown-on media with PGRs.

Hierarchical cluster analysis was also computed for both stem - and leaf callus, which applies an agglomerative (‘bottom-up’) algorithm to determine correlation/similarities within callus grown on media with different combination ratios of PGRs (Granato et al., 2018; Rodriguez et al., 2019). HCA dendrograms were generated from the datasets and these indicated differences in the metabolomic profiles of callus grown on different combination ratios of PGRS for leaf callus (Figure 4.2.3-B) and stem callus (Figure 4.2.3-D). The HCA dendrogram (Figure 4.2.3- B) indicated that leaf callus maintained on media with high cytokinin to auxin ratios, [1:10], [1:20] and [1:40] 2,4-D to BAP, was more closely related compared to media grown on [0:0], [1:2] and [10:1] 2,4-D to BAP, which clustered together. In stem callus culture, the dendrogram (Figure 4.2.3-D) indicated that callus maintained on a high cytokinin to auxin ratio ([1:10] and [1:20] 2,4-D to BAP) were more closely related when compared to callus grown on [0:0], [1:2],

[1:40] and [10:1] 2,4-D to BAP. PGRs were also shown to induce differential responses from leaf- and stem-derived callus. In leaf-derived callus, a ratio with very high cytokinin [1:40] 2,4- D to BAP, induced a similar metabolic response as induced by other ratios with high cytokinins ([1:10] and [1:20] 2,4D to BAP). In contrast, stem-derived callus grown on media with very high cytokinin concentratration was shown to have similar metabolic responsiveness as callus grown on media with high auxin concentration ([10:1], 2,4-D to BAP) and moderate cytokinin concentration [1:2] 2,4-D to BAP. This could indicate some tissue dependent differential responsiveness to PGRs in media. Further metabolomic differences were assessed post metabolite identification through relative quantification of the annotated metabolites from the different growth conditions.

Figure 4.4.3: Unsupervised exploratory statistical analysis of B. pilosa leaf - and stem callus maintained on media with different combination ratios of 2,4-D to BAP ([1:10], [10:1], [0:0], [1:2], [1:20] and [1:40]). (A) A PCA scores scatterplot of the Pareto-scaled data set of leaf callus. The computed model was a 11-component model, with PC 1 and PC 2 explaining 45.1% of the variation. The quality parameters of the model were: explained variation/ goodness of fit R2=0.876 and the predictive variance Q2=0.704. The ellipse in the PCA score scatterplot indicates the Hotellings T2 at 95% confidence interval. (B) The HCA plot shows the hierarchical structure of the leaf callus data in a dendrogram format, showing PGR concentration / ratio-dependent clustering. (C) A PCA score plot scores scatterplot of the Pareto-scaled data set of stem callus. The computed model was a 11-component model, with PC 1 and PC 2 explaining 49.6% of the variation. The quality parameters of the model were: R2=0.9 and Q2=0.78. (D) The HCA plot shows the hierarchical structure of the stem callus data in a dendrogram format, showing PGRs concentration ratio dependent clustering.

4.4.3. Comparative analysis of metabolites identified in callus maintained on media with different PGR ratios

Post the multivariate statistical analysis, metabolites in B. pilosa callus grown on solid agar media with different ratios of PGRs, were putatively annotated to MSI-level 2 (Table 4.4) (Sumner et al., 2007). These were annotated based on accurate mass, Rt and MS fragmentation patterns and compared against literature as described in section 4.1.2. In this study, metabolites identified from methanol extracts of the callus grown under the mentioned PGRs ratios, were found to be hydroxycinnamic acid derivatives. As previously mentioned, HCA derivatives are biologically important metabolites which have been shown to be abundant in tissues (section 4.1) and cell cultures (section 4.2) of B. pilosa. A total of 14 HCA derivatives were identified in these callus cultures.

Table 4.4: Characterisation of hydroxycinnamic acids (HCAs) and caffeoyl/feruloyl quinic acids derivatives (C/FQAs) present in methanol extracts of Bidens pilosa stem- and leaf-derived callus maintained on media with different combination ratios of 2,4-D to BAP*.

No. m/z Rt Fragment ions Molecular Metabolite Abbreviation (min) formulae

1 353.0842 2.77 191, 179, 135 C16H18O9 trans-3-Caffeoylquinic acid trans-3-CQA

2 353.0881 5.48 191 C16H18O9 trans-5-Caffeoylquinic acid trans-5-CQA

3 353.0831 5.87 191, 179, 173, 135 C16H18O9 trans-4-Caffeoylquinic acid trans-4-CQA

4 367.0980 10.03 193, 173 C17H20O9 4-Feruloylquinic acid 4-FQA

5 515.1166 14.01 353, 335, 191, 179, 135 C25H24O12 3,4-di-Caffeoylquinic acid 3,4-diCQA

6 515.1195 14.34 353, 191, 179, 135 C25H24O12 3,5-di-Caffeoylquinic acid 3,5-diCQA

7 515.1219 15.17 353, 335, 191, 179, 173, C25H24O12 4,5-di-Caffeoylquinic acid 4,5-diCQA 135

8 529.1398 15.54 367, 353, 335, 193, 179, C26H26O12 3-Feruloyl-4-caffeoylquinic acid 3F-4CQA 173,134

9 529.1013 15.72 367, 335, 193, 173 C26H26O12 3-Caffeoyl-4-feruloylquinic acid 3C-4FQA

10 529.0983 16.00 367, 193, 134 C26H26O12 3-Feruloyl-5-caffeoylquinic acid 3F-5CQA

11 529.1345 16.11 367, 353, 191, 179 C26H26O12 3-Caffeoyl-5-feruloylquinic acid 3C-5FQA

12 529.1345 16.53 367, 193, 173 C26H26O12 4-Feruloyl-5-caffeoylquinic acid 4F-5CQA

13 529.117 16.68 367, 353, 191, 179, 173, C26H26O12 4-Caffeoyl-5-feruloylquinic acid 4C-5FQA 135

14 677.1561 17.51 515, 353,179, 173 C34H30O15 tri-Caffeoylquinic acid triCQA

* Metabolites listed were identified across all conditions in callus derived from leaves and stems with differential intensities (Figure 4.4.4). B. pilosa callus were maintained on media with PGRs (2,4-D to BAP) at the following ratios [1:10] (1), [10:1] (2), [0:0] (3), [1:2] (4), [1:20] (5) and [1:40] (6). The concentrations were as follows: (1) 0.2 mg/L 2,4-D and 2.0 mg/L BAP, (2) 2.0 mg/L 2,4-D and 0.2 mg/L BAP, (3) 0 mg/L 2,4-D and 0 mg/L BAP (4) 0.45 mg/L 2,4-D and 1.0 mg/L BAP, (5) 0.3 mg/L 2,4-D and 4.0 mg/L BAP and (6) 0.2 mg/L 2,4-D and 8.0 mg/L BAP.

The distribution or accumulation of the HCA derivatives in response to alterations of concentration ratios of 2,4-D to BAP were investigated and highlighted by means of colour- coded heatmaps (Figure 4.2.4). These were generated from MetaboAnalyst (www.metaboanalyst.ca), in which the resulting heatmaps indicated differential metabolite concentration patterns in response to manipulations with PGRs (Chong et al., 2018). A colour gradient was used to indicate abundances of the secondary metabolites where deep red indicates the highest relative abundances and dark blue indicated the lowest abundances.

In leaf-derived callus (Figure 4.4.4-A), differential abundance of HCA derivatives were observed in response to alterations of concentration ratios of 2,4-D to BAP ([1:10], [10:1], [0:0], [1:2], [1:20] and [1:40]). Interestingly, leaf callus grown on media without PGRs ([0:0] 2,4-D to BAP) was found to maintain metabolic responsiveness, as 3C-4FQA and 3C-5FQA were found to be relatively abundant in this callus type. However, these metabolites were also relatively more abundant in callus maintained on media with high cytokinin ([1:10] 2,4-D to BAP). This could indicate that although B. pilosa leaf callus could accumulate some HCA derivatives independent of PGRs in culture media, PGRs were still required to increase the abundances of these metabolites. HCA derivatives were also found to be abundant in culture with other combinations of concentration ratios of PGRs ([10:1], [1:2], [1:20] and [1:40] 2,4- D to BAP).

In contrast, stem-derived calllus (Figure 4.4.4-B) abundantly produced most HCA derivatives independent of PGRs ([0:0] 2,4-D to BAP). Generally, some auxins may upregulate production of phenolics (Park et al., 2017). As also observed in this study, high auxin concentration ([10:1] 2,4-D to BAP) resulted in accumulation of 3,5-diCQA, 3F-4CQA, 4C-5FQA and 4F-5CQA in leaf callus of B. pilosa. Partial suppression of the phenylpropanoid pathway was observed in stem - and leaf callus grown on media with [1:2] 2,4-D to BAP, as some HCA derivatives were reduced (Figure 4.4.4, indicated by black rectangles). This could suggest that other auxin- cytokinin concentration combinations were better for optimal production of HCA derivatives in cell culture of B. pilosa.

This study highlights the differential effects of auxin-cytokinin interactions on secondary metabolite production in plant cell culture of B. pilosa. This study also permits the investigation of optimal PGRs concentration combinations for the biosynthesis of HCA derivatives in cell cultures of B. pilosa.

Figure 4.4.4: Heatmaps generated from hierarchical clustering (indicating group similarities) illustrating the occurrence/ distribution of hydroxycinnamic acids (HCAs) and caffeoyl/feruloyl quinic acids derivatives (C/FQAs) in B. pilosa leaf callus (A) and stem callus (B), maintained on media with different ratios of 2,4-D to BAP ([1:10], [10:1], [0:0], [1:2], [1:20] and [1:40]). Group averages were used to simplify the visualisation of the distribution of these HCA derivatives. Black rectangles indicate groups of HCA derivatives that were significantly reduced in the [1:2] 2,4-D to BAP combination.

Chapter Five: General Conclusions and Future Perspectives

In this study, metabolomics tools and approaches were utilised to investigate and profile the hydroxycinnamic acids and their ester derivatives of Bidens pilosa, an important food and medicinal plant. Comparative metabolomic profiling of different tissues (leaves, stems and roots) indicated that methanol extracts of this plant contain at least 30 HCA derivatives. Herein, chemically diverse esters of quinic acid (chlorogenic acids) and tartaric acid were found to be differentially distributed throughout the tissues. Understanding the bio-distribution of secondary metabolites across tissues is important in understanding the bio-medicinal value of the various tissues (Wang et al., 2005; Rui et al., 2019). Extracts were separated by reversed phase liquid chromatography coupled to mass spectrometry. Multivariate statistical analysis was applied to the data sets and revealed systematic trends in the tissues studied, indicating tissue-specific differences in the distribution of HCA derivatives. Chlorogenic acids are synthesized by the chloroplast-located HQT / HCT enzymes. Most of the HCA derivatives were found to be abundant in the leaves and stems of B. pilosa, compared to the roots. Furthermore, tartaric acid esters (caftaric acid and chicoric acid) were exclusively identified to be present in stems and leaves, thus the aerial parts. Metabolomics is a hypothesis-generating discipline and these findings suggested that the enzyme hydroxycinnamoyl-CoA: tartaric acid hydroxycinnamoyl transferase (HTT), which is proposed to be responsible for the biosynthesis of tartaric acid esters, may be localized in the aerial parts of B. pilosa. However, future studies to isolate and characterize full-length cDNA sequences encoding HTT in tissues of B. pilosa would be necessary to confirm the proposed theory.

CGAs that were identified in B. pilosa were found to be structurally complex metabolites occurring as regio-isomers and also as geometric isomers (Clifford, 2000; Clifford et al., 2003; Mhlongo et al., 2015). Adding to the complexity, CGAs also occur as mixed esters of HCAs and quinic acid (e.g. caffeoylferuloylquinic acid and caffeoylcoumaroylquinic acid) (Jaiswal et al., 2010; Wianowska and Gil, 2019). In this study various regio- and geometric isomers and hetero-acyl CGAs (e.g. caffeoylferuloylquinic acid, caffeoylcoumaroylquinic acid) were identified, which through application of MS fragmentation patterns obtained post-ISCID, were efficiently discriminated. This study demonstrated UHPLC-qTOF-MS/MS-based ISCID as a useful technique for annotation of structurally complex metabolites. Future exploration of computer-assisted structural elucidation (CASE) may simplify structural clarification of complex CGAs. CASE applies artificial intelligence, pattern recognition and spectral simulation to assist in metabolite annotations (Vaniya and Fiehn, 2015). Fragmentation - and mass spectral trees and similarity filters can also be constructed to assist in elucidation of substructures of CGAs (Liu et al., 2018). This could be used to build comprehensive mass

spectral libraries that account for regio- and geometric isomers of CGAs based on data obtained from UHPLC-qTOF-MS/MS-based ISCID.

Centred on metabolomic profiling of the extracted phytochemicals from B. pilosa leaves treated with signal molecules, differential responses of secondary metabolites were observed. MeJA and MeSA were used as elicitors with the aim to enhance the biosynthesis of secondary metabolites in this plant. MeJA and MeSA were shown to enhance levels of some HCA derivatives and flavonoids in leaf tissue of treated plants. Time-dependent differences were also observed in the metabolite levels. These results highlighted the possibility of using phytohormones to enhance the accumulation of bioactive secondary metabolites, such as the CGA phenolics in this plant. In future, the effects of these phytohormones on accumulation of important metabolites of B. pilosa can be studied in vitro. This will allow for homogenous treatment of cells, elicitor concentration optimisation and a possibility of scaling up production of metabolites in appropriate bioreactors (Rao and Ravishankar, 2002; Mendoza et al., 2018).

Plant cell cultures of B. pilosa were successfully initiated from stem and leaf explant material. Interestingly, metabolomic profiling of such generated callus and cell suspensions demonstrated differential cell-line specific metabolite distribution, similar to the tissue-specific distribution initially observed in the corresponding differentiated tissues. This showed that B. pilosa cell cultures are a promising alternative approach for the production of high-value secondary metabolites such as the HCA derivatives. To enhance the biosynthesis of HCA derivatives in vitro, various strategies can be followed as mentioned in section 4.3.3. Future experiments that involve precursor feeding in vitro can be designed where precursors of the phenylpropanoid pathway can be incorporated into the culture media.

This study, also aimed at investigating the effects of auxin (2,4-D) and cytokinin (BAP) on the ability of undifferentiated B. pilosa callus cells to regenerate, and to profile the metabolite distribution patterns resulting from the exogenous applications of plant growth regulators. Although the results indicated that treatment with different combinations of auxin and cytokinins did not initiate organogenesis, the callus demonstrated differential accumulation of HCA derivatives in response to the various phytohormone concentration ratios investigated. This illustrated the importance of optimization of concentration ratios of plant growth regulators for efficient accumulation of phenolic compounds in vitro.

Based on the results obtained in this study, the original hypotheses as stated in the introductory chapter (chapter 1), were partially proven. In this study, a UHPLC-qTOF-MS/MS-based ISCID metabolomics procedure was highlighted to be a powerful technique in the elucidation of structurally complex and closely related isomers of CGAs in plant tissues and undifferentiated tissues. Metabolomics allowed for the evaluation and elucidation of the effects of exogenous treatment with signal molecules (MeJA and MeSA) on the metabolomic profiles of B. pilosa. However, undifferentiated cells of B. pilosa did not undergo organogenesis under the investigated PGRs concentration combinations. However, the PGR manipulated growth did result in differential metabolite profiles as initially hypothesised. Finally, the current study demonstrated deeper insights into the metabolomes of differentiated and undifferentiated

tissues of B. pilosa. This emphasized the need to further elucidate underlying biochemical pathways, networks and inducible responses of B. pilosa to external stimuli.

-o-o-O-o-o-

Chapter 6: References

AbouZid, S. (2014). Yield improvement strategies for the production of secondary metabolites in plant tissue culture: ilymarin from Silybum marianum tissue culture. Natural Product Research, 28, 2102-2110. doi:10.1080/14786419.2014.927465. Achakzai, A. K. K., Achakzai, P., Masood, A., Kayani, S. A., and Tareen, R. B. (2009). Response of plant parts and age on the distribution of secondary metabolites on plants found in Quetta. Pakistan Journal of Botany, 41, 2129-2135. doi:10.1097/IGC.0b013e31821b2568. Adedapo, A., Jimoh, F., and Afolayan, A. (2011). Comparison of the nutritive value and biological activities of the acetone, methanol and water extracts of the leaves of Bidens pilosa and Chenopodium album. Acta Poloniae Pharmaceutica - Drug Research, 68, 83–92. Afendi, F. M., Okada, T., Yamazaki, M., Hirai-Morita, A., Nakamura, Y., et al. (2012). KNApSAcK family databases: Integrated metabolite-plant species databases for multifaceted plant research. Plant Cell Plant and Cell Physiology, 53, 1-12. doi:10.1093/pcp/pcr165. Alagarsamy, K., Shamala, L. F., and Wei, S. (2018). Influence of media supplements on inhibition of oxidative browning and bacterial endophytes of Camellia sinensis var. sinensis. 3 Biotech, 8, 1–7. doi:10.1007/s13205-018-1378-9. Al Kadhi, O., Melchini, A., Mithen, R., and Saha, S. (2017). Development of a LC-MS/MS method for the simultaneous detection of tricarboxylic acid cycle intermediates in a range of biological matrices. Journal of Analytical Methods in Chemistry, 2017, 1–12. doi:10.1155/2017/5391832. Al-sane, K. O., Shibli, R. A., Freihat, N. M., and Hammouri, M. K. (2005). Cell suspension culture and secondary metabolites production in african violet (Saintpaulia ionantha Wendl). Jordan Journal of Agricultural Sciences, 1, 84-92. doi:10.2519/jospt.2012.3941. Alam, M. A., Subhan, N., Hossain, H., Hossain, M., Reza, H. M., et al. (2016). Hydroxycinnamic acid derivatives: A potential class of natural compounds for the management of lipid metabolism and obesity. Nutrition and Metabolism, 13, 1-13. doi:10.1186/s12986-016-0080-3. Alonso, A., Marsal, S., and JuliÃ, A. (2015). Analytical methods in untargeted metabolomics: state of the art in 2015. Frontiers in Bioengineering and Biotechnology, 3, 1-20. doi:10.3389/fbioe.2015.00023. Álvarez-Sánchez, B., Priego-Capote, F. and Luque de Castro, M. D. (2010a) Metabolomics analysis I. Selection of biological samples and practical aspects preceding sample preparation. Trends in Analytical Chemistry, 29, 111–119. doi: 10.1016/j.trac.2009.12.003. Álvarez-Sánchez, B., Priego-Capote, F. and Luque de Castro, M. D. (2010b) Metabolomics analysis II. Preparation of biological samples prior to detection. Trends in Analytical Chemistry, 29, 120–127.

doi.org/10.1016/j.trac.2009.12.004 Andersson, E. R., Day, R. D., Loewenstein, J. M., Woodley, C. M., and Schock, T. B. (2019). Evaluation of sample preparation methods for the analysis of reef-building corals using 1 H-NMR- based metabolomics. Metabolites, 9, 32. doi:10.3390/metabo9020032. Anterola, A. M., Jeon, J. H., Davin, L. B., and Lewis, N. G. (2002). Transcriptional control of monolignol biosynthesis in Pinus taeda: factors affecting monolignol ratios and carbon allocation in phenylpropanoid metabolism. Journal of Biological Chemistry, 277, 18272-18280. doi:10.1074/jbc.M112051200. Arthur, G. D., Naidoo, K. K., and Coopoosamy, R. M. (2012). Bidens pilosa L.: agricultural and pharmaceutical importance. Journal of Medicinal Plants Research, 6, 3282-3287. doi:10.5897/JMPR012.195. Ashafa, A., and Afolayan, A. (2009). Screening the root extracts from Bidens pilosa L. var. radiata (Asteraceae) for antimicrobial potentials. Journal of Medicinal Plants Research, 3, 268-572. doi:10.1007/s12237-016-0207-9. Azay-Milhau, J., Ferrare, K., Leroy, J., Aubaterre, J., Tournier, M., et al. (2013). Antihyperglycemic effect of a natural chicoric acid extract of chicory (Cichorium intybus L.): a comparative in vitro study with the effects of caffeic and ferulic acids. Journal of Ethnopharmacology, 150, 755-760. doi:10.1016/j.jep.2013.09.046. Baira, S. M., Gunupudi, P., Abbaraju, V., Srinivas, R., and Talluri, M. V. N. K. (2018). Study of forced degradation behaviour of cobicistat and atazanavir using LC/ESI/QTOF/MS; a combination of in- sourced and collision-induced dissociation for evaluation of the fragmentation patterns of degradation products. New Journal of Chemistry, 42, 19113-19128. doi:10.1039/c8nj03418d. Balmer, D., Flors, V., Glauser, G., and Mauch-Mani, B. (2013). Metabolomics of cereals under biotic stress: current knowledge and techniques. Frontiers in Plant Science, 4, 1-12. doi:10.3389/fpls.2013.00082. Balmer, A., Pastor, V., Gamir, J., Flors, V., and Mauch-Mani, B. (2015). The “prime-ome”: Towards a holistic approach to priming. Trends Plant Science. 20, 443–452. doi:10.1016/j.tplants.2015.04.002. Bartel, J., Krumsiek, J., and Theis, F. J. (2013). Statistical methods for the analysis of high-throughput metabolomics data. Computational and Structural Biotechnology Journal. 4, e201301009. doi:10.5936/csbj.201301009. Bartolome, A. P., Villaseñor, I. M., and Yang, W. C. (2013). Bidens pilosa L. (Asteraceae): Botanical properties, traditional uses, phytochemistry, and pharmacology. Evidence-based Complementary and Alternative Medicine, 2013, 1-51.doi:10.1155/2013/340215. Berg, M., Vanaerschot, M., Jankevics, A., Cuypers, B., Breitling, R., et al. (2013). LC-MS metabolomics from study design to data-analysis-using a versatile pathogen as a test case. Computational and Structural Biotechnology Journal, 4, e201301002.

doi:10.5936/csbj.201301002. Bhat, R. B., Rubuluza, T., and Jäger, A. K. (2016). The Bio-diversity of traditional vegetables of the Transkei region in the Eastern Cape of South Africa. South African Journal of Botany, 68, 94-99. doi:10.1016/s0254-6299(16)30463-x. Bi, H. H., Zeng, R. Sen, Su, L. M., An, M., and Luo, S. M. (2007). Rice allelopathy induced by methyl jasmonate and methyl salicylate. Journal of Chemical Ecology, 33, 1089–1103. doi:10.1007/s10886-007-9286-1. Bijttebier, S., Van der Auwera, A., Foubert, K., Voorspoels, S., Pieters, L. et al. (2016). Bridging the gap between comprehensive extraction protocols in plant metabolomics studies and method validation. Analytica Chimica Acta, 935, 136-150. doi:10.1016/j.aca.2016.06.047. Bilanda, D. C., Dzeufiet, P. D. D., Kouakep, L., Aboubakar, B. F. O., Tedong, L., et al. (2017). Bidens pilosa Ethylene acetate extract can protect against L-NAME-induced hypertension on rats. BMC Complementary and Alternative Medicine. 17, 1-7. doi:10.1186/s12906-017-1972-0. Bourgaud, F., Gravot, A., Milesi, S., and Gontier, E. (2001). Production of plant secondary metabolites: A historical perspective. Plant Science,161, 839–851. doi:10.1016/S0168-9452(01)00490-3. Brown, M., Dunn, W. B., Ellis, D. I., Goodacre, R., Handl, J., et al. (2005). A metabolome pipeline: from concept to data to knowledge. Metabolomics, 1, 39-51. doi:10.1007/s11306-005-1106-4. Brown, T. J. (1990). The initiation and maintenance of callus culture. in Methods in molecular biology, eds. J. W. Pollard and J. M. Walker (New Jersey), 57–63. Cardoso, J. C., de Oliveira, M. E. B. S., and Cardoso, F. de C. I. (2019). Advances and challenges on the in vitro production of secondary metabolites from medicinal plants. Horticultura Brasileira, 37, 124–132. doi:10.1590/s0102-053620190201. Călinoiu, L. F., and Vodnar, D. C. (2018). Whole grains and phenolic acids: a review on bioactivity, functionality, health benefits and bioavailability. Nutrients, 10, 1615. doi:10.3390/nu10111615. Cambiaghi, A., Ferrario., M., and Masseroli., M. (2017). Analysis of metabolomic data: tools, current strategies and future challenges for omics data integration. Briefings in Bioinformatics, 18, 498-

510. doi: 10.1093/bib/bbw031 Chaleckis, R., Meister, I., Zhang, P., and Wheelock, C. E. (2019). Challenges, progress and promises of metabolite annotation for LC-MS-based metabolomics. Current Opinion in Biotechnology, 55, 44-50. doi:10.1016/j.copbio.2018.07.010. Chang, H. P., Hyeon, J. Y., Ye, E. P., SeWon, C., Yong, S. C., et al. (2019). Influence of chitosan, salicylic acid and jasmonic acid on phenylpropanoid accumulation in germinated Buckwheat (Fagopyrum esculentum Moench). Foods, 8, 1-10. Chang, S. L., Chiang, Y. M., Chang, C. L. T., Yeh, H. H., Shyur, L. F., et al. (2007). Flavonoids, centaurein and centaureidin, from Bidens pilosa, stimulate IFN-γ expression. Journal of Ethnopharmacology, 112, 232-236. doi:10.1016/j.jep.2007.03.001. Chaudhary, G., and Dantu, P. K. (2015). Evaluation of callus browning and develop a strategically

callus culturing of Boerhaavia diffusa L. Journal of Plant Development, 22, 47-58. Chawla, G., and Ranjan, C. (2016). Principle, instrumentation, and applications of UPLC: a novel technique of liquid chromatography. Open Chemistry Journal, 3, 1-16. doi:10.2174/1874842201603010001. Chen, S., Qin, W., Zhang, Q., Kong, M., Wu, H., et al. (2016). An overview of plant phenolic compounds and their importance in human nutrition and management of type 2 diabetes. Molecules, 21, 1374. doi:10.3390/molecules21101374. Chen, Y. L., Huang, S. T., Sun, F. M., Chiang, Y. L., Chiang, C. et al. (2011). Transformation of cinnamic acid from trans- to cis-form raises a notable bactericidal and synergistic activity against multiple-drug resistant Mycobacterium tuberculosis. European Journal of Pharmaceutical Sciences, 43, 188-194. doi:10.1016/j.ejps.2011.04.012. Chen, S., Lin, J., Liu, H., Gong, Z., Wang, et al. (2018). Insights into tissue-specific specialized metabolism in Tieguanyin tea cultivar by untargeted metabolomics. Molecules, 23. doi:10.3390/molecules23071817. Chen, Z., Tang, N., You, Y., Lan, J., Liu, Y., et al. (2015). Transcriptome analysis reveals the mechanism underlying the production of a high quantity of chlorogenic acid in young leaves of Lonicera macranthoides Hand.-Mazz. Plos One, 10, 1-18. doi:10.1371/journal.pone.0137212. Chen, Y., Yu, H., Wu, H., Pan, Y., Wang, K., et al. (2015). Characterization and quantification by LC- MS/MS of the chemical components of the heating products of the flavonoids extract in Pollen typhae for transformation rule exploration. Molecules, 20, 18352-18366. doi:10.3390/molecules201018352. Chhipa, N. M. R., Patel, K. M., Ganchi, S. P., Dhrubo, P., and Sen, J. (2014). Chicoric acid and its analogues as an anti-HIV integrase agents. World Journal of Pharmacy and Pharmaceutical Sciences, 3, 2321–2335. Chiang, Y. M., Chuang, D. Y., Wang, S. Y., Kuo, Y. H., Tsai, P. W., et al. (2004). Metabolite profiling and chemopreventive bioactivity of plant extracts from Bidens pilosa. Journal of Ethnopharmacology, 95, 409-419. doi:10.1016/j.jep.2004.08.010. Chien, S., Young, P. H., Hsu, Y., Chen, C., Tien, Y., et al. (2009). Phytochemistry anti-diabetic properties of three common Bidens pilosa variants in Taiwan. Phytochemistry, 70, 1246-1254. doi:10.1016/j.phytochem.2009.07.011. Chong, J., Soufan, O., Li, C., Caraus, I., Li, S., Bourque, G., et al. (2018). MetaboAnalyst 4.0: Towards more transparent and integrative metabolomics analysis. Nucleic Acids Research. 46, 486–494. doi:10.1093/nar/gky310. Chkhikvishvili, I. D., and Kharebava, G. I. (2001). Chicoric and chlorogenic acids in plant species from Georgia. Applied Biochemistry and Microbiology, 37, 214-217. doi:10.1023/A:1002888016985 Clifford, M. N. (2000). Chlorogenic acids and other cinnamates - Nature, occurrence, dietary burden, absorption and metabolism. Journal of the Science of Food and Agriculture, 80, 1033-1043.

doi:10.1002/(SICI)1097-0010(20000515)80:7<1033::AID-JSFA595>3.0.CO;2-T. Clifford, M. N., Jaganath, I. B., Ludwig, I. A., and Crozier, A. (2017). Chlorogenic acids and the acyl- quinic acids: Discovery, biosynthesis, bioavailability and bioactivity. Natural Product Reports, 34, 1391-1421. doi:10.1039/c7np00030h. Clifford, M. N., Johnston, K. L., Knight, S., and Kuhnert, N. (2003). Hierarchical scheme for LC-MSn identification of chlorogenic acids. Journal of Agricultural and Food Chemistry, 51, 2900-2911. doi:10.1021/jf026187q. Clifford, M. N., Kirkpatrick, J., Kuhnert, N., Roozendaal, H., and Salgado, P. R. (2008). LC-MSn analysis of the cis isomers of chlorogenic acids. Food Chemistry, 106, 379-385. doi:10.1016/j.foodchem.2007.05.081. Clifford, M. N., Knight, S., and Kuhnert, N. (2005). Discriminating between the six isomers of dicaffeoylquinic acid by LC-MSn. Journal of Agricultural and Food Chemistry, 53, 3821-3832. doi:10.1021/jf050046h. Cliford, M.N., and Madala, E.N. (2017). Surrogate standards: a cost-effective strategy for identification of phytochemicals. Journal of Agricultural and Food Chemistry, 65, 3589-3590. doi:10.1021/acs.jafc.7b01588. Clifford, M. N., Marks, S., Knight, S., and Kuhnert, N. (2006). Characterization by LC-MSn of four new classes of p-coumaric acid-containing diacyl chlorogenic acids in green coffee beans. Journal of Agricultural and Food Chemistry, 54, 4095-4101. doi:10.1021/jf060536p. Clifford, M. N., Wu, W., Kirkpatrick, J., and Kuhnert, N. (2007). Profiling the chlorogenic acids and other caffeic acid derivatives of herbal chrysanthemum by LC-MSn. Journal of Agricultural and Food Chemistry, 55, 929-936. doi:10.1021/jf062314x. Coenen, C., and Lomax, T. L. (1997). Auxin-cytokinin interactions in higher plants: old problems and new tools. Trends Plant Science, 2, 351-355. doi: 10.1016/S1360-1385(97)84623-7. Comino, C., Lanteri, S., Portis, E., Acquadro, A., Romani, A., et al. (2007). Isolation and functional characterization of a cDNA coding a hydroxycinnamoyltransferase involved in phenylpropanoid biosynthesis in Cynara cardunculus L.. BMC Plant Biology, 7, 14. doi:10.1186/1471-2229-7-14. Cortés-Rojas, D. F., Chagas-Paula, D. A., Da Costa, F. B., Souza, C. R. F., and Oliveira, W. P. (2013). Bioactive compounds in Bidens pilosa L. populations: A key step in the standardization of phytopharmaceutical preparations. Brazilian Journal of Pharmacognosy, 23, 28-35. doi:10.1590/S0102-695X2012005000100. Daub, C. O., Kloska, S., and Selbig, J. (2003). MetaGeneAlyse: analysis of integrated transcriptional and metabolite data. Bioinformatics, 19, 2332-2333. doi:10.1093/bioinformatics/btg321. de Wet, H., Nzama, V. N., and Van Vuuren, S. F. (2012). Medicinal plants used for the treatment of sexually transmitted infections by lay people in northern Maputaland, KwaZulu – Natal province, South Africa. South African Journal of Botany, 78, 12-20. doi:10.1016/j.sajb.2011.04.002. Dicke, M., and Hilker, M. (2003). Induced plant defences: From molecular biology to evolutionary

100

ecology. Basic and Applied Ecology, 4, 3-14. doi:10.1078/1439-1791-00129. Dörnenburg, H. (2008). Plant cell culture technology-harnessing a biological approach for competitive cyclotides production. Biotechnology Letters, 30, 1311-1321. doi:10.1007/s10529-008-9704-7. Dorenenburg, H., and Knorr, D. (1995). Strategies to improve secondary metabolites in plant cultures. Enzyme and Microbial Technology, 17, 674-684. dos Santos, F. A., Sousa, I. P., Furtado, N. A. J. C., da Costa, F. B. (2018). Combined OPLS-DA and decision tree as a strategy to identify antimicrobial biomarkers of volatile oils analyzed by gas chromatography-mass spectrometry. Brazilian Journal of Pharmacognosy, 28, 647-653. doi:10.1016/j.bjp.2018.08.006. Douglas, C. J. (1996). Phenylpropanoid metabolism and lignin biosynthesis: from weeds to trees. Trends in Plant Science, 1, 171-178. doi:10.1016/1360-1385(96)10019-4. Dunn, W. B., and Ellis, D. I. (2005). Metabolomics: Current analytical platforms and methodologies. TrAC - Trends in Analytical Chemistry, 24, 285-294. doi:10.1016/j.trac.2004.11.021. Dunn, W. B., Erban, A., Weber, R. J. M., Creek, D. J., Brown, M., et al. (2013). Mass appeal: Metabolite identification in mass spectrometry-focused untargeted metabolomics. Metabolomics, 9, 44-66. doi:10.1007/s11306-012-0434-4. Efferth, T. (2019). Biotechnology applications of plant callus cultures. Engineering, 5, 50-59. doi:10.1016/j.eng.2018.11.006. El-Seedi, H. R., Taher, E. A., Sheikh, B. Y., Anjum, S., Saeed, A., et al. (2017). Hydroxycinnamic acids: natural sources, biosynthesis, possible biological activities, and roles in Islamic medicine. Studies in Natural Products Chemistry, 55, 1-29. doi:10.1016/B978-0-444-64068-0.00008-5. Eibl, R., Meier, P., Stutz, I., Schildberger, D., Hühn, T., et al. (2018). Plant cell culture technology in the cosmetics and food industries: current state and future trends. Applied Microbiology and Biotechnology, 102, 8661-8675. doi:10.1007/s00253-018-9279-8. Elaleem, K. G. A., Modawi, R. S., and Khalafalla, M. M. (2009). Effect of plant growth regulators on callus induction and plant regeneration in tuber segment culture of potato (Solanum tuberosum L.) cultivar Diamant. African Journal of Biotechnology, 8, 2529-2534. Erbilgin, O., Rübel, O., Louie, K. B., Trinh, M., Raad, M., et al. (2019). MAGI: a method for metabolite annotation and gene integration. ACS Chemical Biology, 14, 704-714. doi:10.1021/acschembio.8b01107. Eriksson, L., Trygg, J., and Wold, S. (2008). CV-ANOVA for significance testing of PLS and OPLS® models. Journal of Chemometrics, 22, 594-600. doi:10.1002/cem.1187. Ferrare, K., Bidel, L. P. R., Awwad, A., Poucheret, P., Cazals, G., et al. (2018). Increase in insulin sensitivity by the association of chicoric acid and chlorogenic acid contained in a natural chicoric acid extract (NCRAE) of chicory (Cichorium intybus L.) for an antidiabetic effect. Journal of Ethnopharmacology, 215, 241–248. doi:10.1016/j.jep.2017.12.035. Figueiredo, A. C., Barroso, J. G., Pedro, L. G., and Johannes, J. C. S. (2008). Factors affecting secondar

101

metabolite production in plants: volatile components and essential oils. Flavour and Fragrance Journal, 23, 213-226. doi:10.1002/ffj.1875. Filova, A. (2014). Production of secondary metabolites in plant tissue cultures. Research Journal of Agricultural Science, 46, 236-245. Fischer, R., Emans, N., Schuster, F., Hellwig, S., and Drossard, J. (1999). Towards molecular farming in the future: using plant-cell-suspension cultures as bioreactors. Biotechnology and Applied Biochemistry, 30, 109-12. doi:10.1093/nar/gku825. Fukushima, A., and Kusano, M. (2013). Recent progress in the development of metabolome databases for plant systems biology. Frontiers in Plant Science, 4, 1-11. doi:10.3389/fpls.2013.00073. Gardinassi, L. G., Xia, J., Safo, S. E., and Li, S. (2017). Bioinformatics tools for the interpretation of metabolomics data. Current Pharmacology Reports, 3, 374-83. doi:10.1007/s40495-017-0107-0. Gaspar, T., Kevers, C., Bisbis, B., Franck, T., Crevecoeur, M., et al. (2000). Special symposium : in vitro plant recalcitrance. In Vitro Cellular & Developmental Biology, 36, 171–181. Gbashi, S., Njobeh, P., Steenkamp, P., and Madala, N. (2017). Pressurized hot water extraction and chemometric fingerprinting of flavonoids from Bidens pilosa by UPLC-tandem mass spectrometry . CyTA – Journal of Food, 15, 171-180. doi:10.1080/19476337.2016.1230151. Geissberger, P., and Séquin, U. (1991). Constituents of Bidens pilosa L.: do the components found so far explain the use of this plant in traditional medicine?. Acta Tropica, 48, 251-261. doi:10.1016/0001-706X(91)90013-A. Ghanati, F., and Bakhtiarian, S. (2014). Effect of methyl jasmonate and silver nanoparticles on production of secondary metabolites by Calendula Officinalis L (Asteraceae). Tropical Journal of Pharmaceutical Research, 13, 1783-1789. doi:10.4314/tjpr.v13i11.2. Ghasemzadeh, A., and Ghasemzadeh, N. (2011). Flavonoids and phenolic acids: role and biochemical activity in plants and human. Journal of Medicinal Plants Research, 3, 6697-6703. doi:10.5897/JMPR11.1404. Gika, H. G., Zisi, C., Theodoridis, G., and Wilson, I. D. (2016). Protocol for quality control in metabolic profiling of biological fluids by U(H)PLC-MS. Journal of Chromatography B: Analytical Technologies in the Biomedical and Life Sciences, 1008, 15-25. doi:10.1016/j.jchromb.2015.10.045. Godzien, J., Ciborowski, M., Angulo, S., and Barbas, C. (2013). From numbers to a biological sense: How the strategy chosen for metabolomics data treatment may affect final results. A practical example based on urine fingerprints obtained by LC-MS. Electrophoresis, 34, 2812-2826. doi:10.1002/elps.201300053. Goleniowski, M., Bonfill, M., Cusido, R., and Palazón, J. (2013). Phenolic acids, In Natural Products: Phytochemistry, Botany and Metabolism of Alkaloids, Phenolics and Terpenes; Ramawat, K.G., Merlillon, J.M., Eds.; Springer-Verla:Berlin Heidelbergen, Germany, 1951-1973. doi:10.1007/978-3-642-22144-6_64.

102

Gong, Z. G., Hu, J., Wu, X., and Xu, Y. J. (2017). The recent developments in sample preparation for mass spectrometry-based metabolomics. Critical Reviews in Analytical Chemistry, 47, 325-331. doi:10.1080/10408347.2017.1289836. Goodacre, R., Broadhurst, D., Smilde, A. K., Kristal, B. S., Baker, J. D., et al. (2007). Proposed minimum reporting standards for data analysis in metabolomics. Metabolomics, 3, 231-241. doi:10.1007/s11306-007-0081-3. Gowda, H., Ivanisevic, J., Johnson, C. H., Kurczy, M. E., Benton, H. P., et al. (2014). Interactive XCMS online: Simplifying advanced metabolomic data processing and subsequent statistical analyses. Analytical Chemistry, 86, 6931-6939. doi:10.1021/ac500734c. Granato, D., Santos, J. S., Escher, G. B., Ferreira, B. L., and Maggio, R. M. (2018). Use of principal component analysis (PCA) and hierarchical cluster analysis (HCA) for multivariate association between bioactive compounds and functional properties in foods: a critical perspective. Trends Food Science and Technology,72, 83-90. doi:10.1016/j.tifs.2017.12.006. Guan, Y., Hu, W., Jiang, A., Xu, Y., Sa, R., Feng, K., et al. (2019). Effect of methyl jasmonate on phenolic accumulation in wounded Broccoli. Molecules, 24, 3537. Guijas, C., Montenegro-Burke, J. R., Warth, B., and Spilker, M. E. (2018). Metabolomics activity screening for identifying metabolites that modulate phenotype. Nature Biotechnology, 36, 316- 320. doi:10.1038/nbt.4101. Hall, R., Beale, M., Fiehn, O., Hardy, N., Sumner, L. W., et al. (2002). Plant metabolomics: the missing link in functional genomics strategies. Plant Cell Online, 14, 1437-1440. doi:10.1105/tpc.140720. Hamany Djande, C. Y., Steenkamp, P. A., Steenkamp, A., Piater, L. A., and Madala, N. E. (2019). Habituated Moringa oleifera callus retains metabolic responsiveness to external plant growth regulators. Plant Cell, Tissue and Organ Culture, 137, 249-264. doi:10.1007/s11240-019-01565- y. Healy, E. F., Sanders, J., King, P. J., and Robinson, W. E. (2009). A docking study of l-chicoric acid with HIV-1 integrase. Journal of Molecular Graphics and Modeling, 27, 584-589. doi:10.1016/j.jmgm.2008.09.011. Hee, G., Woong, H., Ki, M., Young, S., Bak, A. R., Jin, D., et al. (2018). Characterization and quantification of flavonoid glycosides in the Prunus genus by UPLC-DAD- QTOF / MS. Saudi Journal of Biologica Sciences, 25, 1622-1631. doi:10.1016/j.sjbs.2016.08.001. Hegeman, A. D. (2010). Plant metabolomics-meeting the analytical challenges of comprehensive metabolite analysis. Briefings in Functional Genomics and Proteomics, 9, 139-148. doi:10.1093/bfgp/elp053. Heleno, S. A., Martins, A., Queiroz, M. J. R. P., and Ferreira, I. C. F. R. (2015). Bioactivity of phenolic acids: metabolites versus parent compounds: a review. Food Chemistry, 173, 501–513. doi:10.1016/j.foodchem.2014.10.057. Hellwig, S., Drossard, J., Twyman, R. M., and Fischer, R. (2004). Plant cell cultures for the production

103

of recombinant proteins. Nature Biotechnology, 22, 1415-1422. doi:10.1038/nbt1027. Hernández-Sotomayor, S., Siddiqui, K., Guerriero, G., Abdel-Salam, E., Hausman, J.-F., et al. (2018). Production of plant secondary metabolites: examples, tips and suggestions for biotechnologists. Genes, 9, e309. doi:10.3390/genes9060309. Herrmann, K. M. (1995). The shikimate pathway: early steps in the biosynthesis of aromatic compounds.The Plant Cell, 7, 907. doi:10.2307/3870046. Heuberger, A. L., Robison, F. M., Lyons, S. M. A., Broeckling, C. D., and Prenni, J. E. (2014). Evaluating plant immunity using mass spectrometry-based metabolomics workflows. Front. Plant Science,5, 1-11. doi:10.3389/fpls.2014.00291. Hoffmann, L., Maury, S., Martz, F., Geoffroy, P., and Legrand, M. (2003). Purification, cloning, and properties of an acyltransferase controlling shikimate and quinate ester intermediates in phenylpropanoid metabolism. Journal of Biological Chemistry, 278, 95–103. doi:10.1074/jbc.M209362200. Hong, J., Yang, L., Zhang, D., and Shi, J. (2016). Plant metabolomics: an indispensable system biology tool for plant science. International Journal of Molecular Sciences, 17, 767. doi:10.3390/ijms17060767. Huan, T., Forsberg, E., Rinehart, D., Johnson, C. H., Ivanisevic, J., et al. (2017). Systems biology guided by XCMS Online metabolomics. Nature Methods, 14, 461-462. doi:10.1038/s41591-018- 0092-9.Senolytics. Hudson, A. O., and Prabhu, P. R. (2010). Identification and partial characterization of an L-tyrosine aminotransferase (TAT) from Arabidopsis thaliana. Biochemistry Research International, 1-11. doi:10.1155/2010/549572. Hussain, M. S., Fareed, S., Ansari, S., Rahman, M. A., Ahmad, I. Z., et al. (2012). Current approaches toward production of secondary plant metabolites. Journal of Pharmacy and Bioallied Sciences, 4, 10-20. doi:10.4103/0975-7406.92725. Ikeuchi, M., Ogawa, Y., Iwase, A., and Sugimoto, K. (2016). Plant regeneration : cellular origins and molecular mechanisms. Development, 143, 1442-1451. doi:10.1242/dev.134668. Ikeuchi, M., Sugimoto, K., and Iwase, A. (2013). Plant callus: mechanisms of induction and repression. Plant Cell, 25, 3159-73. doi:10.1105/tpc.113.116053. Iriti, M., and Faoro, F. (2009). Chemical diversity and defence metabolism: how plants cope with pathogens and ozone pollution. International Journal of Molecular Sciences, 10, 3371-3399. doi:10.3390/ijms10083371. Ivanisevic, J., Benton, H. P., Rinehart, D., Epstein, A., Kurczy, M. E., et al. (2015). An interactive cluster heat map to visualize and explore multidimensional metabolomic data. Metabolomics, 11, 1029-1034. doi:10.1007/s11306-014-0759-2. Jain, C., Khatana, S., and Vijayvergia, R. (2019). Bioactivity of secondary metabolites of various plants: a review. International Journal Pharmaceutical Science and Research, 10, 494-504.

104

doi:10.13040/IJPSR.0975-8232.10(2).494-04. Jaiswal, R., and Kuhnert, N. (2011). Identification and characterization of five new classes of chlorogenic acids in burdock (Arctium lappa L.) roots by liquid chromatography/tandem mass spectrometry. Food and Function, 2, 63-71. doi:10.1039/c0fo00125b. Jaiswal, R., Müller, H., Müller, A., Gamaleldin, M., Karar, E., et al. (2014). Phytochemistry identification and characterization of chlorogenic acids , chlorogenic acid glycosides and flavonoids from Lonicera henryi L . (Caprifoliaceae) leaves by LC-MSn. Phytochemistry, 108, 252-263. doi:10.1016/j.phytochem.2014.08.023. Jaiswal, R., Sovdat, T., Vivan, F., and Kuhnert, N. (2010). Profiling and characterization by LC-MSnof the chlorogenic acids and hydroxycinnamoylshikimate esters in maté (Ilex paraguariensis). J. Agric. Food Chem. 58, 5471–5484. doi:10.1021/jf904537z. Jamil, S. Z. M. R., Rohani, E. R., Baharum, S. N., and Noor, N. M. (2018). Metabolite profiles of callus and cell suspension cultures of mangosteen. 3 Biotech, 8, 1-14. doi:10.1007/s13205-018-1336-6. Jamwal, K., Bhattacharya, S., and Puri, S. (2018). Plant growth regulator mediated consequences of secondary metabolites in medicinal plants. Journal of Applied Research on Medicinal and Aromatic Plants, 9, 26-38. doi:10.1016/j.jarmap.2017.12.003. Jones, W. P., and Kinghorn, D. A. (2006). Extraction of Plant Secondary Metabolites, In natural products isolation, eds. S. D. Sarker, Z. Latif, and A. Gray. Human Press Inc (New Jersey), 279- 327. doi:10.1007/978-1-59259-256-2_10. Jorge, T. F., Mata, A. T., and António, C. (2016). Mass spectrometry as a quantitative tool in plant metabolomics. Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences, 374, 20150370. doi:10.1098/rsta.2015.0370. Juhascik, M. P., and Jenkins, A. J. (2009). Comparison of liquid/liquid and solid-phase extraction for alkaline drugs. Journal of Chromatographic Science, 47, 553-557. doi.org/10.1093/chromsci/47.7.553. Kaiser, N., Birkholz, D., Colomban, S., Navarini, L., and Engelhardt, U. H. (2013). A new method for the preparative isolation of chlorogenic acid lactones from coffee and model roasts of 5- caffeoylquinic acid. Journal of Agricultural and Food Chemistry, 61, 6937-6941. doi:10.1021/jf4011356. Kalaivani, K., Kalaiselvi, M. M., and Senthil-Nathan, S. (2016). Effect of methyl salicylate (MeSA), an elicitor on growth, physiology and pathology of resistant and susceptible rice varieties. Scientific Reports, 6, 1-11. doi:10.1038/srep34498. Kapoore, R. V., and Vaidyanathan, S. (2016). Towards quantitative mass spectrometry-based metabolomics in microbial and mammalian systems. Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences, 374, 2079. doi:10.1098/rsta.2015.0363. Karuppusamy, S. (2009). Review on trends in production of secondary metabolites from higher plants

105

by in vitro tissue, organ and cell cultures. Journal of Medicinal Plants Research, 3, 1222-1239. doi:DOI: 10.5897/JMPR. Kasai, H., Fukada, S., Yamaizumi, Z., Sugie, S., and Mori, H. (2000). Action of chlorogenic acid in vegetables and fruits as an inhibitor of 8-hydroxydeoxyguanosine formation in vitro and in a rat carcinogenesis model. Food and Chemical Toxicology, 38, 467-471. doi:10.1016/S0278- 6915(00)00014-4. Kaufmann, B., and Christen, P. (2002). Recent extraction techniques for natural products: microwave- assisted extraction and pressurised solvent extraction. Phytochemical Analysis, 13, 105-113. doi:10.1002/pca.631. Khoza, B. S., Gbashi, S., Steenkamp, P. A., Njobeh, P. B., and Madala, N. E. (2016). Identification of hydroxylcinnamoyl tartaric acid esters in Bidens pilosa by UPLC-tandem mass spectrometry. South African Journal of Botany, 103, 95-100. doi:10.1016/j.sajb.2015.08.018. Kim, H. J., Chen, F., Wang, X., and Rajapakse, N. C. (2006). Effect of methyl jasmonate on secondary metabolites of sweet basil (Ocimum basilicum L.). Journal of Agricultural and Food Chemistry, 54, 2327-2332. doi:10.1021/jf051979g. Kim, H. J., Fonseca, J. M., Choi, J.-H., and Kubota, C. (2007). Effect of methyl jasmonate on phenolic compounds and carotenoids of Romaine lettuce (Lactuca sativa L.). Journal of Agricultural and Food Chemistry, 55, 10366-10372. Kim, S. J., Kim, S. H., Kim, J. H., Hwang, S., and Yoo, H. J. (2016). Understanding Metabolomics in Biomedical Research. Endocrinology and Metabolism, 31, 7. doi:10.3803/enm.2016.31.1.7. Kim, H. K., and Verpoorte, R. (2010). Sample preparation for plant metabolomics. Phytochemical Analysis, 21, 4-13. doi:10.1002/pca.1188. Koornneef, A., Verhage, A., Leon-reyes, A., Snetselaar, R., Loon, L. C. Van, and Pieterse, C. M. J. (2008). Towards a reporter system to identify regulators of cross-talk between salicylate and jasmonate signaling pathways in Arabidopsis. Plant Signal Behavior, 3, 543-546. Kouassi, N.J., Bocca, C., Simard, G., Prunier-Mirebeau, D., Chao De La Barca, J. M., et al. (2017). A nontargeted UHPLC-HRMS metabolomics pipeline for metabolite identification: application to cardiac remote ischemic preconditioning. Analytical Chemistry, 89, 2138-2146. doi:10.1021/acs.analchem.6b04912. Kuban-Jankowska, A., Sahu, K. K., Gorska, M., Tuszynski, J. A., and Wozniak, M. (2016). Chicoric acid binds to two sites and decreases the activity of the YopH bacterial virulence factor. Oncotarget, 7, 2229-2238. doi:10.18632/oncotarget.6812. Kumar, R., Bohra, A., Pandey, A. K., Pandey, M. K., and Kumar, A. (2017). Metabolomics for Plant Improvement: Status and Prospects. Frontiers in Plant Science, 8, 1-27. doi:10.3389/fpls.2017.01302. Kumlay, A. M., and Ercisli, S. (2015). Callus induction, shoot proliferation and root regeneration of potato (Solanum tuberosum L.) stem node and leaf explants under long-day conditions.

106

Biotechnology and Biotechnological Equipment. 29, 1075-1084. doi:10.1080/13102818.2015.1077685. Kundu, a., and Vadassery, J. (2019). Chlorogenic acid-mediated chemical defence of plants against insect herbivores. Plant Biology, 21, 185-189. doi:10.1111/plb.12947. Lallemand, L. A., Zubieta, C., Lee, S. G., Wang, Y., Acajjaoui, S., Timmins, J., et al. (2012). A structural basis for the biosynthesis of the major chlorogenic acids found in coffee. Plant Physioliology,160, 249260. doi:10.1104/pp.112.202051. Laura, C. R., Mauricio, R. L., Urrea-Turjillo, A. I., and Lucia, A. G. (2018). Effect of nutrients starvation in the cell cycle synchronization of cell suspension cultures of Jatropha curcas L. Current Plant Biology, 13, 23-31. doi:10.1016/j.cpb.2018.03.002. Lee, D. Y., Bowen, B. P., and Northen, T. R. (2010). Mass spectrometry-based metabolomics, analysis of metabolite-protein interactions, and imaging. BioTechniques, 49, 557-565. doi:10.2144/000113451. Lee, Z. H., Hirakawa, T., Yamaguchi, N., and Ito, T. (2019). The roles of plant hormones and their interactions with regulatory genes in determining meristem activity. International Journal of Molecular Sciences, 20, 1-19. Lee, Y., Lee, D. E., Lee, H. S., Kim, S. K., Lee, W. S., et al. (2011). Influence of auxins, cytokinins, and nitrogen on production of rutin from callus and adventitious roots of the white mulberry tree (Morus alba L.). Plant Cell. Tissue and Organ Culture, 105, 9–19. doi:10.1007/s11240-010-9832- 3. Lee, J., and Scagel, C. F. (2010). Chicoric acid levels in commercial basil (Ocimum basilicum) and Echinacea purpurea products. J. Funct. Foods 2, 77–84. doi:10.1016/j.jff.2009.11.004. Lee, J., and Scagel, C. F. (2013). Chicoric acid: chemistry, distribution, and production. Frontiers in Chemistry, 1, 1-17. doi:10.3389/fchem.2013.00040. Legrand, G., Delporte, M., Khelifi, C., Harant, A., Vuylsteker, C., et al. (2016). Identification and characterization of five BAHD acyltransferases involved in hydroxycinnamoyl ester metabolism in chicory. Frontiers in Plant Science, 7, 1-16. doi:10.3389/fpls.2016.00741. Lepelley, M., Cheminade, G., Tremillon, N., Simkin, A., Caillet, V., et al. (2007). Chlorogenic acid synthesis in coffee: an analysis of CGA content and real-time RT-PCR expression of HCT, HQT, C3H1, and CCoAOMT1 genes during grain development in C. canephora. Plant Science, 172, 978-996. doi:10.1016/j.plantsci.2007.02.004. Li, S., Park, Y., Duraisingham, S., Strobel, F. H., Khan, N., et al. (2013). Predicting Network Activity from High Throughput Metabolomics. PLoS Computational Biology, 9. doi:10.1371/journal.pcbi.1003123. Li, Z., Lu, Y., Guo, Y., Cao, H., Wang, Q., et al. (2018). Comprehensive evaluation of untargeted metabolomics data processing software in feature detection, quantification and discriminating marker selection. Analytica Chimica Acta, 1029, 50-57. doi:10.1016/j.aca.2018.05.001.

107

Li, X., Zhang, L., Zhang, L., Yan, P., Ahammed, G. J., et al. (2019). Methyl salicylate enhances flavonoid biosynthesis in tea leaves by stimulating the phenylpropanoid pathway. Molecules, 24, 24020362. doi:10.3390/molecules24020362. Liang, L., Xu, J., Zhou, W. W., Brand, E., Chen, H. B., et al. (2018). Integrating targeted and untargeted metabolomics to investigate the processing chemistry of Polygoni multiflori radix. Frontiers in Pharmacology, 28, 934. doi:10.3389/fphar.2018.00934. Liang, N., and Kitts, D. D. (2016). Role of chlorogenic acids in controlling oxidative and inflammatory stress conditions. Nutrients, 8, 25. doi:10.3390/nu8010016. Liang, Y. C., Yang, M. T., Lin, C. J., Chang, C. L. T., and Yang, W. C. (2016). Bidens pilosa and its active compound inhibit adipogenesis and lipid accumulation via down-modulation of the C/EBP and PPARγ pathways. Scientific Reports, 6, 1-10. doi:10.1038/srep24285. Lin, S., Sung, J., and Chen, C. (2011). Effect of drying and storage conditions on caffeic acid derivatives and total phenolics of Echinacea Purpurea grown in Taiwan. Food Chemistry. 125, 226-231. doi:10.1016/j.foodchem.2010.09.006. Ling, A. P. K., Ong, S. L., and Sobri, H. (2011). Strategies in enhancing secondary metabolites production in plant cell cultures. Medicinal and Aromatic Plant Science and Biotechnology, 5, 94- 101. Linić, I., Šamec, D., Grúz, J., Vujčić Bok, V., Strnad, M., et al. (2019). Involvement of phenolic acids in short-term adaptation to salinity stress is species-specific among Brassicaceae. Plants, 8, 155. doi:10.3390/plants8060155. Liu, Q., Liu, F., Zhang, L., Niu, Y., Liu, Z.,et al. (2017). Comparison of chicoric acid , and its metabolites caffeic acid and caftaric acid : in vitro protection of biological macromolecules and inflammatory responses in BV2 microglial cells. Food Science and Human Wellness, 6, 155-166. doi:10.1016/j.fshw.2017.09.001. Liu, X., and Locasale, J. W. (2017). Metabolomics :a primer. Trends in Biochemical Sciences, 4, 1-11. doi:10.1016/j.tibs.2017.01.004. Liu, L., Zhang, J., Zheng, B., Guan, Y., Wang, L., et al. (2018). Rapid characterization of chlorogenic acids in Duhaldea nervosa based on ultra-high-performance liquid chromatography-linear trap quadropole-Orbitrap-mass spectrometry and mass spectral trees similarity filter technique. Journal of Separation Science, 41, 1764-1774. doi:10.1002/jssc.201701047. Luczkiewicz, M., Kokotkiewicz, A., and Glod, D. (2014). Plant growth regulators affect biosynthesis and accumulation profile of isoflavone phytoestrogens in high-productive in vitro cultures of Genista tinctoria. Plant Cell Tissue and Organ Culture, 118, 419-429. doi:10.1007/s11240-014- 0494-4. Lukitasari, M., Nugroho, D. A., and Widodo, N. (2018). Chlorogenic acid: the conceivable chemosensitizer leading to cancer growth suppression. Journal of Evidence-Based Integrative Medicine, 23, 1-6. doi:10.1177/2515690X18789628.

108

Ma, F., Cui, Q., and Bai, G. (2019). Combining UPLC/Q-TOF-MS/MS with biological evaluation for NF-κB inhibitors in uyghur medicine Althaea rosea flowers. Frontiers in Plant Science, 9, 1-9. doi:10.3389/fpls.2018.01975. Maanda, M., and Bhat, R. (2010). Wild vegetable use by Vhavenda in the Venda region of Limpopo Province , South Africa. International Journal of Experimental Botany, 79, 189-194. Madala, N. E., Piater, L. A., Steenkamp, P. A., and Dubery, I. A. (2014a). Multivariate statistical models of metabolomic data reveals different metabolite distribution patterns in isonitrosoacetophenone- elicited Nicotiana tabacum and Sorghum bicolor cells. SpringerPlus, 3, 1-10. doi:10.1186/2193- 1801-3-254. Madala, N. E., Tugizimana, F., and Steenkamp, P. A. (2014b). Development and optimization of an UPLC-QTOF-MS/MS method based on an in-source collision induced dissociation approach for comprehensive discrimination of chlorogenic acids isomers from momordica plant species. Journal of Analytical Methods in Chemistry, 2014. doi:10.1155/2014/650879. Maier, T. S., Kuhn, J., and Müller, C. (2010). Proposal for field sampling of plants and processing in the lab for environmental metabolic fingerprinting. Plant Methods, 6, 1-14. doi:10.1186/1746- 4811-6-6. Makola, M. M., Dubery, I. A., Koorsen, G., Steenkamp, P. A., Kabanda, M. M., et al. (2016). The effect of geometrical isomerism of 3,5-dicaffeoylquinic acid on its binding affinity to hiv-integrase enzyme: a molecular docking study. Evidence-based Complementary and Alternative Medicine, 2016, 1-10. doi:10.1155/2016/4138263. Malik, S., Zia, M., Rehman, R., and Chaudhary, F. (2007). In vitro plant regeneration from direct and indirect organogenesis of Memordica charantia. Pakistan Journal of Biological Sciences, 10, 4118- 4122. doi:10.3923/pjbs.2007.4118.4122. Mandal, S. M., Chakraborty, D., and Dey, S. (2010). Phenolic acids act as signaling molecules in plant- microbe symbioses. Plant Signaling and Behavior, 5, 359-368. doi:10.4161/psb.5.4.10871. Marco-Ramell, A., Palau-Rodriguez, M., Alay, A., Tulipani, S., Urpi-Sarda, M., et al. (2018). Evaluation and comparison of bioinformatic tools for the enrichment analysis of metabolomics data. BMC Bioinformatics, 19, 1-11. doi:10.1186/s12859-017-2006-0. Marzouk, M. M., Hussein, S. R., Elkhateeb, A., El-shabrawy, M., Abdel-Hameed, E. S. S., et al. (2018). Comparative study of Mentha species growing wild in Egypt: LC-ESI-MS analysis and chemosystematic significance. Journal of Applied Pharmaceutical Science, 8, 116-122. doi:10.7324/JAPS.2018.8816. Masike, K., Dubery, I., Steenkamp, P., Smit, E., and Madala, E. (2018). Revising reverse-phase chromatographic behavior for efficient differentiation of both positional and geometrical isomers of dicaffeoylquinic acids. Journal of Analytical Methods in Chemistry, 2018, 1-11. doi:10.1155/2018/8694579. Masike, K., Mhlongo, M. I., Mudau, S. P., Nobela, O., Ncube, E. N., et al. (2017). Highlighting mass

109

spectrometric fragmentation differences and similarities between hydroxycinnamoyl-quinic acids and hydroxycinnamoyl-isocitric acids. Chemistry Central Journal, 11,-1–7. doi:10.1186/s13065- 017-0262-8. Mathesius, U. (2018). Flavonoid functions in plants and their interactions with other organisms. Plants, 7, 7-9. doi:10.3390/plants7020030. Matsuda, F., Shinbo, Y., Oikawa, A., Hirai, M. Y., Flehn, O., et al. (2009). Assessment of metabolome annotation quality: A method for evaluating the false discovery rate of elemental composition searches. PLoS ONE, 4, 10-15. doi:10.1371/journal.pone.0007490. Meinhart, A. D., Damin, F. M., Caldeirão, L., da Silveira, T. F. F., Filho, J. T., et al. (2017). Chlorogenic acid isomer contents in 100 plants commercialized in Brazil. Food Research International, 99, 522-530. doi:10.1016/j.foodres.2017.06.017. Mendoza, D., Cuaspud, O., Arias, J. P., Ruiz, O., and Arias, M. (2018). Effect of salicylic acid and methyl jasmonate in the production of phenolic compounds in plant cell suspension cultures of Thevetia peruviana. Biotechnology Reports, 19, e00273. doi:10.1016/j.btre.2018.e00273 Meng, S., Cao, J., Feng, Q., Peng, J., and Hu, Y. (2013). Roles of chlorogenic acid on regulating glucose and lipids metabolism: a review. Evidence-based Complementary and Alternative Medicine, 2013, e801457. doi:10.1155/2013/801457. Mhlongo, M. I., Piater, L. A., Steenkamp, P. A., Madala, N. E., and Dubery, I. A. (2015). Metabolomic fingerprinting of primed tobacco cells provide the first evidence for the biological origin of cis- chlorogenic acid. Biotechnology Letters, 37, 205-209. doi:10.1007/s10529-014-1645-8. Mhlongo, M. I., Steenkamp, P. A., Piater, L. A., Madala, N. E., and Dubery, I. A. (2016). Profiling of altered metabolomic states in Nicotiana tabacum cells induced by priming agents. Frontiers in Plant Science, 11, e0151350. doi:10.3389/fpls.2016.01527. Mondolot, L., La Fisca, P., Buatois, B., Talansier, E., De Kochko, A., et al. (2006). Evolution in caffeoylquinic acid content and histolocalization during Coffea canephora leaf development. Annals of Botany, 98, 33–40. doi:10.1093/aob/mcl080. Morton, J. F. (1962). Spanish needles (Bidens pilosa L.) as a wild food resource. Economic Botany, 16, 173-179. doi:10.1007/BF02860036. Moubayidin, L., Mambro, R. Di, and Sabatini, S. (2009). Cytokinin - auxin crosstalk. Trends Plant Science, 14, 557-562. doi:10.1016/j.tplants.2009.06.010. Mulabagal, V., and Tsay, H.-S. (2004). Plant cell cultures - an alternative and efficient source for the production of biologically important secondary metabolites. International Journal of Applied Science and Engineering, 2, 29-48. doi:10.1016/S0168-1656(98)00195-3. Murashige T., and Skoog F. (1962). A revised medium for rapid growth and bio assays with tobacco tissue cultures. Physiologia Planturum, 15, 473–497. doi:10.1111/j.1399-3054.1962.tb08052.x Murata, M., Nishimura, M., Murai, N., Haruta, M., and Itoh, Y. (2014). A transgenic apple callus showing reduced polyphenol oxidase activity and lower browning potential. Bioscience,

110

Biotechnology, and Biochemistry, 65, 383-388. doi:10.1271/bbb.65.383. Murthy, H. N., Lee, E. J., and Paek, K. Y. (2014). Production of secondary metabolites from cell and organ cultures: Strategies and approaches for biomass improvement and metabolite accumulation. Plant Cell. Tissue and Organ Culture, 118, 1-16. doi:10.1007/s11240-014-0467-7. Mushtaq, M. Y., Choi, Y. H., Verpoorte, R., and Wilson, E. G. (2014). Extraction for metabolomics: access to the metabolome. Phytochemical Analysis, 25, 291–306. doi:10.1002/pca.2505. Ncube, E. N., Mhlongo, M. I., Piater, L. A., Steenkamp, P. A., Dubery, I. A., et al. (2014). Analyses of chlorogenic acids and related cinnamic acid derivatives from Nicotiana tabacum tissues with the aid of UPLC-QTOF-MS/MS based on the in-source collision-induced dissociation method. Chemistry Central Journal, 8, 1-10. doi:10.1186/s13065-014-0066-z. Nobela, O. (2018). ‘Phytochemical characterization of an underutilized green leafy vegetable, Sonchus oleraceus with the aid of liquid chromatography ion trap - time of flight - mass spectrometry (LCIT-TOF-MS)’ Masters dissertation. University of Johannesburg. Johannesburg, South Africa. Nomura, T., Ogita, S., and Kato, Y. (2018). Rational metabolic-flow switching for the production of exogenous secondary metabolites in bamboo suspension cells. Scientific Reports, 8, 1-11. doi:10.1038/s41598-018-31566-4. Noumedem., A.C.N., Wabo P.J., Kumar, K.N., Mohana, K.D., Mittal, G., et al. (2017). In vitro antiplasmodial activity and cytotoxicity of extracts and fractions of Bidens pilosa. Asian Journal of Biomedical and Pharmaceutical Sciences, 7, 28-34. Nugroho, L. H., Verberne, M. C., and Verpoorte, R. (2002). Activities of enzymes involved in the phenylpropanoid pathway in constitutively salicylic acid-producing tobacco plants. Plant Physiology and Biochemistry, 40, 755-760. doi:10.1016/S0981-9428(02)01437-7. Nüsslein, B., Kurzmann, M., Bauer, R., and Kreis, W. (2000). Enzymatic degradation of cichoric acid in Echinacea purpurea preparations. Journal of Natural Products, 63, 1615-1618. doi:10.1021/np0002839. Ochoa-Villarreal, M., Howat, S., Hong, S. M., Jang, M. O., Jin, Y. W., et al. (2016). Plant cell culture strategies for the production of natural products. BMB Reports, 49, 149–158. doi:10.5483/BMBRep.2016.49.3.264. Odhav, B., Beekrum, S., Akula, U., and Baijnath, H. (2007). Preliminary assessment of nutritional value of traditional leafy vegetables in KwaZulu-Natal, South Africa. Journal of Food Composition and Analysis, 20, 430-435. doi:10.1016/j.jfca.2006.04.015. Ong, K. W., Hsu, A., and Tan, B. K. H. (2013). Anti-diabetic and anti-lipidemic effects of chlorogenic acid are mediated by AMPK activation. Biochemical Pharmacology, 85, 1341–1351. doi:10.1016/j.bcp.2013.02.008. Owoyemi, O. O., and Oladunmoye, M. K. (2017). Phytochemical screening and antibacterial activities of Bidens pilosa L. and Tridax procumbens L. on skin pathogens. International Journal of Modern Biological Research, 8, 24-46.

111

Ozcan, T., Akpinar-Bayizit, A., Yilmaz-Ersan, L., and Delikanli, B. (2014). Phenolics in Human Health. International Journal of Chemical Engineering and Applications, 5, 393-396. doi:10.7763/IJCEA.2014.V5.416. Papazian, S., Girdwood, T., Wessels, B. A., Poelman, E. H., Dicke, et al. (2019). Leaf metabolic signatures induced by real and simulated herbivory in black mustard (Brassica nigra). Metabolomics, 15, 1-16. doi:10.1007/s11306-019-1592-4. Park, C. H., Yeo, H. J., Park, Y. J., Morgan, A. M. A., Arasu, M. V., et al. (2017). Influence of indole- 3-acetic acid and gibberellic acid on phenylpropanoid accumulation in common buckwheat (Fagopyrum esculentum Moench) Sprouts. Molecules, 22, e374 doi:10.3390/molecules22030374. Payyavula, R. S., Shakya, R., Sengoda, V. G., Munyaneza, J. E., Swamy, P., et al. (2015). Synthesis and regulation of chlorogenic acid in potato: rerouting phenylpropanoid flux in HQT-silenced lines. Plant Biotechnology Journal, 13, 551-564. doi:10.1111/pbi.12280. Pernisova, M., Klima, P., Horak, J., Valkova, M., Malbeck, J., et al. (2009). Cytokinins modulate auxin- induced organogenesis in plants via regulation of the auxin efflux. Proceedings of the National Academy of Sciences, 106, 3609-3614. doi:10.1073/pnas.0811539106. Piasecka, A., Kachlicki, P., and Stobiecki, M. (2019). Analytical methods for detection of plant metabolomes changes in response to biotic and abiotic stresses. International Journal of Molecular Sciences, 20, 379. doi:10.3390/ijms20020379. Pischke, M. S., Huttlin, E. L., Hegeman, A. D., and Sussman, M. R. (2006). A transcriptome-based characterization of habituation in plant tissue culture. Plant Physiology, 140, 1255-1278. doi:10.1104/pp.105.076059.Habituation. Plazas, M., Andújar, I., Vilanova, S., Hurtado, M., Gramazio, P., et al. (2013). Breeding for chlorogenic acid content in eggplant: interest and prospects. Notulae Botanicae Horti Agrobotanici Cluj- Napoca, 41, 26-35. Qu, J., Zhang, W., and Yu, X. (2011). A combination of elicitation and precursor feeding leads to increased anthocyanin synthesis in cell suspension cultures of Vitis vinifera. Plant Cell Tissue and Organ Culture, 107, 261-269. doi:10.1007/s11240-011-9977-8. Ramakrishna, A., and Ravishankar, G. A. (2011). Influence of abiotic stress signals on secondary metabolites in plants. Plant Signaling and Behavior, 6, 1720-1731. doi:10.4161/psb.6.11.17613. Ramulifho, E., Goche, T., As, J. Van, Tsilo, T. J., et al. (2019). Establishment and characterization of callus and cell suspension cultures of selected Sorghum bicolor ( L .) Moench varieties: A resource for gene discovery in plant stress biology. Agronomy, 9, 1-18. doi:doi:10.3390/agronomy905021. Randhir, R., Lin, Y. T., and Shetty, K. (2004). Stimulation of phenolics, antioxidant and antimicrobial activities in dark germinated mung bean sprouts in response to peptide and phytochemical elicitors. Process Biochemistry, 39, 637-646. doi:10.1016/S0032-9592(03)00197-3. Rao, S. R., and Ravishankar, G. A. (2002). Plant cell cultures: Chemical factories of secondary metabolites. Biotechnology Advances, 20, 101-153. doi:10.1016/S0734-9750(02)00007-1.

112

Repka, V., Fischerová, I., and Šilhárová, K. (2004). Methyl jasmonate is a potent elicitor of multiple defense responses in grapevine leaves and cell-suspension cultures. Biologia Plantarum, 48, 273- 283. doi:10.1023/B:BIOP.0000033456.27521.e5. Ribbenstedt, A., Ziarrusta, H., and Benskin, J. P. (2018). Development, characterization and comparisons of targeted and non-targeted metabolomics methods. PLoS ONE, 13, 1-18. doi:10.1371/journal.pone.0207082. Robbins, R. J. (2003). Phenolic acids in foods: An overview of analytical methodology. Journal of Agricultural and Food Chemistry, 51, 2866-2887. doi:10.1021/jf026182t. Roberts, L. D., Souza, A. L., Gerszten, R. E., and Clish, C. B. (2012). Targeted metabolomics. Current Protocols in Molecular Biology, 30, 1-34. doi:10.1002/0471142727.mb3002s98. Robinson, W. E., Reinecke, M. G., Abdel-Malek, S., Jia, Q., and Chow, S. A. (1996). Inhibitors of HIV- 1 replication that inhibit HIV integrase. Proceedings of the National Academy of Sciences of the United States of America, 93, 6326-6331. doi:10.2307/39304. Rodriguez, M. Z., Comin, C. H., Casanova, D., Bruno, O. M., Amancio, D. R., et al. (2019). Clustering algorithms: A comparative approach. PLoS One 14, 1-34. doi:10.1371/journal.pone.0210236. Roessner, U., and Bacic, A. (2009). Metabolomics in plant research. Australian Biochemist, 40, 9-12. Rosato, A., Tenori, L., Cascante, M., De Atauri Carulla, P. R., Martins dos Santos, V. A. P., et al (2018). From correlation to causation: analysis of metabolomics data using systems biology approaches. Metabolomics, 14, 1-20. doi:10.1007/s11306-018-1335-y. Rui, W., Xia, W., Zhao, W., Li, B., Li, J., et al. (2019). Differential constituents in roots, stems and leaves of Polygonum multiflorum Thunb. screened by UPLC / ESI-Q-TOF-MS and multivariate statistical analysis. Journal of Chromatographic Science, 2019, 1-8. doi:10.1093/chromsci/bmz086. Saccenti, E., Hoefsloot, H. C. J., Smilde, A. K., Westerhuis, J. A., and Hendriks, M. M. W. B. (2014). Reflections on univariate and multivariate analysis of metabolomics data. Metabolomics, 10, 361- 374. doi:10.1007/s11306-013-0598-6. Santos, R. B., Abranches, R., Fischer, R., Sack, M., and Holland, T. (2016). Putting the spotlight back on plant suspension cultures. Frontiers in Plant Science, 7, 1-12. doi:10.3389/fpls.2016.00297. Sato, Y., Itagaki, S., Kurokawa, T., Ogura, J., Kobayashi, M., et al. (2011). In vitro and in vivo antioxidant properties of chlorogenic acid and caffeic acid. International Journal of Pharmaceutics, 403, 136-138. doi:10.1016/j.ijpharm.2010.09.035. Sauerschnig, C., Doppler, M., Bueschl, C., and Schuhmacher, R. (2018). Methanol generates numerous artifacts during sample extraction and storage of extracts in metabolomics research. Metabolites, 8, 1-19. doi:10.3390/metabo8010001. Sawada, Y., Akiyama, K., Sakata, A., Kuwahara, A., Otsuki, H., et al. (2009). Widely targeted metabolomics based on large-scale MS/MS data for elucidating metabolite accumulation patterns in plants. Plant and Cell Physiology, 50, 37-47. doi:10.1093/pcp/pcn183.

113

Schaller, G. E., Bishopp, A., and Kieber, J. J. (2015). The yin-yang of hormones: cytokinin and auxin interactions in plant development. Plant Cell Online, 27, 44-63. doi:10.1105/tpc.114.133595. Schrimpe-Rutledge, A. C., Codreanu, S. G., Sherrod, S. D., and Mclean, J. A. (2016). Untargeted metabolomics strategies-challenges and emerging directions. Journal of the American Society for Mass Spectrometry, 27, 1897-1905. doi:10.1007/s13361-016-1469-y. Schütz, W., and Rave, G. (1999). The effect of cold stratification and light on the seed germination of temperate sedges (Carex) from various habitats and implications for regenerative strategies. Plant Ecology, 144, 215-230. doi:10.1023/A:1009892004730. Scossa, F., Brotman, Y., de Abreu E Lima, F., Willmitzer, L., Nikoloski, Z., et al. (2016). Genomics- based strategies for the use of natural variation in the improvement of crop metabolism. Plant Science, 242, 47-64. doi:10.1016/j.plantsci.2015.05.021. Sello, S., Moscatiello, R., La Rocca, N., Baldan, B., and Navazio, L. (2017). A rapid and efficient method to obtain photosynthetic cell suspension cultures of Arabidopsis thaliana. Frontiers in Plant Science, 8, 1-8. doi:10.3389/fpls.2017.01444. Semenya, S., Maroyi, A., Potgieter, M., and Erasmus, L. (2013). Herbal medicines used by Bapedi traditional healers to treat reproductive ailments in the Limpopo province, South Africa. African Journal of Traditional, Complementary and Alternative Medicines, 10, 331-339. doi:10.4314/ajtcam.v10i2.19. Shilpa, K., Varun, K., and Lakshmi, B. S. (2010). An alternate method of natural drug production: Elciting secondary metabolite production using plant cell culture. Journal of Plant Sciences, 5, 222-247. doi:10.3923/jps.2010.222.247. Silva, F. L., Fischer, D. C. H., Tavares, J. F., Silva, M. S., De Athayde-Filho, P. F., et al. (2011). Compilation of secondary metabolites from Bidens pilosa L. Molecules 16, 1070-1102. doi:10.3390/molecules16021070. Singh, G., Passsari, A. K., Singh, P., Leo, V. V., Subbarayan, S., et al. (2017). Pharmacological potential of Bidens pilosa L. and determination of bioactive compounds using UHPLC-QqQLIT-MS/MS and GC/MS. BMC Complementary and Alternative Medicine, 17, 1-16. doi:10.1186/s12906-017- 2000-0. Siqueira, C. F. D. Q., Cabral, D. L. V., Peixoto Sobrinho, T. J. D. S., De Amorim, E. L. C., De Melo, J. G., et al. (2012). Levels of tannins and flavonoids in medicinal plants: Evaluating bioprospecting strategies. Evidence-based Complementary and Alternative Medicine, 2012, 7. doi:10.1155/2012/434782. Sivanandhan, G., Selvaraj, N., Ganapathi, A., and Manickavasagam, M. (2014). Enhanced biosynthesis of withanolides by elicitation and precursor feeding in cell suspension culture of Withania somnifera (L.) dunal in shake-flask culture and bioreactor. PLoS ONE, 9, doi:10.1371/journal.pone.0104005. Skibiński, R., and Komsta, L. (2015). Optimization of data acquisition and sample preparation methods

114

for LC-MS urine metabolomic analysis. Open Chemistry, 13, 763-768. doi:10.1515/chem-2015- 0096. Sonnante, G., D’Amore, R., Blanco, E., Pierri, C. L., De Palma, M., et al. (2010). Novel hydroxycinnamoyl-coenzyme a quinate transferase genes from artichoke are involved in the synthesis of chlorogenic acid. Plant Physiology, 153, 1224-1238. doi:10.1104/pp.109.150144. Srivastava, P., Sisodia, V., and Chaturvedi, R. (2011). Effect of culture conditions on synthesis of triterpenoids in suspension cultures of Lantana camara L. Bioprocess and Biosystems Engineering, 34, 75-80. doi:10.1007/s00449-010-0448-0. Stella de Freitas, T. F., Stout, M. J., and Sant’Ana, J. (2019). Effects of exogenous methyl jasmonate and salicylic acid on rice resistance to Oebalus pugnax. Pest Management,75, 744-752. doi:10.1002/ps.5174. Stringer, K. A., McKay, R. T., Karnovsky, A., Quémerais, B., and Lacy, P. (2016). Metabolomics and its application to acute lung diseases. Frontiers in Immunology, 7, 1-22. doi:10.3389/fimmu.2016.00044. Su, Y. H., Liu, Y. B., and Zhang, X. S. (2011). Auxin-cytokinin interaction regulates meristem development. Molecular Plant, 4, 616-625. doi:10.1093/mp/ssr007. Sullivan, M. L. (2014). Perennial peanut (Arachis glabrata Benth.) leaves contain hydroxycinnamoyl- CoA:tartaric acid hydroxycinnamoyl transferase activity and accumulate hydroxycinnamoyl- tartaric acid esters. Planta, 239, 1091-1100. doi:10.1007/s00425-014-2038-x. Sumner, L.W., Amberg, A., Barrett, D., Beale, M.H., Beger, R., et al. (2007). Proposed minimum reporting standards for chemical analysis Chemical Analysis Working Group (CAWG) Metabolomics Standards Initiative (MSI). Metabolomics, 3, 211-221. doi:10.1007/s11306-007- 0082-2. Tan, S. Z., Begley, P., Mullard, G., Hollywood, K. A., and Bishop, P. N. (2016). Introduction to metabolomics and its applications in ophthalmology. Eye (Basingstoke), 30, 773-783. doi:10.1038/eye.2016.37. Taofiq, O., González-Paramás, A. M., Barreiro, M. F., Ferreira, I. C. F. R., and McPhee, D. J. (2017). Hydroxycinnamic acids and their derivatives: Cosmeceutical significance, challenges and future perspectives, a review. Molecules, 22, e281. doi:10.3390/molecules22020281. Tautenhahn, R., Cho, K., Uritboonthai, W., Zhu, Z., Patti, G. J., and Siuzdak, G. (2012a). An accelerated workflow for untargeted metabolomics using the METLIN database. Nature Biotechnology, 30, 826-828. doi:10.1038/nbt.2348. Tautenhahn, R., Patti, G. J., Rinehart, D., and Siuzdak, G. (2012b). XCMS online: a web-based platform to process untargeted metabolomic data. Analytical Chemistry, 84, 5035-5039. doi:10.1021/ac300698c. Teixeira, J., Gaspar, A., Garrido, E. M., Garrido, J., and Borges, F. (2013). Hydroxycinnamic acid antioxidants: an electrochemical overview. BioMed Research International, 2013.

115

doi:10.1155/2013/251754. Tiwari, R., and Rana, C. S. (2015). Plant secondary metabolites : a review. International Journal of Engineering Research and General Science, 3, 661-670. doi:10.5511/plantbiotechnology.14.1002a. Tomita, M., Robert, M., Kawakami, M., Sugimoto, M., and Soga, T. (2012). Bioinformatics tools for mass spectroscopy-based metabolomic data processing and analysis. Current Bioinformatics, 7, 96-108. doi:10.2174/157489312799304431. Treutter, D. (2005). Significance of flavonoids in plant resistance and enhancement of their biosynthesis. Plant Biology, 7, 581-591. doi:10.1055/s-2005-873009. Tsai, K.-L., Kao, C.-L., Hung, C.-H., Cheng, Y.-H., Lin, H.-C., et al. (2017). Chicoric acid is a potent anti-atherosclerotic ingredient by anti-oxidant action and anti-inflammation capacity. Oncotarget, 8, 29600-29612. doi:10.18632/oncotarget.16768. Tsai, Y. L., Chiu, C. C., Yi-Fu Chen, J., Chan, K. C., and Lin, S. D. (2012). Cytotoxic effects of Echinacea purpurea flower extracts and cichoric acid on human colon cancer cells through induction of apoptosis. Journal of Ethnopharmacology, 143, 914-919. doi:10.1016/j.jep.2012.08.032. Tugizimana, F., Piater, L., and Dubery, I. (2013). Plant metabolomics: a new frontier in phytochemical analysis. South African Journal of Science, 109, 18-20. doi:10.1590/sajs.2013/20120005. Tugizimana, F., Steenkamp, P. A., Piater, L. A., and Dubery, I. A. (2016). A conversation on data mining strategies in LC-MS untargeted metabolomics: Pre-processing and pre-treatment steps. Metabolites 6, 1–18. doi:10.3390/metabo6040040. Turner, J. G., Ellis, C., and Devoto, A. (2002). The jasmonate signal pathway. The Plant Cell, 14, 153- 164. doi:10.1105/tpc.000679. Tzin, V., and Galili, G. (2010). New insights into the shikimate and aromatic amino acids biosynthesis pathways in plants. Molecular Plant, 3, 956-972. doi:org/10.1093/mp/ssq048 Tzin, V., and Galili, G. (2011). The biosynthetic pathways for shikimate and aromatic amino acids in Arabidopsis thaliana. The Arabidopsis book, 8. doi:10.1199/tab.0132. Ulbrich, B., and Zenk, M. H. (1979). Partial purification and properties of hydroxycinnamoyl-CoA: quinate hydroxycinnamoyl transferase from higher plants. Phytochemistry, 18, 929-933. doi.org/10.1016/S0031-9422(00)91451-1. van der Berg, R. A., Hoofsloot, H. C., Westerhuis, J. A., Smilde, A. K., and van der Werf, M. (2006). Centering, scaling, and transformations: improving the biological information content of metabolomics data. Cancer Researc, 7, 1-15. doi:10.1186/1471-2164-7-142. Vaniya, A., and Fiehn, O. (2015). Using fragmentation trees and mass spectral trees for identifying unknown compounds in metabolomics. TrAC Trends in Analytical Chemistry, 69, 52-61. doi:10.1016/j.trac.2015.04.002. Vanzo, A., Cecotti, R., Vrhovsek, U., Torres, A. M., Mattivi, F., et al (2007). The fate of trans-caftaric

116

acid administered into the rat stomach. Journal of Agricultural and Food Chemistry, 55, 1604- 1611. doi:10.1021/jf0626819. Vargas-Tah, A., and Gosset, G. (2015). Production of cinnamic and p-hydroxycinnamic acids in engineered microbes. Frontiers in Bioengineering and Biotechnology, 3, 1-10. doi:10.3389/fbioe.2015.00116. Vasilev, N., Boccard, J., Lang, G., Grömping, U., Fischer, R., et al. (2016). Structured plant metabolomics for the simultaneous exploration of multiple factors. Scientific Reports, 6, 1-15. doi:10.1038/srep37390. Viant, M. R., Kurland, I. J., Jones, M. R., and Dunn, W. B. (2017). How close are we to complete annotation of metabolomes?. Current Opinion in Chemical Biology, 36, 64-69. doi:10.1002/cncr.27633. Vrhovšek, U. (1998). Extraction of hydroxycinnamoyltartaric acids from berries of different grape varieties. Journal of Agricultural and Food Chemistry, 46, 4203-4208. doi:10.1021/jf980461s. Wang, K., Jin, P., Cao, S., Shang, H., Yang, Z., et al. (2009). Methyl jasmonate reduces decay and enhances antioxidant capacity in Chinese bayberries. Journal of Agricultural and Food Chemistry, 57, 5809-5815. doi:10.1021/jf900914a. Wang, M., Lamers, R. J. A. N., Korthout, H. A. A. J., Van Nesselrooij, J. H. J., Witkamp, R. F., et al. (2005). Metabolomics in the context of systems biology: bridging traditional chinese medicine and molecular pharmacology. Phytotherapy Research, 19, 173-182. doi:10.1002/ptr.1624. Wang, S., Liu, L., Wang, L., Hu, Y., Zhang, W., et al. (2012). Structural characterization and identification of major constituents in jitai tablets by high-performance liquid chromatography/diode-array detection coupled with electrospray ionization tandem mass spectrometry. Molecules ,17, 10470-10493. doi:10.3390/molecules170910470. Wang, J., Qian, J., Yao, L., and Lu, Y. (2015). Enhanced production of flavonoids by methyl jasmonate elicitation in cell suspension culture of Hypericum perforatum. Bioresources and Bioprocessing, 2, 1-9. doi:10.1186/s40643-014-0033-5. Watson, D. G. (2013). A rough guide to metabolite identification using high resolution liquid chromatography mass spectrometry in metabolomic profiling in Metazoans. Computational and Structural Biotechnology Journal, 4, e201301005. doi:10.5936/csbj.201301005. Wianowska, D., and Gil, M. (2019). Recent advances in extraction and analysis procedures of natural chlorogenic acids. Phytochemistry Reviews, 18, 273-302. doi:10.1007/s11101-018-9592-y. Widemann, E., Miesch, L., Lugan, R., Holder, E., Heinrich, C., et al. (2013). The amidohydrolases iar3 and ill6 contribute to jasmonoyl-isoleucine hormone turnover and generate 12-hydroxyjasmonic acid upon wounding in Arabidopsis leaves. Journal of Biological Chemistry, 288, 31701-31714. doi: 10.1074/jbc.M113.499228. Wiklund, S., Johansson, E., Sjo, L., Shockcor, J. P., Gottfries, J., et al. (2008). Visualization of GC/TOF-MS-based metabolomics data for identification of biochemically interesting compounds

117

using OPLS class models. Analytical Chemistry, 80, 115-122. doi:10.1021/AC0713510. Worley, B., and Powers, R. (2015). Multivariate analysis in metabolomics. Current Metabolomics, 1, 92-107. doi:10.2174/2213235X11301010092. Wu, J., and Lin, L. (2002). Ultrasound-induced stress responses of panax ginseng cells : enzymatic browning and phenolics production. Biotechnology Progess, 18, 862-866. doi: 10.1021/bp0255210 Xia, J., Psychogios, N., Young, N., and Wishart, D. S. (2009). MetaboAnalyst: a web server for metabolomic data analysis and interpretation. Nucleic Acids Research, 37, 652-660. doi:10.1093/nar/gkp356. Xia, J., Sinelnikov, I. V, Han, B., and Wishart, D. S. (2015). MetaboAnalyst 3 . 0 – making metabolomics more meaningful. Nucleic Acids Research, 43, 251-257. doi:10.1093/nar/gkv380. Xia, J., and Wishart, D. S. (2011). Web-based inference of biological patterns , functions and pathways from metabolomic data using MetaboAnalyst. Nature Protocols, 6, 743-760. doi:10.1038/nprot.2011.319. Xie, C., Yu, K., Zhong, D., Yuan, T., Ye, F., et al. (2011). Investigation of isomeric transformations of chlorogenic acid in buffers and biological matrixes by ultraperformance liquid chromatography coupled with hybrid quadrupole / ion mobility / orthogonal acceleration time-of-flight mass spectrometry. Journal of Agricultural and Food Chemistry, 59, 11078–11087. doi:10.1021/jf203104k. Xuan, T. D., and Khanh, T. D. (2016). Chemistry and pharmacology of Bidens pilosa: an overview. Journal of Pharmaceutical Investigation,46, 91-132. doi:10.1007/s40005-016-0231-6. Yan, K., Cui, M., Zhao, S., Chen, X., and Tang, X. (2016). Salinity stress is beneficial to the accumulation of chlorogenic acids in honeysuckle (Lonicera japonica Thunb.). Frontiers in Plant Science, 7, 1563. doi:10.3389/fpls.2016.01563 Yang, W. C. (2014). Botanical, pharmacological, phytochemical, and toxicological aspects of the antidiabetic plant Bidens Pilosa L. Evidence-based Complementary and Alternative Medicine, 2014. doi:10.1155/2014/698617. Yoo, H., Widhalm, J. R., Qian, Y., Maeda, H., Cooper, B. R., et al. (2013). An alternative pathway contributes to phenylalanine biosynthesis in plants via a cytosolic tyrosine:phenylpyruvate aminotransferase. Nature Communications, 4, 1-11. doi:10.1038/ncomms3833. Zhang, H.-L., Dai, L.-H., Wu, Y.-H., Yu, X.-P., Zhang, Y.-Y., et al. (2014). Evaluation of hepatocytoprotective and anti-hepatitis B virus properties of cichoric acid from Cichorium intybus leaves in cell culture. Biological and Pharmaceutical Bulletin, 37, 1214-1220. doi:10.1248/bpb.b14-00137. Zhang, X., Ding, X., Ji, Y., Wang, S., Chen, Y., et al. (2018). Measurement of metabolite variations and analysis of related gene expression in Chinese liquorice (Glycyrrhiza uralensis) plants under UV-B irradiation. Science Reports, 8, 1-17. doi:10.1038/s41598-018-24284-4. Zhang, T. Q., Lian, H., Tang, H., Dolezal, K., Zhou, C. M., et al. (2015). An intrinsicmicroRNA timer

118

regulates progressive decline in shoot regenerative capacity in plants. Plant Cell, 27, 349-360. doi:10.1105/tpc.114.135186. Zhang, A., Sun, H., Wang, P., Han, Y., and Wang, X. (2012). Modern analytical techniques in metabolomics analysis. Analyst, 137, 293-300. doi:10.1039/c1an15605e. Zhang, J., Wu, M., Li, W., and Bai, G. (2017). Regulation of chlorogenic acid biosynthesis by hydroxycinnamoyl CoA quinate hydroxycinnamoyl transferase in Lonicera japonica. Plant Physiology and Biochemistry, 121, 74-79. doi:10.1016/j.plaphy.2017.10.017. Zhang, C., Yan, Q., Cheuk, W., and Wu, J. (2004). Enhancement of tanshinone production in Salvia miltiorrhiza hairy root culture by Ag+ elicitation and nutrient feeding. Planta Medica, 70, 147-151. doi:10.1055/s-2004-815492. Zhao, J., Davis, L. C., and Verpoorte, R. (2005). Elicitor signal transduction leading to production of plant secondary metabolites. Biotechnology Advances, 23, 283-333. doi:10.1016/j.biotechadv.2005.01.003. Zeiss, D. R., Mhlongo, M. I., Tugizimana, F., Steenkamp, P. A., and Dubery, I. A. (2018). Comparative metabolic phenotyping of tomato (Solanum lycopersicum) for the identification of metabolic signatures in cultivars differing in resistance to ralstonia solanacearum. International Journal of Molecular Sciences,19. doi:10.3390/ijms19092558. Zhao, Y., Wang, J., Ballevre, O., Luo, H., and Zhang, W. (2011). Antihypertensive effects and mechanisms of chlorogenic acids. Hypertension Research, 1-5. doi:10.1038/hr.2011.195. Zheng, X., Renslow, R. S., Makola, M. M., Webb, I. K., Deng, L., et al. (2017). Structural elucidation of cis/trans dicaffeoylquinic acid photoisomerization using ion mobility spectrometry-mass spectrometry. Journal of Physical Chemistry Letters, 8, 1381–1388. doi:10.1021/acs.jpclett.6b03015. Zhou, B., Xiao, F. J., Tuli, L., and Ressom, H. M. (2013). LC-MS-based metabolomics. National Institute of Health, 8, 470-481. doi:10.1039/c1mb05350g. Zhu, F., Xi, D. H., Yuan, S., Xu, F., Zhang, D. W., and Lin, H. H. (2014). Salicylic acid and jasmonic acid are essential for systemic resistance against tobacco mosaic virus in Nicotiana benthamiana. Molecular Plant-Microbe Interactions, 27, 567-577. doi:10.1094/MPMI-11-13-0349-R.

119