Using Frontier Technologies for the Quality Assurance of Medicinal Herbs

Home , Agrimonia pilosa, Alchemilla xanthochlora, Allocasuarina paradoxa, Allocasuarina striata, Angelica dahurica, Aralia chinensis

RIRDC Publication No. 11/093

RIRDCInnovation for rural Australia

Using Frontier Technologies for the Quality Assurance of Medicinal Herbs

by Associate-Professor Eddie Pang

November 2011

RIRDC Publication No. 11/093 RIRDC Project No. PRJ-000763

ISBN 978-1-74254-273-7 ISSN 1440-6845

Using Frontier Technologies for the Quality Assurance of Medicinal Herbs Publication No. 11/093 Project No. PRJ-000763

The information contained in this publication is intended for general use to assist public knowledge and discussion and to help improve the development of sustainable regions. You must not rely on any information contained in this publication without taking specialist advice relevant to your particular circumstances.

While reasonable care has been taken in preparing this publication to ensure that information is true and correct, the Commonwealth of Australia gives no assurance as to the accuracy of any information in this publication.

The Commonwealth of Australia, the Rural Industries Research and Development Corporation (RIRDC), the authors or contributors expressly disclaim, to the maximum extent permitted by law, all responsibility and liability to any person, arising directly or indirectly from any act or omission, or for any consequences of any such act or omission, made in reliance on the contents of this publication, whether or not caused by any negligence on the part of the Commonwealth of Australia, RIRDC, the authors or contributors.

The Commonwealth of Australia does not necessarily endorse the views in this publication.

This publication is copyright. Apart from any use as permitted under the Copyright Act 1968, all other rights are reserved. However, wide dissemination is encouraged. Requests and inquiries concerning reproduction and rights should be addressed to the RIRDC Publications Manager on phone 02 6271 4165.

Researcher Contact Details

A/Prof Eddie Pang Royal Melbourne Institute of Technology School of Applied Sciences PO BOX 71 BUNDOORA VIC 3083

Phone: +61 3 99257137 Fax: +61 3 99257110 Email: [email protected]

In submitting this report, the researcher has agreed to RIRDC publishing this material in its edited form.

RIRDC Contact Details

Rural Industries Research and Development Corporation Level 2, 15 National Circuit BARTON ACT 2600

PO Box 4776 KINGSTON ACT 2604

Phone: 02 6271 4100 Fax: 02 6271 4199 Email: [email protected]. Web: http://www.rirdc.gov.au

Electronically published by RIRDC in November 2011 Print-on-demand by Union Offset Printing, Canberra at www.rirdc.gov.au or phone 1300 634 313

Foreword

Ensuring the safety and efficacy of herbal medicines has increased in relevance following negative media coverage, particularly in relation to the adulteration and substitution of herbal products. Calls for the stricter regulation of the herbal industry have subsequently increased. However, to police the industry, regulators require the necessary tools for the rapid identification of herbal species and products. As many commercial herbal products are available in either powdered or shredded form, authentication by morphological and histological methods is extremely difficult, if not impractical. Chemical analysis is restricted in many circumstances as growth environment significantly affects the profile and amount of chemical compounds in herbal medicine plants. Recently, the identification of herbal products via DNA-based fingerprinting has been developed to complement the existing methods of microscopic examination and chemical analyses.

A fingerprinting microarray was developed in this three-year project to identify important medicinal species from around the world, with an emphasis on herbal species employed in Western and Chinese medicine. In this project, distinct fingerprints were generated for more than 40 herbal species from the flowering plant clades, Asterids and Rosids. Additionally, a sub-array was constructed to fingerprint different chemotypes of Salvia miltiorrhiza (Danshen). A significant advantage of the developed microarrays was their ability to fingerprint species of medicinal and non-medicinal plants which were not used in microarray construction—potentially widening their application to general plant fingerprinting. Further, employing a modular sub-array design, the fingerprinting microarray may be further expanded to include other important species from the Monocots, Core Eudicots and Magnoliids.

The existing microarrays are fully functional and with the eventual incorporation of the remaining sub- arrays will be a valuable tool for the medicinal plant industry for quality assurance and identification purposes. Once this is in place, the technology will be published in the scientific literature and a fee- for-service facility would be offered to the medicinal herb industry, i.e. for growers, manufacturers, pharmacological companies, or the Therapeutic Goods Administration—any body interested in validating the identity of medicinal herbs. It is also envisioned that this array will be useful for researchers and herb breeders interested in the genetic diversity of their breeding lines.

This project was funded from RIRDC Core Funds which are provided by the Australian Government. Additional funding and in-kind support was also provided by MediHerb, Botanical Resources Australia, Southern Cross University and RMIT University.

This report, an addition to RIRDC’s diverse range of over 2000 research publications, forms part of our Essential Oils and Plant Extracts R&D program which aims to provide the knowledge and skills base for industry to provide high, consistent and known qualities in their essential oils and plant extracts products that respond to market opportunities and enhance profitability.

Most of RIRDC’s publications are available for viewing, free downloading or purchasing online at www.rirdc.gov.au. Purchases can also be made by phoning 1300 634 313.

Craig Burns Managing Director Rural Industries Research and Development Corporation

iii

About the Author

Associate-Professor Eddie Pang obtained a Bachelor of Agricultural Science with Honours at the University of Melbourne, and subsequently a PhD in Plant Breeding from the same institution in 1992. After working at Agriculture Victoria (Department of Primary Industries) for several years as a pea geneticist, he returned to academia, firstly at the University of Melbourne, and subsequently at RMIT University, where he has lectured and conducted research into molecular plant breeding since 1998. Acknowledgments

My research team and I would like to gratefully acknowledge the funding received from the RIRDC, without which the research detailed in this report would not have been possible. We would like to acknowledge the financial assistance granted to us by MediHerb Pty Ltd, and for the invaluable advice and guidance of Dr Reg Lehman of MediHerb (later part of the Integria Healthcare Group). We would also like to gratefully acknowledge the advice, and the provision of plant materials by Dr Hans Wohlmuth of Southern Cross University and Mr Tim Groom of Botanical Resources Australia. Lastly, we would like to thank the University for the provision of a VRII grant as seed funding for this project. Abbreviations

BSA bovine serum albumin

DNA deoxyribonucleic acid

EDTA ethylenediaminetetraacetic acid gDNA genomic DNA

PCA principal component analysis

PCR polymerase chain reaction

PMT photomultiplier tube

SDA subtracted diversity array

SDS sodium dodecylsulphate

SSC standard saline citrate

SSH suppression subtraction hybridisation

USDA United States Department of Agriculture

Contents

Foreword ...... iii About the Author ...... iv Acknowledgments ...... iv Abbreviations ...... iv Executive Summary ...... vii Introduction ...... 1 Objectives ...... 2 Methodology ...... 3 Refining the fingerprinting microarray ...... 3 Construction of the Asterid sub-array ...... 5 Construction of the Rosid sub-array ...... 6 Construction of the Salvia miltiorrhiza chemotyping sub-array ...... 6 Construction of the Echinacea fingerprinting sub-array ...... 6 Results ...... 7 The Asterid sub-array ...... 7 The Rosid sub-array ...... 8 The Salvia chemotyping sub-array ...... 13 The Echinacea fingerprinting sub-array ...... 16 Implications ...... 17 Recommendations ...... 18 Appendix 1 ...... 19 References ...... 21

Tables

Table 1 Important features from hybridisation patterns of 25 Asterid species ...... 11

Table 2 The number of probes on the Rosid sub-array which were specific to a particular species...... 13

Table 3 Variety-specific sequences ...... 15

Table 4 Chemotype-specific sequences ...... 15

Figures

Figure 1 A schematic of the ‘Medicinal Plant Fingerprinting Microarray’ ...... 3

Figure 2 Subtraction suppression hybridisation (SSH) as employed by Jayasinghe et al. (2007) .... 4

Figure 3 Hierarchical clustering of the 25 Asterid species produced by their DNA hybridisation on the Asterid sub-array ...... 10

Figure 4 Hierarchical dendrogram of 14 Rosid species ...... 12

Figure 5 Hierarchical cluster analysis of 10 Salvia species ...... 15

Figure 6 Hierarchical cluster analysis of five chemotypes of S. Miltiorrhiza and S. sinica (outgroup control) ...... 15

Figure 7 Hierarchical cluster analysis of Echinacea species ...... 16

Executive Summary

What the report is about

This report details research conducted to produce a microarray-based fingerprinting system for medicinal plants. It summarises the methods used and results to date. The implications of the main findings as they relate to the application of this technology for routine fingerprinting of herbal products for the industry are discussed.

Who is the report targeted at?

This report is targeted at growers, manufacturers and pharmacological companies who may be interested in validating the identity of their medicinal herbs. Further, the technology outlined in this report may be of relevance to researchers and herb breeders interested in the genetic diversity of their breeding lines.

Background

Ensuring the safety and efficacy of herbal medicines has increased in relevance following negative media coverage, particularly in relation to the adulteration and substitution of herbal products. Calls for the stricter regulation of the herbal industry have subsequently increased. However, to police the industry, regulators require the necessary tools for the rapid identification of herbal species and products. As many commercial herbal products are available in either powdered or shredded form, authentication by morphological and histological methods is extremely difficult, if not impractical. Chemical analysis is restricted in many circumstances as growth environment significantly affects the profiles and the amount of chemical compounds in herbal medicine plants. Recently, the identification of herbal products via DNA-based fingerprinting has been developed to complement the existing methods of microscopic examination and chemical analyses.

Aims/objectives

The aim of the project was to develop a single microarray capable of fingerprinting several hundred medicinal plants of economical importance. The main beneficiaries from this research would be manufacturers and pharmacological companies who may be interested in validating the identity of their medicinal herbs.

Methods used

Suppression subtraction hybridisation (SSH) was performed on pools of DNA representing different clades of flowering plants or single genera of medicinal plants. The products from subtraction were subsequently gridded onto glass slides and used to generate hybridisation patterns for 47 species of medicinal plants from the Asterid and Rosid clades, as well as the genus Salvia. Hierarchical dendrograms were generated, and principal component analysis was performed to identify highly informative polymorphic features which may be converted into PCR-based markers.

Results/key findings

Asterid and Rosid sub-arrays were generated and tested. A high level of subtraction of common sequences (over 90 per cent) was obtained in each instance. Reproducible hybridisation-based fingerprints were generated for 22 medicinal plants from the Asterids and 15 from the Rosids. However, fingerprints from all species employed in the construction of both arrays have not yet been obtained. Separate sub-arrays for chemotyping Salvia spp. and Echinacea spp. were produced as part of a PhD project. Currently, the Salvia array has been employed to fingerprint 10 species of Salvia used for medicinal purposes around the world, as well as five regional varieties of S. miltiorrhiza

vii

(Danshen). The Echinacea array is the latest to be completed and tested, and has successfully been employed to discriminate between five different species of Echinacea and their genotypes.

Implications for relevant stakeholders

The accurate and efficient identification of herbal products is a prerequisite for quality control. Our project demonstrated that a microarray-based system may provide an accurate single-platform solution for this purpose. Currently, the costs of performing a replicated test on a single sample, including labour, is about $200, although PCR-based testing may be implemented for about $70. These costs may come down in the future as technological advancements are rapid in this area.

Recommendations

While the sub-arrays produced from the study are fully functional, the overall microarray is incomplete, as the Monocot, Core Eudicot, and Magnoliid arrays have not yet been fabricated. More research and funding are therefore required to complete the project.

viii

Introduction

More than 50 per cent of Australians use complementary medicines (including Chinese herbal remedies) and spend $2.2 billion a year (www.tdtctrade.com 2005). Of this, the use of Chinese herbal medicines accounts for 3.2 per cent (www.australianprescriber.com 2005), translating to a domestic market of approximately $70 million per year. Less than 10 per cent of herbal material used in Australia is locally produced, with China the largest supplier (Wilk & Dingle 2004). The global complementary medicine market was worth $62 billion in 1998 and growing at 10 per cent per annum (Eisenberg et al. 1998). The medicinal herb industry in Australia has expanded and gained wide recognition in recent years. It has been recognised as a market segment with significant potential for Australian exports. However, currently most growers are engaged at the cottage-industry level as a supplementary source of income. Although potential opportunities exist for industry to expand locally and internationally, its success will depend on many factors; one such factor would be the investment in quality assurance.

Ensuring the correct species identification of the herbs is a pre-requisite for assessing quality of any medicinal plant products. This should be therefore a critical step that should be addressed in any quality assurance program. The confirmation of the species provides full confidence to consumers and ensures that international market requirements are met. Failure to properly authenticate herbal raw materials is a breach of the Australian Code of Good Manufacturing Practice for Medicinal Herbs and poses serious risks to consumers (for example, potentially serious adverse reactions and treatment failure arising from the administration of substitute plant material). Therefore quality assurance must be improved urgently to promote the safe use of herbal products.

Traditionally, the authentication of medicinal herbs has relied upon morphological as well as histological inspection and on chemical analysis. As many commercial herbal products are available in either powdered or shredded form, authentication by morphological and histological method is extremely difficult, if not impractical. The chemistry within plant species is variable due to genetic, environmental, developmental and biological factors. Morphological identification depends on access to voucher materials and/or accurate literature, which do not always exist. Often conventional means of herbal authentication is time-consuming and hence expensive.

Authentication of medicinal herbal materials at the DNA level provides a perfect answer to the botanical identity of medicinal plant products, as the genetic composition of an herbal species does not vary with physical form. Array-based genotyping, when fully established, will permit botanical identification and origin to be established with higher accuracy, enhancing quality assurance. With effective implementation, hundreds of individual samples may be processed rapidly and cost- effectively. Compared with other molecular techniques, array-based genotyping is more robust, accurate, reproducible, reliable and sensitive.

In 2006 my research group at RMIT University completed work on a prototype microarray capable of generating fingerprints for a small number of medicinal species (Jayasinghe et al. 2007). This microarray was produced using a modification of an existing technique for DNA/RNA subtraction, and was only capable of discriminating between species from different families, and in certain cases between different species within the same genus. In this project, we refined the existing techniques we had developed to produce a larger, more sensitive microarray for fingerprinting important herbal species employed in Western, Aryuvedic and Chinese Medicine.

Objectives

The objectives of this project were therefore to:

1. Refine the prototype DNA fingerprinting microarray developed for Chinese medicinal herbs, expanding its functionality to include other herbs important to the Australian herbal industry, e.g. Ayurvedic, Western and Pan-American herbs.

2. Develop a database of DNA fingerprints for economically important species and chemotypes of medicinal herbs. Using the enhanced microarray, we plan to fingerprint over 20 species of medicinal herbs and their chemotypes. Generated molecular fingerprints and other relevant information will be stored in a comprehensive database.

Methodology

Refining the fingerprinting microarray

The original prototype microarray employed a replicated grid of approximately 300 spots (probes) representing different fragments of DNA which were cloned after the subtraction process (Jayasinghe et al. 2007, 2009). However, this design was not fully effective for discriminating different species within the same plant family. To achieve further sensitivity, we redesigned and reorganised the new microarray into sub-arrays, where each sub-array may be produced, validated and refined independently. The final microarray would therefore be constructed from printing all the refined sub- arrays onto a single chip (Figure 1). The first sub-array allowed an unknown plant to be assigned to one of the main clades of flowering plants—we have previously identified the necessary probes from the prototype array which may perform this function (Jayasinghe et al. 2009) thus it was unnecessary for us to produce new clade-specific probes for the final, completed microarray. This project, constrained by time, funding and resources, therefore focused on developing the sub-arrays for the two largest clades of flowering plants, the Asterids and Rosids, and chemotyping sub-arrays for Salvia miltiorrhiza and Echinacea purpurea (in progress). It is hoped that future funding and resources will allow the construction of the remaining sub-arrays.

Figure 1 A schematic of the ‘Medicinal Plant Fingerprinting Microarray’

Subtraction suppression hybridisation (SSH), which was employed for the construction of the original prototype microarray, was used to produce each sub-array. The methodology is outlined in Figure 2. Instead of the angiosperm ‘A’ and the non-angiosperm ‘NA’ DNA pools (as shown in Figure 2), clade, or chemotype-specific pools were made, and then subtracted from a large pool of DNA consisting of ‘all-others’, essentially non-angiosperms + other clades. This methodology will be explained in greater detail in subsequent sections. A full listing of the species employed for the construction of the DNA pools is given in Appendix 1.

Figure 2 Subtraction suppression hybridisation (SSH) as employed by Jayasinghe et al. (2007). Reproduced with permission from Clontech.

Construction of the Asterid sub-array

A genomic DNA (gDNA) pool of 45 Asterid species representing 19 families was subtracted from a Non-Asterid gDNA pool of 107 species representing other angiosperms (Monocots, Rosids, Core Eudicots, Magnoliids and Basal Angiosperms) and non-angiosperm species. Subsequently, 283 subtracted fragments were successfully amplified from the subtraction library and spotted onto microarray slides along with potential positive controls (transcripts representing chlorophyll a/b binding protein, ribosomal RNA, RUBISCO gene and Cy3 dye) and negative controls (nested primers 1 and 2, T7 primer, SP6 primer, DMSO buffer, Cy5 dye). Successful subtraction of Asterid-specific spots was first validated using the original Asterid gDNA pool and Non-Asterid gDNA pool. Subsequently, 25 species (23 medicinal species) representing 20 families and nine orders within the Asterid clade were hybridised onto the array to reveal the sensitivity of species discrimination and identification possible with this array.

The target preparation, hybridisation, washing and scanning protocol was modified from Jayasinghe et al. (2009). Briefly, 1 μg of gDNA from each species was restriction digested with HaeIII and AluI overnight at 37oC. The digested gDNA was purified using a Qiagen PCR purification kit and successful digestion confirmed by 1% agarose gel electrophoresis. The digested gDNA was biotin labelled with a Biotin Decalabel Kit (Fermentas, Ontario, Canada) according to manufacturer’s guidelines. The labelled sample was purified with a Qiagen PCR purification kit and dried down to a volume of 16 μL in a thermocycler. Corning GAPSII coated slides were spotted with 294 spots (282 subtracted fragments and 11 controls) as mentioned above. The array was printed in 11 x 4 x 7 format with six technical replicates per spot. Printed slides were pre-hybridised at 42°C for 45 minutes in a pre-warmed solution containing 5X standard saline citrate (SSC), 0.1% sodium dodecylsulphate (SDS), 1% bovine serum albumin (BSA) and 25% formamide, rinsed with sterile MilliQ water and dried with an air gun. The biotin-labeled targets (dried to 16 μL) were added to 17.5 μL of fresh 2X hybridisation buffer (500 μL of formamide, 500 μL of 10X SSC, 20 μL of 10% SDS), 0.5 μL of 5 μg/μL Cot1 DNA (Sigma-Aldrich, St Louis, MO), 0.5 μL of 10 mg/mL Poly A (Sigma-Aldrich) and 0.5 μL of 10 mg/mL salmon sperm DNA (Sigma-Aldrich). The mixture was denatured at 100°C for 2 minutes and immediately applied onto pre-hybridised slides under a 22 × 25-mm lifter slip (Grale Scientific, Ringwood, Victoria, Australia). The slides were then placed in waterproof, humidified hybridisation chambers (Corning Incorporated Life Sciences) and incubated overnight in a 42°C water bath. After hybridisation, each slide was washed in each of these solutions for 5 minutes: 1X SSC/0.1% SDS, 1X SSC/0.1% SDS, 0.1X SSC/0.1% SDS, and 0.1X SSC. The slides were subsequently transferred to 6X SSPE-T buffer (0.9 M NaCl, 60 mM NaH2PO4, 6mM EDTA, 0.005% Triton, pH 7.4) for 5 minutes. Cy3 labelling was performed in the dark using the FluroLinkTM Streptavidin-Cy3 kit, followed by incubation for an hour in the dark and washing with three changes of 6X SSPE-T for 5 minutes each. Finally the slides were rinsed with sterile MilliQ water and dried with an air gun before scanning.

The hybridisation patterns were determined by scanning each slide in a ScanArray Express scanner (PerkinElmer, USA) at 532 nm wavelength, 50–55 PMT and 5 μm resolutions. Spot intensity values were calculated by the software and bad spots were automatically flagged taking signal to background measurements into account. Automatic flagging eliminated empty spots, negative spots (signal mean less than background mean) and ‘poor’ spots (contaminated background, ignored pixels >25 per cent, open perimeter >25 per cent, offset from expected position >60 per cent). Subsequently, spots with high intensity caused by dust particles or other artifacts were manually flagged.

All the hybridisations were performed with six technical replications (each feature printed six times on the array) for each of the two biological replications (separately synthesised target) per species tested. The spot intensity data was exported to Microsoft Excel for further analysis. Spots with a median signal to median background ratio of ≥3.0 were considered to represent positive spots, and their values were converted Log2. Spots with a median signal to median background ratio of <2.0 were converted to zero. Only those features with ‘good’ spots of consistent classification (positive or negative) in all six technical replicates of each hybridisation were accepted for the data analysis. Subsequently, the six

replicates for each spot were combined to get average Log2 median signal. Finally, the spot intensity values were compared across the two biological replications for each species tested. For the spots that were consistent in both biological replications, an average value was calculated whilst inconsistent spot values were converted to zero. Average Log2 spot intensity values for all the spots (except control spots) for all 26 species tested were compared using SPSS v. 17. A dendrogram was generated using ‘Hierarchical clustering’ function with ‘between groups linkage’ and ‘Euclidean distance’.

Construction of the Rosid sub-array

A gDNA pool of 26 Rosid species, representing 11 of the 13 recognised families within this group was subtracted from a Non-Rosid gDNA pool of 108 species representing other angiosperms (Monocots, Asterids, Core Eudicots, Magnoliids and Basal Angiosperms) and non-angiosperm species. Subsequently, 166 subtracted fragments were successfully amplified from the subtraction library and spotted onto microarray slides along with potential positive controls (transcripts representing chlorophyll a/b binding protein, ribosomal RNA, RUBISCO gene and Cy3 dye) and negative controls (Nested primers 1 and 2, T7 primer, SP6 primer, DMSO buffer, Cy5 dye). Successful subtraction of Rosid-specific spots was first validated using the original Rosid gDNA pool and Non-Rosid gDNA pool. Subsequently, 10 medicinal species representing 10 families within the Rosids were hybridised onto the array to produce their individual fingerprints. The methodology for array construction, target preparation, hybridisation, washing and scanning was as described above for the Asterid sub-array.

Construction of the Salvia miltiorrhiza chemotyping sub-array

A gDNA pool of 10 Salvia species, including five chemotypes of S. miltiorrhiza was subtracted from a Non-Salvia gDNA pool of 108 species representing other angiosperms (Monocots, Asterids, Rosids, Core Eudicots, Magnoliids and Basal Angiosperms) and non-angiosperm species. Subsequently, 285 subtracted fragments were successfully amplified from the subtraction library and spotted onto microarray slides along with potential positive controls (transcripts representing Chlorophyll a/b binding protein, ribosomal RNA, RUBISCO gene and Cy3 dye) and negative controls (Nested primers 1 and 2, T7 primer, SP6 primer, DMSO buffer, Cy5 dye). Successful subtraction of Salvia-specific spots was first validated using the original Salvia gDNA pool and Non-Salvia gDNA pool. Subsequently, all 10 Salvia species, and all S. miltiorrhiza chemotypes were hybridised onto the array to produce their individual fingerprints. The methodology for array construction, target preparation, hybridisation, washing and scanning was as described above for the Asterid sub-array.

Construction of the Echinacea fingerprinting sub-array

A gDNA pool of five Echinacea species: E. angustifolia (two genotypes), E. pallida (three genotypes), E. paradoxa (two genotypes), E. purpurea (four genotypes) and E. tennesseensis (one genotype) obtained from a USDA collection and used in a previous study (Flagel et al. 2008) was subtracted from a Non-Echinacea gDNA pool of 108 species representing other angiosperms (Monocots, Asterids, Rosids, Core Eudicots, Magnoliids and Basal Angiosperms) and non-angiosperm species. Subsequently, 285 subtracted fragments were successfully amplified from the subtraction library and spotted onto microarray slides along with potential positive controls (transcripts representing Chlorophyll a/b binding protein, ribosomal RNA, RUBISCO gene and Cy3 dye) and negative controls (Nested primers 1 and 2, T7 primer, SP6 primer, DMSO buffer, Cy5 dye). Successful subtraction of Echinacea-specific spots was first validated using the original Echinacea gDNA pool and Non- Echinacea gDNA pool. Subsequently, all Echinacea genotypes were hybridised onto the array to produce their individual fingerprints. The methodology for array construction, target preparation, hybridisation, washing and scanning was as described above for the Asterid sub-array.

Results

The original objectives of the project were to refine the existing microarray, and then to generate at least 20 unique fingerprints for important medicinal species. At the conclusion of the project, we have produced fingerprints for 47 species, and five chemotypes of S. miltiorrhiza. Continuing work to fingerprint about 10 species of Echinacea and several chemotypes of E. purpurea should be concluded by the end of 2010.

The Asterid sub-array

The 283 subtracted fragments were successfully amplified from the subtraction library and spotted onto microarray slides along with 11 positive and negative controls. The validation of subtraction revealed a nearly perfect subtraction of the 45 Asterid species gDNA from the pool of 107 Non- Asterid angiosperm and non-angiosperm species. Only one out of 283 subtracted fragments hybridised with the Non-Asterid angiosperm and non-angiosperm gDNA pool. This efficiency is considerably higher than that observed in previous subtraction of non-angiosperm gDNA from angiosperm gDNA (Jayasinghe et al. 2007). This may be chiefly attributed to the modified hybridisation, washing, scanning and stringent quality control method developed by Dr Nitin Mantri in January 2009. However, 33 out of 283 spots did not hybridise with the Asterid or Non-Asterid gDNA pools. This could either be due to ‘dilution effect’ (low frequency sequences remain undetected in complex targets) as explained by Jayasinghe et al. (2009) or because the subtracted sequences are of bad quality. The stringent hybridisation and analysis conditions used here may have eliminated these bad sequences.

Subsequently, all the 25 Asterid species tested using this array generated different hybridisation patterns allowing to successfully differentiate between them. Cornus sp. hybridised to the least number of spots (one spot) whilst Coffea arabica hybridised to the largest number of spots (80 spots). Three Asterid species, Cornus sp., Gardenia jasminoides and Lonicera japonica that were not used to make the initial gDNA pool for subtraction suppressive hybridisation also hybridised to spots on the array and produced unique fingerprints, allowing for their differentiation. Overall 142 spots out of 283 (50 per cent) revealed polymorphism between the 25 species tested. The hierarchical clustering of average median Log2 values of good spots using between groups linkage and Euclidean distance is presented in Figure 3.

Hierarchical clustering results reveal that both the Ericales species, Camellia sinensis and Impatiens sp., clustered together. Similarly, the two Solanales species Withania somnifera and Lycium barbarum clustered together. Further, five out of seven Lamiales species tested clustered together. However, even though Symphytum spp. (Boraginaceae) is placed in the Lamiales by APG II (The Angiosperm Phylogeny Group 2003), its exact position is still unclear and it has been previously reported to be an outgroup to other Lamiales (Bremer et al. 2002). Further, even though the two Lamiaceae species tested (Vitex agnus-castus and Leonurus cardiaca) clustered separately, they were closer to each other relative to the other members of the Lamiales. In some texts, V. agnus-castus is placed in the Verbenaceae rather than the Lamiaceae, thus our results support the contention that, at least at the DNA level, this species may not be typical member of the Lamiaceae.

Interesting observations were made with regard to members of Gentianales. Although Gardenia jasminoides (Rubiaceae) and Trachaelospermum jasminoides (Apocynaceae) clustered together, another member of the Rubiaceae (Coffea arabica), was clearly separated from this pair. Since this was totally unexpected, four biological replications were performed for Coffea arabica, all revealing same result. The Rubiaceae is one of the largest of the angiosperm families with ~10 000 species. The family is easily recognised, but it has a problematic and much-discussed intra-familial phylogeny and issues with classification (Bremer et al. 1999).

The hybridised spots for the 25 species tested using the array were further compared to reveal species- specific or family-specific features. In addition, a principal component analysis (PCA) was performed to reveal spots (features) that displayed the maximum amount of variation between these 25 species (figure not shown). The results are summarised in Table 1; the first 10 features that explained the maximum amount of variation from PCA analysis are listed. These features hybridised to most of the species tested using the array. They may putatively serve as Asterid-specific features because except for 5TP230, none of them hybridised to Non-Asterid gDNA. The comparison also revealed two features specific to Camellia sinensis (tea) and 10 features specific to Coffea arabica (coffee). These features did not hybridise with any other species tested using the array. More importantly, one feature 5TP104 hybridised only to tea and coffee gDNA. In fact, the signal strength from coffee hybridisation was greater than that of tea. Such features can potentially be related to the distinguishing characteristic of these plants, i.e. caffeine production. In addition, we identified features that were specific to Taraxacum officinale (Asteraceae), Achellia millefolium (Asteraceae), Leonurus cardiaca (Lamiaceae), Lonicera japonica (Caprifoliaceae), Valeriana officinalis (Valerianaceae), Angelica archangelica (Apiaceae) and Ilex paraguariensis (Aquifoliaceae). Sequencing these important features may reveal important information relating to distinguishing characteristics of these plants, this work is currently in progress.

The Rosid sub-array

The Rosid sub-array was produced by Ms Urvi Desai (Masters by Coursework) and Mr Vincent Corneille (Honours student). The 166 subtracted fragments were successfully amplified from the subtraction library and spotted onto microarray slides along with 11 positive and negative controls. The validation of subtraction revealed a good subtraction efficiency of 91 per cent, the Rosid DNA hybridised with 151 probes (91 per cent) of the 166 available, whilst the Non-Rosid DNA pool hybridised with only 16 probes (10 per cent). These results indicated that the Rosid microarray contained DNA fragments which were highly specific to members of this clade, and may therefore be effective for generating specific fingerprints for these species, and at the same time excluding Non- Rosids.

Fingerprints were generated for 10 medicinal species used in the generation of the Rosid pool as well as four medicinal species which were not—Citrus medica, Eriobotrya japonica, Sophora japonica and Trigonella foenum-graecum. Hierarchical cluster analysis (Figure 4) involving all 14 species revealed that much like the accepted phylogenetic model the dendrogram separated most of the species into the Eurosids I and Eurosids II sub-clades with the exceptions of Passiflora edulis (Malpighiales order) and Oenothera biennis (Myrtales). There was significant clustering seen between the two Rosales species, Rosa chinensis and Morus alba; however, the two Malpighiales species, Passiflora edulis and Hypericum perforatum, failed to cluster together. This may be because Passiflora edulis hybridised with significantly more probes (35) than Hypericum perforatum (16) causing significant dissimilarity between the two genotypes. The species which were not used in the construction of the microarray clustered as expected. Citrus medica clustered with Citrus reticulata, both from the Sapindales. Eriobotrya japonica, a Rosales, clustered with the other Rosales species Rosa chinensis and Morus alba. Sophora japonica and Trigonella foenum-graecum clustered with the other Fabales species: Glycyrrhiza glabra. As before there was a significant separation of the Eurosids I and Eurosids II sub- clades. These results indicate that although there were a few inconsistencies in where each species was placed, the Rosid array was able to generate good, reproducible fingerprints for the species tested.

As part of the secondary aim to this experiment the genotype data for all fourteen species was examined to identify the factors (probes) that contributed to the arrangement of the dendrogram. This included categorising the exclusive probes by which species/family/order they were unique to (Table 2). A total of 56 probes (34 per cent of the possible 166) were found to be unique to a single species. Seven species did not have any unique probes on the array whilst Trigonella foenum-graecum had the most unique probes: 21. The majority of the unique probes were found in species belonging to the Fables order. It was expected that species that are evolutionarily similar would hybridise with a common set of probes whilst species that were evolutionarily distant would share fewer common

probes. As outlined in Table 2, the Rosid array identified 13 probes that were exclusive to species within the same order. Probes that are exclusive to certain orders/families hold significant value as they may be eligible to be used as unique molecular markers, or be used in the development of rapid classification tests (Jayasinghe et.al. 2007). In order to confirm the exclusivity of these probes more species need to be hybridised with the array.

Figure 3 Hierarchical clustering of the 25 Asterid species produced by their DNA hybridisation on the Asterid sub-array. The dendrogram was generated using average linkage (between groups) and Euclidean distance. Species marked with asterisk (*) were not used in original subtraction pool.

Table 1 Important features from hybridisation patterns of 25 Asterid species Clone ID Importance

All Asterids

5TP230, 5TP235, 5TP111, 5TP286, Spots from PCA analysis revealing maximum amount of 5TP296, 5TP218, 4TP117, 5TP249, variation. These spots hybridised with most of the Asterid 5TP179, 5TP236 species. Order: Ericales Specific to Camellia sinensis (Theaceae). Did not hybridise 5TP204, 5TP228 with any other species tested. Specific to only Camellia sinensis and Coffea arabica. 5TP104 Higher signal strength in coffee than tea. Order: Gentianales 4TP143, 4TP168, 5TP115, 5TP250, Specific to Coffea arabica (Rubiaceae). Did not hybridise 5TP282, 4TP122, 5TP112, 5TP113, with any other species tested. 5TP114, 5TP248 Order: Asterales Specific to Taraxacum officinale (Asteraceae). Did not 4TP138 hybridise with any other species tested. 4TP174, 5TP162, 5TP212, 5TP281, Specific to Achellia millefolium (Asteraceae). Did not 5TP292 hybridise with any other species tested. Specific to Leonurus cardiaca (Lamiaceae). Did not 4TP131, 4TP155, 5TP274 hybridise with any other species tested. Order: Solanales Hybridised with both Solanaceae species (Withania somnifera & Lycium barbarum). Did not hybridise with any 4TP140, 4TP132, 4TP103 other species tested except Lonicera japonica (Caprifoliaceae) Order: Dipsacales Specific to Lonicera japonica (Caprifoliaceae). Did not 4TP170, 5TP110, 5TP275 hybridise with any other species tested. Specific to Valeriana officinalis (Valerianaceae). Did not 4TP173 hybridise with any other species tested. Hybridised with Lonicera japonica (Caprifoliaceae) and 5TP106 Sambucus nigra (Adoxaceae). Did not hybridise with any other species tested. Order: Apiales Specific to Angelica archangelic (Apiaceae). Did not 4TP163, 5TP135 hybridise with any other species tested. Order: Aquifoliales 4TP106, 4TP111, 4TP178, 5TP219, Specific to Ilex paraguariensis (Aquifoliaceae). Did not 5TP227 hybridise with any other species tested.

Figure 4 Hierarchical dendrogram of 14 Rosid species. The four species that were not used to create the probes have clustered into their expected positions.

Table 2 The number of probes on the Rosid sub-array which were specific to a particular species Specific to species Number of unique probes Althaea officinalis 6 Armoracia rusticana 1 Catha edulis 0 Citrus medica 0 Citrus reticulata 0 Eriobotrya japonica 0 Glycyrrhiza glabra 13 Hypericum perforatum 0 Morus alba 1 Oenothera biennis 0 Passiflora edulis 13 Sophora japonica 0 Rosa chinensis 1 Trigonella foenum-graecum 21 Total 56

The Salvia chemotyping sub-array

The subtraction efficiency obtained for Salvia-specific sequences was lower than that obtained for the Asterid and Rosid sub-arrays. Of the 285 features, 152 were found to be polymorphic between the Salvia and the Non-Salvia DNA pools. Based on these results, it may be possible to conclude that the subtraction technique was able to isolate highly polymorphic Salvia-specific DNA sequences with only 53 per cent efficiency. This was less than that obtained for the prototype subtracted diversity array (SDA) for angiosperms (Jayasinghe et al. 2007), where over 90 per cent of the features were found to be polymorphic between the angiosperm and non-angiosperm samples. Further, the Asterid and Rosid subtractions also produced subtraction efficiencies of over 90 per cent. However, a direct comparison between the Salvia subtraction and these other studies may not be valid, as the Salvia subtraction involved the subtraction of much more closely related DNA sequences than the angiosperm/non-angiosperm subtraction or the clade/non-clade subtractions.

In terms of the discrimination power of the Salvia sub-array, it was observed that 166 features (58 percent) were polymorphic among the fifteen Salvia genotypes. Previous studies with the original prototype microarray found higher polymorphic frequencies (68 per cent) when genotyping the six main clades of angiosperms (Jayasinghe et al. 2007). However, the percentage of polymorphic sequences depends on the genetic diversity of the initial pool used to develop the array (Gupta et al. 2008) and the number of clones screened (James et al. 2008), therefore it is possible that the use of over 40 genetically diverse species in the construction of the prototype array may have produced significantly more polymorphic features than the Salvia sub-array.

Hierarchical cluster analysis was performed as outlined in previous sections. Two dendrograms were produced, firstly for a comparison between different species, and secondly for the five chemotypes of S. miltiorrhiza. The dendrogram elucidating the genetic relationships between the different Salvia species showed that the first cluster included Salvia sp. native to Asia; the second cluster was formed by S. lavandulaefolia and S. officinalis (three varieties), and the third cluster was formed by the native American and African species along with S. sclarea which is native to Europe and Asia (Figure 5). The analysis therefore revealed genetic relationships consistent with the geographical origin of each Salvia sp. The fifteen genotypes tested were divided into three distinctive clusters—Cluster A

contained species native to the Mediterranean region; Cluster C grouped all Salvia from China; and Cluster B grouped the entire genotyped species native to Africa and America. This genetic correlation found between African and America Salvia has also been acknowledged in a molecular analysis performed with random amplified polymorphic DNA markers (Bruna et al. 2006). Furthermore, taxonomical studies performed by Walker and Elisens, (2001) classified into the Section Heterosphace nine species native to southern Africa, one native to central and eastern Africa, along with species native to south-western North America, including S. lyrata. In these species the upper two stamens are reduced to staminodes and a fertile posterior anther thecae is found (Walker and Elisens 2001).Therefore it seems that the morphological and molecular data corroborates the hypothesis that there is a close relationship between African and American species.

The hierarchical cluster analysis presented a clear differentiation among the chemotypes/varieties of S. Miltiorrhiza (Figure 6). The varieties from Shandong province (SDL) and Hebei province cluster together and were closely related with the variety from Henan province. Cluster 2 contained the variety from Shandong province (var. alba) and that from Shanxi province. Finally, consistent with expectation, Salvia sinica, the outgroup control was found to be distantly associated with the S. miltiorrhiza clusters.

The normalised mean data obtained from the varieties was used to perform principal-component and bivariate analyses. These analyses allowed the identification of highly informative polymorphic features (10 spots), which may be used for chemotyping purposes. The normalised mean data from these polymorphic features was correlated with the data from chemical analysis obtained from several previous studies on S. miltiorrhiza (Li et al. 2009a, b). Correlations from phenotypic agricultural traits (aerial weight, number of roots, maximum root diameter and root weight.) and hybridisation data were performed. The signal of two of the features (C11-K2) was found to be inversely correlated to all the growth parameters. Therefore, the higher the hybridisation signals the lower the root weight, root diameter, number of roots and aerial weight of the plant.

Bioactive compound concentrations were also correlated with the 10 microarray features. From the studies by Li et al. (2009a, b) root weight and root diameter was found to be inversely correlated with the content of tanshinones. This implied that plants with small roots produced higher levels of tanshinones per gram of fresh tissue. In the current study, positive linear correlations were found between the signal strength K2 and the level of cryptotanshinone (r=0.84) and tanshinone IIA (0.69). K2 may therefore be a potential molecular marker for predicting levels of cryptotanshinone and tanshinone IIA. However, further studies, including the analysis of segregating progeny populations derived from a cross between high- and low-bioactive compound producing varieties will need to be conducted to confirm the co-segregation of this DNA marker with the gene(s) responsible for the production of these compounds.

The Salvia microarray was also effective in finding species-specific features. It was possible to screen these specific sequences by calculating the variance of the signal-to-noise ratios across the species. Spots with higher variances showed higher values of signal-to-noise ratios for one or two species. Specific features for species were found for S. elegans, S. officinalis, S. sclarea, S. przewalskii, S. runcinata, S. lyrata and S. fruticosa. These species-specific features and the more polymorphic spots useful for fingerprinting species (eight spots) and chemotyping (10 spots) were sequenced and analysed using BLASTN and BLASTX (www.angis.org.au). Only six features found matched with homologous sequences in GeneBank (Tables 3 and 4).

Table 3 Variety-specific sequences Feature ID Length Putative identity E value (bp) I5 398 Forsythia europaea psaA-psbB fragment, chloroplast 1e-113 L5 407 18S rRNA gene 0.0 O14 504 chloroplast, complete genome 0.0 Table 4 Chemotype-specific sequences Feature ID Length Putative identity E value (bp) P4 526 Forsythia europaea psaA-psbB fragment, chloroplast 7e-84 P10 472 Chloroplast sequence. trnL-trnF, trnF -ndhJ intergenic 7e-22 spacer J4 (specific 407 18S rRNA gene 0.0 for S. lyrata)

Figure 5 Hierarchical cluster analysis of 10 Salvia species

Figure 6 Hierarchical cluster analysis of five chemotypes of S. miltiorrhiza, and S. sinica (outgroup control)

The Echinacea fingerprinting sub-array

The subtraction efficiency obtained for Echinacea-specific sequences was as high as that achieved for the Asterid and Rosid sub-arrays. Of the 285 features, over 95 per cent were found to be polymorphic between the Echinacea and the Non-Echinacea DNA pools. This was consistent with that obtained for the prototype SDA for angiosperms, (Jayasinghe et al. 2007) where over 90 per cent of the features were found to be polymorphic between the angiosperm and non-angiosperm samples. Comparing this result with the Salvia sub-array (where a subtraction efficiency of only 53 per cent was obtained), it may be inferred that this higher efficiency may be due to better control of the subtraction process, and also perhaps the result of the addition of higher amounts of driver-DNA than was employed for the Salvia subtraction.

Hierarchical cluster analysis was performed as outlined in previous sections. The dendrogram (Figure 7) elucidating the genetic relationships between the different Echinacea species showed that the first cluster included E. tennesseensis, E. angustifolia, E. pallida and E. paradoxa. This clustering is consistent with what was observed in a previous fingerprinting study employing sequences from three nuclear and two plastid genes (Flagel et al. 2008). The second cluster comprised only E. purpurea, again consistent with results by Flagel et al. (2008). It may be inferred from these results therefore, that the Echinacea sub-array is producing fingerprints consistent with previous work, though it must be stated that the discrimination between the different species using the current sub-array is much more overt than that obtained by Flagel et al. (2008).

Figure 7 Hierarchical cluster analysis of Echinacea species

Implications

The accurate and efficient identification of herbal products is a prerequisite for quality control. Our project demonstrated that a microarray-based system may provide an accurate single-platform solution for this purpose. At the conclusion of the project, we were able to generate reproducible DNA fingerprints for 47 species of medicinal plants, as well as five provincial varieties of Danshen (S. miltiorrhiza) and Echinacea spp. using four sub-arrays. Further, a significant finding from this project was that our microarray system is capable of generating fingerprints for species which were not in its original construction. This may mean that the existing, and future microarrays may be used more broadly, for example, to include the fingerprinting of culinary herbs.

Currently, the cost of implementing this system for routine fingerprinting is relatively high. At current prices, the cost of materials for DNA extraction, coated microarray slides, and fluorescent dyes for a replicated test of a single sample is about $50. The cost of the six hours of labour, at casual technical assistant rates, is about $150, giving a total price of about $200 per sample. As indicated in the Results section, however, we are also identifying a smaller subset of highly informative DNA features which may be employed in PCR-based fingerprinting. It may be feasible, from our current studies, to identify about 20 highly polymorphic features which may be converted into PCR-based markers for the generation of fingerprints for medicinal species in the Asterid and Rosid clades. Chemotypes and species of Salvia may already be effectively differentiated using a smaller subset of about 10 features. If these features may be converted to PCR markers, then the costs may be reduced to about $20 for the materials (10 x 2 replicated PCRs x $1 per PCR using, for example, the latest PHIRE® kit.) and about $50 for labour, with an estimated final cost of $70–100 per sample. It is worth mentioning that instrumentation-heavy high-throughput PCR systems are available that significantly reduce the cost of each analysis, however, the initial outlay for such devices is very high, and may not be easily recouped unless thousands of samples are analysed weekly.

Microarray technology and high-throughput PCR systems are normally not part of the infrastructure and skill-set for most industry partners in the herbal industry. For routine testing, therefore, the technology developed by this project should be implemented by either a third party, e.g. DPI Victoria, or through RMIT University. In the first instance, however, much more research is required to complete the microarray system, and to produce a more comprehensive database of fingerprints. The recommendations for future research are outlined in the next section.

Recommendations

At present, the ‘Medicinal Plant Fingerprinting Microarray’ is only partly completed, and in its present state, cannot be used for routine fingerprinting of all medicinal plants. In 2010, further funding for this research will be sought from a variety of sources to complete the fingerprinting microarray. In particular, the Monocot, Core Eudicot and Magnoliid sub-arrays will need to be fabricated and tested. Once this is accomplished, the most informative features from each sub-array will be integrated and printed onto a single array. This final product, capable of fingerprinting at least 200–300 medicinal species and chemotypes will then be used to generate a set of definitive fingerprints for each of the species currently employed in the construction of the array, and also other closely related species not used in its construction. Further, multiple samples of herbal raw materials from economically important species should also be fingerprinted to obtain baseline variability information for each species. A database of fingerprints will thus be generated, and in collaboration with the bioinformaticists at RMIT University, we will produce a software package which will allow for the rapid retrieval of this data. The software will also allow the future input of new fingerprints. Concurrently, we will convert the most informative features into PCR-based markers, as this may be a more cost-effective means of implementing DNA-based fingerprinting for the herbal industry. Finally, in the future, RMIT University is interested in providing such a service to the industry, contingent upon receiving the necessary accreditation from the Therapeutic Goods Administration.

Appendix 1

The plant species employed in the construction of the microarrays. Where medicinal species were not available, a non-medicinal species was used to represent a certain plant family to ‘fill-in’ the gaps in the DNA content.

CLADE SPECIES MAGNOLIIDS Cinnamomum verum Magnolia denudata & Houttuynia cordata Peumus boldus BASAL ANGIOSPERMS Illicium anisatum Nymphae caerulea (6 species) Acorus calamus Iris versicolor MONOCOTS Acorus lepot Lilium longiflorum (23 species) Aloe vera Lomandra longiflora Bambusa beecheycimus Ophiopogon japonicus Belamcanda chinensis Pinellia cordata Bletilla striata Polygonatum multiflorum Coix lacryma-jobi Ruscus aculeatus Colocasia esculenta Serenoa repens Curcuma longa Trachycarpus fortunei Dioscorea batatas Zea mays Fritillaria thunbergia Zephyranthes sp. (hybrid) Zingiber officinale Adiantum raddianumv Hymenophyton flabellatum NON ANGIOSPERMS Azolla filicuiloides Lopidium coninnum (26 species) Blechnum chambersi Marchantia polymorpha Blechnum fluviatile Microsorum divesifolium Bryum billardierei Polystichum proliferum Catagonium nitens Racopilum convolutaceum Cyathea cooperi Riccardia eriocaula Cyathophorum sp Selaginella sp. Dawsonia superba Sphagnum australe Dicksonia antarctica Sticherus tener Equisetum hymale Thuidium farfurosum Ginkgo biloba Weymouthia cochlearifolia Grammitis billardieri Wollemia nobilis Aconitum carmichaelii Dianthus superbus CORE EUDICOTS Aquilegia vulgaris Eschscholzia californica (19 species) Buxus sempervirens Grevillea robusta Chelidonium majus Gypsophila oldhamii Clematis hexapetala Hamamelis virginiana Clematis montana Mahonia fortunei Clematis serratifolia Mahonia japonica Clematis songarica Phytolacca acinosa Dianthus caryophyllus Ranunculus sp. Rumex crispus Abutilon avicennae Firmiana simplex ROSIDS Agrimonia eupatoria Pelargonium x hortorum (33 species) Agrimonia pilosa Glycyrrhiza glabra Albizzia julibrissin Glycyrrhiza uralensis Alchemilla xanthochlora Gynostemma pentaphyllum Althaea officinalis Humulus lupulus Armoracia rusticana Hypericum perforatum

Astragalus membranaceus Isatis tinctoria Baptisia tinctoria Oenothera biennis Allocasuarina verticillata Oenothera odorata Catha edulis Oxalis pes-caprae Citrus aurantium Passiflora edulis Citrus reticulata Poncirus trifolia Crataegus monogyna Rosa chinensis Dichroa febrifuga Ruta graveolens Filipendula ulmaria Sophora flavescens Urtica dioica Achillea millefolium Eupatorium perfoliatum ASTERIDS Adenophora potaninii Ilex paraguariensis (45 species) Angelica archangelica Impatiens sp. Angelica dahurica Inula helenium Aralia chinensis Lycium barbarum Artemisia abrotanum Physalis peruviana Artemisia absinthium Plantago major Artemisia lactiflora Platycodon apoyama Artemisia pontica Platycodon grandiflorum Atropa belladonna Pyrethrum tanacetum Bacopa monnieri Sambucus nigra Camellia sinensis Scrophularia ningpoensis Centella asiatica Scrophularia nodosa Chamaemelum nobile Solidago canadensis Trachaelospermum Symphytum officinale jasminoides Syringa vulgaris Clerodendrum trichotomum Tanacetum parthenium Codonopsis mollis Taraxacum officinale Codonopsis pilosula Tussilago farfara Coffea arabica Valeriana officinalis Cynara scolymus Verbascum thapsus Digitalis purpurea Withania somnifera Digitalis purpurea S. officinalis (five plants) S. przewalskii SALVIA SPP. S. lyrata S. miltiorrhiza (five plants (10 species) S. elegans from five different S. sclarea provinces): S. mexicana Shandong province S. runcinata Shanxi province S. lavandulaefolia Henan province S. sinica (six plants) Hebei province Zhejiang province

References

The Angiosperm Phylogeny Group 2003, ‘An update of the Angiosperm Phylogeny Group classification for the orders and families of flowering plants: APG II’, Botanical Journal of the Linnean Society, vol. 141, pp. 399–436. Bruna, S, Giovannini, A, Benedetti, LD, Principato, MC & Ruffoni, B 2006, ‘Molecular analysis of Salvia spp. through RAPD markers’, ISHS Acta Horticulturae, vol. 723, pp. 157–160. Bremer, B, Jansen, RK, Oxelman, B, Backlund, M, Lantz, H & Kim, KJ 1999, ‘More characters or more taxa for a robust phylogeny—case study from the coffee family (Rubiaceae)’, Systematic Biology, vol. 48, p. 413. Bremer, B, Bremer, K, Heidari, N, Erixon, P, Olmstead, RG, Anderberg, AA, Källersjö, M & Barkhordarian, E 2002, ‘Phylogenetics of asterids based on 3 coding and 3 non-coding chloroplast DNA markers and the utility of non-coding DNA at higher taxonomic levels’, Molecular Phylogenetics and Evolution, vol. 24, pp. 274–301. Eisenberg, DM, Davis, RB, Ettner, SL, Appel, S, Wilkey, S, Van Rompay, M & Kessler, RC 1998, ‘Trends in alternative medicine use in the United States, 1990–1997: results of a follow-up national survey’, Journal of the American Medical Association, vol. 280, no. 18, pp. 1569–75. Flagel, LE, Rapp, RA, Grover, CE, Wirdrlechner, MP, Hawkins, J, Grafenberg, JL, Alvarez, I, Chung, GY & Wendel, JF 2008, ‘Phylogenetic, morphological, and chemotaxonomic incongruence in the North American endemic genus Echinacea’, American Journal of Botany, vol. 95, no. 6, pp. 756–65. Gupta, PK, Rustgi, S & Mir, RR 2008, ‘Array-based high-throughput DNA markers for crop improvement’, HEREDITY, vol. 101, pp. 5–18. James, KE, Schneider, H, Ansell, SW, Evers, M, Robba, L, Uszynski, G, Pedersen, N, Newton, AE, Russell, SJ, Vogel, JC & Kilian, A 2008, ‘Diversity arrays technology (DArT) for pan-genomic evolutionary studies of non-model organisms’, PLoS One, vol. 3, e1682. Jayasinghe, R, Kong, S, Coram, TE, Kaganovitch, J, Xue, CCL, Li, CG & Pang, ECK 2007, ‘Construction and validation of a novel microarray for efficient high-throughput genotyping of Angiosperms’, Plant Biotechnology Journal, vol. 5, no. 2, pp. 282–89 Jayasinghe, R, Niu, LH, Coram, TE, Kong, S, Kaganovitch, J, Xue, CC, Li, CG & Pang, ECK 2009, ‘Effectiveness of an innovative prototype subtracted diversity array (SDA) for fingerprinting plant species of medicinal importance’, Planta medica, vol. 75, pp. 1180–85. Li, CG, Sheng, S-J, Pang, ECK, Marriott P, May, B, Zhou SF, Story, D & Xue, CCL 2009a, ‘Cultivar variations of Australia-grown Danshen (Salvia miltiorrhiza): bioactive markers and root yields’, Chemistry and Biodiversity, vol. 6, pp. 170–81. Li, CG, Sheng, S-J, Pang, ECK, May, B & Xue, CCL 2009b, ‘HPLC profiles and biomarker contents of Australian-grown Salvia miltiorrhiza f. alba roots’, Chemistry & Biodiversity, vol. 6, no. 6, pp. 1077–86. PHIRE 2010, www.finnzymes.com/pcr/phire_hot_start_dna_polymerase.html. Walker, JB & Elisens, WJ 2001, ‘A revision of Salvia section Heterosphace (Lamiaceae) in Western North America’, Sida, vol. 19, pp. 571–589. Wilk, P & Dingle, W 2004, Proceedings of 3rd National Herb, Native Foods and Essential Oils Convention, 14-16 August 2003, RIRDC publication no. 04/059, Rural Industries Research and Development Corporation, Canberra.

21 Using Frontier Technologies for the Quality Assurance of Medicinal Herbs

by Associate-Professor Eddie Pang

Publication No. 11/093

Authentication of herbal medicines and products is critical for RIRDC is a partnership between government and industry consumer confidence. This report describes a fingerprinting to invest in R&D for more productive and sustainable rural microarray that was developed during a three year project, to industries. We invest in new and emerging rural industries, a accurately identify important medicinal species and products suite of established rural industries and national rural issues. from around the world. Most of the information we produce can be downloaded for free This report is targeted at growers, manufacturers and or purchased from our website . pharmacological companies who may be interested in validating the identity of their medicinal herbs. Further, the technology RIRDC books can also be purchased by phoning outlined in this report may be of relevance to researchers and 1300 634 313 for a local call fee. herb breeders interested in the genetic diversity of their breeding lines.

Contact RIRDC: Level 2 15 National Circuit Ph: 02 6271 4100 Most RIRDC publications can be viewed and purchased at Barton ACT 2600 Fax: 02 6271 4199 our website: Email: [email protected] PO Box 4776 web: www.rirdc.gov.au www.rirdc.gov.au Kingston ACT 2604 Bookshop: 1300 634 313

RIRDCInnovation for rural Australia