Head-to-Head Comparison of Ultra-High-Performance Liquid

Jan 23, 2016 - The assays were applied to industrial S. marianum extracts (AC samples) and to ... Thao L. Nguyen , Duma Hlangothi , Mahmoud A. Saleh...
0 downloads 0 Views 829KB Size
Subscriber access provided by NEW MEXICO STATE UNIV

Article

Head to head comparison of uHPLC-DAD versus qNMR for the quantitative analysis of the silymarin complex in Silybum marianum fruit extracts Antigoni Cheilari, Sonja Sturm, Daniel Intelmann, Christoph Seger, and Hermann Stuppner J. Agric. Food Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jafc.5b05494 • Publication Date (Web): 23 Jan 2016 Downloaded from http://pubs.acs.org on January 25, 2016

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Journal of Agricultural and Food Chemistry is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 28

Journal of Agricultural and Food Chemistry

Head to Head Comparison of uHPLC-DAD versus qNMR for the Quantitative Analysis of the Silymarin Complex in Silybum marianum Fruit Extracts

Antigoni Cheilari1, Sonja Sturm1*, Daniel Intelmann2, Christoph Seger1, Hermann Stuppner1

Affiliation Institute of Pharmacy, Department of Pharmacognosy, CCB – Centrum of Chemistry and Biomedicine, University of Innsbruck, Innrain 80/82, 6020 Innsbruck, Austria Bionorica research GmbH, Mitterweg 24, 6020 Innsbruck, Austria

* Correspondence: Dr. Sonja Sturm, Institute of Pharmacy, Department of Pharmacognosy, CCB – Centrum of Chemistry and Biomedicine, University of Innsbruck, Innrain 80/82, 6020 Innsbruck, Austria Email: [email protected], Tel: 0043 512 507 58408, Fax: 0043 512 507 58499

ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

1

Abstract

2

Quantitative NMR spectroscopy (qNMR) is known as excellent alternative to chromatography based

3

mixture analysis. NMR spectroscopy is a non-destructive method, needs only limited sample

4

preparation and can be readily automated. A head-to-head comparison of qNMR to an uHPLC-DAD

5

based quantitative analysis of six flavonolignan congeners (silychristin, silydianin, silybin A, silybin B,

6

isosilybin A, isosilybin B) of the Silybum marianum silymarin-complex is presented. Both assays

7

showed similar performance characteristics (linear range, accuracy, precision, LOQs) with analysis

8

times below 30 min / sample. The assays were applied to industrial S. marianum extracts (AC

9

samples) and to extracts locally prepared from S. marianum fruits (PL samples). Assay comparison by

10

Bland-Altman plots (relative method bias AC samples = -0.1%, 2S range ± 5.1%, relative method bias

11

PL samples = -0.3%, 2S range ± 7.8%) and Passing-Bablok regression analysis (slope and intercept for

12

AC and PL samples not significantly different from 1.00 and 0.00, respectively; Spearman’s coefficient

13

of rank correlation >0.99) did show, that qNMR and uHPLC-DAD can be used interchangeably to

14

quantitate flavonolignans in the silymarin complex.

15 16 17 18 19 20

Key words

21

Silybum marianum, Asteraceae, silymarin, flavonolignans, uHPLC-DAD, qNMR, quantitation

22

ACS Paragon Plus Environment

Page 2 of 28

Page 3 of 28

23

Journal of Agricultural and Food Chemistry

Introduction

24 25

NMR spectroscopy based analysis of mixtures is an appealing alternative to chromatographic analysis

26

– both in the qualitative (metabolic profiling)1,2 and the quantitative (qNMR) context.3 The analytical

27

concept of qNMR, usually performed using 1H NMR resonances, can be considered an untargeted

28

approach (any analyte combination can be quantified if appropriate undisturbed signals can be

29

identified) and offers some advantages over chromatographic analyses.4 It is non-destructive, sample

30

preparation is of limited complexity and generally compatible to chromatography, can be automated,

31

and provides – without the need of chromatographic separation or additional analytical devices

32

(hyphenation) – both qualitative and quantitative analyte information. Aside quantitative

33

information derived from carefully selected single NMR signals the complete 1H NMR spectrum of

34

the targeted compound allows a deep insight into its chemical structure. A set of 1H NMR resonances

35

can be utilized to confirm the identity of a targeted entity in a mixture – e.g. a methine 1H-NMR

36

resonance is used for quantitation and the methyl group 1H-NMR signal pattern is used for

37

qualification. In addition, an unbiased and holistic view of the total sample composition (alike the

38

total ion current in LC-MS) is obtained and allows checking for sample adulterations, unusual signal

39

pattern, solvent residuals and many other features.4,5,6

40 41

From a practical point of view, quantitative and qualitative phytochemical analysis in research,

42

production, and quality control is always constrained by limited availability of resources including

43

sample material, access to instrumentation, and investigation time. Hence replacing technologies

44

with low sample throughput, high personnel demand, and extensive sample workup requirements by

45

alternative approaches is a central demand in such processes. Due to the advantages of NMR

46

spectroscopy over chromatographic assays as listed above qNMR can be understood as such an

47

alternative. However, although both qNMR and HPLC-UV are well established and frequently applied

48

technologies in this context, head-to-head comparisons of quantitative assays have hardly been

49

performed.

50 51

The group around Verpoorte7 did report on artemisinin quantitation in eight samples by qNMR,

52

HPLC-ELSD, HPLC-MS, and TLC regarding qNMR as gold standard. They reported qNMR easy to use

53

and found all methods quantitatively comparable. Pieri et al. did compare NMR derived quantitative

54

results to findings from chromatographic assays in comprehensive Thymus vulgaris metabolite

55

profiling,8 in a targeted analysis of cynaropicrin from artichoke leaf extracts,9 and in an investigation

56

devoted to quantitate steviol glycosides from Stevia rebaudiana.10 Pauli and co-workers did devote a

57

recent publication to the analysis of green tea catechins by qNMR and LC-MS/MS.11 These authors

ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

58

used the qNMR/HiFSA (1H-NMR iterative full spin analysis) approach4 for NMR signal selection and

59

validated their approach against the NIST reference material SRM3255. Seven analytes were

60

quantified in two extracts and the SRM with LC-MS/MS and qNMR results showing high congruency.

61 62

Milk thistle (Silybum marianum (L) GAERTN.) preparations have been used for centuries to treat a

63

variety of illnesses, particularly liver maladies.12 The effective biological activities of the plant are

64

credited to silymarin, an isomeric mixture of at least six flavonolignans (Figure 1), namely silychristin

65

(1), silydianin (2), silybin A (3), silybin B (4), isosilybin A (5) and isosilybin B (6), and additionally

66

taxifolin, a flavonoid. The fruits contain up to 9% of those compounds, which mainly accumulate in

67

their external cover.13,14 The hepatoprotective activity of silymarin can be explained by the

68

antioxidant properties due to the phenolic nature of flavonolignans, stimulating liver cells

69

regeneration and cell membrane stabilization to prevent hepatotoxic agents from entering

70

hepatocytes.15 Numerous studies have demonstrated the efficacy of silymarin as a hepatoprotective,

71

but also as an anti-inflammatory, antioxidant, antifibrotic and cancer chemopreventive agent.15-18

72 73

In herbal supplement preparations the amount of silymarin in S. marianum extracts varies from 55-

74

85% with the exact composition still not completely unravelled.19 Up to now a number of HPLC-UV

75

methods have been published to support quality control measures; some however with either rather

76

long analysis times (30 min – 100 min)20-26 or limited validation data presentation.21,26 In addition LC-

77

MS/MS instrumentation was used to characterize S. marianum extract constituents27-29 or to follow

78

them in bio-analytical relevant matrices like plasma.30

79 80

With these reports at hands it was envisioned to perform a head-to-head comparison of an HPLC

81

platform with qNMR. Since there is still the need to prove the equivalence of qNMR to the

82

phytoanalytical “gold-standard” methods due to its inherent orthogonality to HPLC, the

83

establishment of a thoroughly validated HPLC method was considered as prerequisite in such

84

approach and not as an “add-on” to prove the validity on NMR derived quantitative data.

85

Consequently, it was decided to establish a state-of-the-art HPLC assay utilizing a modern stationary

86

phase. Due to the chosen research object (the S. marianum derived silymarin complex) MS/MS

87

detection was deemed unnecessary and UV/VIS detection was seen sufficient.

88 89

Materials and Methods

90

Chemicals and Reference Compounds

91

Methanol and formic acid were of HPLC grade and purchased from Merck (Darmstadt, Germany) and

92

Sigma-Aldrich (Vienna, Austria) respectively. Methanol-d4 (99.90% containing 0.03% TMS) was from

ACS Paragon Plus Environment

Page 4 of 28

Page 5 of 28

Journal of Agricultural and Food Chemistry

93

Euriso-Top (Saint-Aubin, France). Water for HPLC was produced on site using an Arium611UV

94

(Sartorius Stedim Biotech, Göttingen, Germany) water purification system. Reference compounds 1 -

95

6 were purchased from PhytoLab (Vestenbergsgreuth, Germany) and anthracene from Sigma Aldrich

96

(Vienna, Austria). Purity of all standard compounds was ≥98.5% (determined by HPLC and NMR).

97 98

Plant material and sample preparation

99

Bionorica Research GmbH (Innsbruck, Austria) provided dry S. marianum fruit extracts (extraction

100

solvent acetone, AC) and dry S. marianum fruits (PL). Voucher specimens of these materials were

101

deposited at the Institute of Pharmacy/Pharmacognosy. AC samples were weighted (10.0 ± 0.1 mg)

102

into 1.5 ml polyethylene microcentrifuge tubes (Eppendorf, Hamburg, Germany). 1 mL methanol-d4

103

containing anthracene (0.8 mg/mL) as internal standard (IS) was added and after vigorous mixing (1

104

min) on a Vortex mixer (VWR, Vienna, Austria) samples were incubated at 24°C under mild agitation

105

overnight to complete solvatization. After centrifugation (10 min at 24 oC and 20817 g) 650 μL of the

106

supernatants was transferred to NMR tubes, 300 µl to HPLC vials. Each sample was prepared in

107

triplicate. All samples were stored at 4°C until analysis. PL samples were ground and 100.0 ± 0.1 mg

108

were extracted five times with 1 mL methanol in the Ultra Sonic bath for 15 minutes. After each

109

extraction cycle, samples were centrifuged as given for the AC samples. Centrifugation supernatants

110

were combined, evaporated, and redissolved in 2 mL of methanol-d4 fortified with the IS (0.8

111

mg/mL). After centrifugation (5 min at 24 oC and 20817 g) 650 μL of the supernatants was transferred

112

to NMR tubes, 300 µl to HPLC vials. Each sample was prepared in triplicate. All samples were stored

113

at 4°C until analysis.

114 115

NMR experimental parameters

116

NMR spectra were acquired at 300 K with a Bruker Avance II 600 spectrometer (Bruker Biospin,

117

Rheinstetten, Germany) equipped with a 5 mm Prodigy Cryo-probehead with Z-gradient.

118

Experiments were performed in automation mode, using a Bruker BACS-60 sample changer operated

119

by Bruker IconNMR. Data acquisition and processing were done with Bruker Topspin 3.1. 1H-NMR

120

spectra were acquired using the Bruker zg0pr pulse program with the following settings: relaxation

121

delay (d1) = 15 s; flip angle = 45o; acquisition time = 2.99 s; FID (free induction decay) data points = 64

122

K; spectral width = 20 ppm; number of scans = 96. For experiments using presaturation, the

123

transmitter offset was manually set in order to achieve optimal suppression of the residual water

124

signal. In all cases the acquired FIDs were Fourier transformed to yield spectra with 128 K data points

125

(zero filling). Manual phase correction and automatic polynomial baseline correction were always

126

used. Chemical shift values were referenced to the calibration standard (TMS) signal. Inversion-

127

recovery experiments were performed using the Bruker t1ir pulse program, with standard acquisition

ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

128

parameters, setting the τ (tau) parameter to 0.01, 0.05, 0.1, 0.25, 0.5, 1, 2, 4, 8, and 15 seconds. T1

129

(longitudinal relaxation) values were calculated using the T1 relaxation routine (Topspin 3.1).

130 131

HPLC-DAD and LC-MS conditions

132

HPLC analyses were performed with a HP1260 system equipped with binary pump, auto sampler,

133

column thermostat, and photodiode array detector (Agilent, Waldbronn, Germany). An Agilent

134

Poroshell 120 EC-C18 column (3.0 x 50 mm, 2.7 μm particle size) was used as stationary phase and

135

water (A) / methanol (B), both containing 0.1% formic acid, as mobile phase solvents. Separation of

136

the analytes (1 µl injection volume) was achieved by applying a solvent gradient of 0 min 80% A; 3

137

min 70% A; 6 min 60% A; 8 min 60% A; 12 min 55% A; 13 min 2% A, 17 min 2% A delivered at a flow

138

rate of 0.6 ml/min. The gradient program was followed by a re-equilibration period of 10 min at 80%

139

A. The column oven temperature was set to 50°C, the detection wavelength to 286 nm. HPLC-MS/MS

140

experiments for HPLC-DAD peak assignment confirmation were performed on an amaZon SL ion-trap

141

mass spectrometer (Bruker-Daltonics, Bremen, Germany) hyphenated to the Agilent HP1260 system

142

operated under the given chromatography conditions. MS parameters: ESI alternating mode; spray

143

voltage = 4.5 kV; dry gas = 8 L/min at 250°C; nebulizer = 30 psi; full scan mode = m/z 100-1500.

144

MS/MS experiments were performed in the automatic mode.

145 146

Calibration and Validation

147

Both uHPLC and NMR methods were validated according to the ICH guidelines ‘’Validation of

148

Analytical Procedures: Text and Methodology Q2 (R1)’’31 for linearity, limit of detection and

149

quantitation, peak purity, accuracy, precision and repeatability. Standard stock solutions were

150

prepared by dissolving compounds 1 – 6 in methanol-d4 containing the internal standard (IS) (0.8

151

mg/mL). From these stock solutions nine calibrator levels (2000 μg/mL, 1000 μg/mL, 700 μg/mL, 500

152

μg/mL, 310 μg/mL, 200 μg/mL, 45 μg/mL, 25 μg/mL, 10 μg/mL) were prepared. Calibration curves

153

were obtained by plotting the peak areas of analytes 1 – 6 versus the concentrations of 1 – 6 in

154

uHPLC and by plotting the respective ratios of areas and concentrations of 1 – 6 and IS in NMR. The

155

regression parameters intercept, slope and correlation coefficient (R2) were calculated by linear

156

regression analysis. The limit of detection (LOD) and the limit of quantitation (LOQ) for each analyte

157

were calculated from the regression models including only the lowest three dilution levels. From the

158

obtained regression equation data the LOD was calculated as three times the standard deviation (SD)

159

of the y-intercept divided by the slope, whereas the LOQ was calculated as ten times the SD of the y-

160

intercept divided by the slope. Accuracy was determined by recovery experiments spiking a dry

161

extract (AC 01) and a fruit batch (PL 01) with known amounts of selected analytes prior to sample

162

workup. All samples were prepared in triplicate. By analysing the variability (triplicate analysis) of all

ACS Paragon Plus Environment

Page 6 of 28

Page 7 of 28

Journal of Agricultural and Food Chemistry

163

recovery samples on three consecutive days intra-day and inter-day precision of the assays was

164

assessed. Quantitative results have not been corrected for residual traces of moisture or solvents.

165 166

Statistics

167

Data analysis for validation and calculation of analyte concentrations was done within Microsoft

168

Excel 2011 (Redmond, WA, USA) spread sheets. Statistical data analysis (comparative Bland-Altman

169

plots, Passing Bablok regression, rank correlation analysis) was done with MedCalc for Windows,

170

version 15.8 (MedCalc Software, Ostend, Belgium)

171 172

Results and Discussion

173

Sample preparation for NMR and HPLC

174

Two sample types were used in this investigation. Beside industry based plant extracts (extraction

175

solvent acetone, AC) typically used as formulation basis in commercial milk thistle medications (>50%

176

silymarin), laboratory prepared fruit extracts (PL) were additionally investigated to introduce broader

177

sample variability into the investigation. Due to the distinctively different matrix properties, tailored

178

sample preparation protocols for AC and PL had to be established. In a series of preliminary 1H-NMR

179

experiments several deuterium labeled solvents including methanol-d4, aceton-d6, DMSO-d6,

180

acetonitrile-d3, and deuterium oxide as well as mixtures thereof were tested as potential extraction

181

solvents for both AC and PL samples and as solvents for pure reference materials. The obtained

182

spectra showed that methanol-d4 was the solvent of choice enabling sufficient solubility of reference

183

compounds and extracts. A good separation of NMR key resonances and a clean baseline in the

184

relevant spectral regions was obtained. Anthracene was chosen as internal standard for qNMR due

185

to its solubility and stability in methanol, its low volatility and its aromatic nature resulting neither in

186

1

187

were directly re-dissolved in methanol-d4, the preparation of PL extracts was optimized to get a

188

maximum of secondary metabolites into solution. Different extraction solvents (ethanol, methanol,

189

chloroform, hexane and mixtures of them with water) and extraction procedures (sonication,

190

Eppendorf shaker) were evaluated. Sonication supported quantitative extraction of the ground plant

191

material with methanol as solvent was found to be most efficient.

H NMR signal overlap with signals of 1 - 6 nor in HPLC co-elution with 1 - 6. Whereas AC extracts

192 193

Validation of quantitative NMR analysis

194

The application of quantitative 1H NMR (qNMR) requires that at least one non-overlapping signal for

195

each molecule to be quantified is available for integration.5 The research efforts were concentrated

196

to the principal components of silymarin – silybin A (1), silybin B (2), isosilybin A (3), isosilybin B (4),

197

silychristin (5), silydianin (6) (Figure 1). All compounds are isobaric (MW 482) with compounds 3-6

ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

Page 8 of 28

198

even representing isobaric pairs of diastereoisomers (3/4 and 5/6) with very similar 1H and 13C NMR

199

spectra. Silybins (3/4) and isosilybins (5/6) have the same trans-conformation in the C ring of the

200

flavonoid skeleton and opposite conformation and substitution pattern in the dioxan scaffold with

201

1

202

Pauli and co-workers have shown that applying quantum mechanical driven HiFSA even almost

203

overlapping proton signals can be used for analyte quantitation – but this approach was not

204

extended to S. marianum extracts.32 Having the complexity of an extract compared to a reference

205

compound standard mixture in mind and since the HiFSA approach depends on software only

206

commercially available; it was decided not to take up this approach but to quantitate 3,4 and 5,6 as

207

compound pairs.

H-NMR shift differences (Δδ) not exceeding 0.02 ppm at 600 MHz 1H-NMR resonance frequency.

208 209

Careful analysis of the 1H-NMR spectra of the single compounds and of mixtures in the extraction

210

solvent led to a series of methine resonances suitable for analyte quantitation in mixtures (Table 1).

211

For analyte pairs 3,4 and 5,6 aromatic doublets (JHH ≈ 2.1 Hz) between 7.08 ppm and 7.15 ppm have

212

been chosen, for 1 an OCHCH doublet at 5.55 ppm (JHH ≈ 6.3 Hz) and for 2 a broadened methine

213

singlet at 2.89 ppm (Figure 2). The peak purity of these signals was assured by the aid of homo- and

214

heteronuclear 2D NMR experiments (COSY, HSQC, HMBC), which enabled unambiguous assignment

215

of these resonances in the proton scaffold of compounds 1 – 6. The patterns of these key signals

216

were identical in pure compounds and matrix containing sample materials, an influence of

217

satellite signals in the signal pairs 3,4 and 5,6 can be excluded, since the distance between these

218

resonances does not match the anticipated 1J(13C,1H) ≈ 120 – 150 Hz.

13

C

219 220

A crucial parameter in qNMR is the relaxation delay, which enables equilibrium of the magnetization

221

between pulses when accumulating co-added FIDs. It depends on the relaxation properties of the

222

various nuclei in the mixture and is determined by inversion-recovery T1 experiments for all

223

resonances of interest. The highest T1-value was calculated to be 4.7 s for the selected signal of

224

anthracene; all Silybum constituents showed shorter T1-values between 1.54 s and 2.22 s (Table 1).

225

In order to obtain a full relaxation of all nuclei of interest a relaxation delay of 5*T1 = 23.5 s was

226

calculated for a flip angle of 90o. After optimization experiments a flip angle of 45o and a relaxation

227

delay of 15 s (acquisition time 2.99 s) were chosen and resulted in a 99.9% recovery of the internal

228

standard signal with the longest T1. The total analysis time including sample transfer, sample

229

temperature equilibration, and sample shimming was about 35 min with a net acquisition time of

230

about 29 min.

231

ACS Paragon Plus Environment

Page 9 of 28

Journal of Agricultural and Food Chemistry

232

NMR method validation covered assay linearity, limit of detection and quantitation, accuracy,

233

precision, and repeatability. Based on the calibration models, linearity of the method was confirmed

234

for the concentration range of 10 µg/mL to 2000 μg/mL for compound 2, 3, and 6 and 25 µg/mL to

235

2000 μg/mL for 1, 4, and 5 (Table 2). In these ranges the correlation coefficient of the calibration

236

function was better than 0.9986 for all analytes. Limits of detection (LOD) and limits of quantitation

237

(LOQ) determined by an independent approach were found to range from 0.5 µg/mL to 3.4 µg/mL

238

and 1.6 µg/mL to 10.4 μg/mL, respectively. The discrepancy between found LOQs and the lower limit

239

of the linear range does prove, that for analyte concentration ranges between these figures of merit

240

a separate calibration model must be implemented. Consequently in this investigation the LLOQs

241

(lower limits of quantitation) were set to the lowest calibrator concentration in the linear range. The

242

signal to noise ratios for these concentrations did exceed 20 units.

243 244

Accuracy was determined in recovery experiments with three different concentrations of standard

245

compounds spiked to AC samples or added to the plant material (PL) prior to extraction. All results

246

were within the range of 90.7% – 106.4% of the amount fortified with no observable bias between

247

the methods of the materials (Table 4). Hence for both assays the analytical analyte recovery is

248

quantitative and the LOQs established for the solvent-based calibrators can be used for the AC and

249

PL samples. Precision of the assay was determined by preparation and analysis of fortification

250

samples on three days. Intra-day precision was better than 7.7% over all analytes and materials;

251

inter-day precision was found to be better than 8.6% (Table 5). There was no precision difference

252

between the materials AC and PL, result scatter for analyte 2, present in rather low concentrations

253

(Table 6 and Table 7), was slightly increased compared to the other analytes.

254 255

Validation of quantitative uHPLC analysis

256

To allow head-to-head comparison of the silymarin complex qNMR assay with a more conventional

257

analytical technique an uHPLC-DAD assay was established. As to be expected, the quite similar

258

physicochemical properties of the diastereomeric analyte pairs 3/4 and 5/6 were the major hurdle in

259

the development of a sufficient selective method. In a first round of experiments several silica based

260

reverse phase materials (RP-12, RP-18, phenyl-, phenyl-hexyl) as stationary phases and different

261

mobile phase conditions were evaluated. The use of acetonitrile in the mobile phase was clearly not

262

appropriate, but methanol alone was also not giving satisfactory separation. Acidic mobile phase

263

additives were improving peak shape; optimal results were achieved with water to methanol

264

gradient elution using formic acid as modifier. Conventional stationary phase materials as the

265

endcapped RP-18 column (Zorbax Eclipse) were found insufficient. Even under optimized and lengthy

266

chromatographic conditions signal overlap at silydianin (2) led to a remarkable method bias with 2-3-

ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

267

fold concentrations in the chromatographic assay compared to qNMR (data not shown). Taking into

268

account the selectivity of the chosen wavelength (286 nm) and the possibility of differences in

269

UV/VIS detection response factors (molar extinction coefficients) it is likely that an unaddressed

270

silydianin isomer or silydianin degradation product33 was co-eluting with silydianin (2) in this method.

271

Only replacing the fully porous Zorbax column by a superficially porous particle stationary phase

272

(Poroshell C18) led to baseline separation of all analytes including silydianin (2). To prove satisfactory

273

analyte separation in chromatography in addition to spiking experiments and comparison of

274

retention times and UV-spectra to reference materials, HPLC-MS/MS analysis was utilized to confirm

275

the peak assignment of 2 (tR = 6.40 min) as silydianin (Figure 3) since this analyte yields a distinctive

276

fragmentation pattern in MS/MS analysis (m/z = 151 and 169). The peak eluting shortly after 2 (tR =

277

6.60 min) showed an identical mass spectrum (negative ESI base peak with m/z = 481) and a similar

278

fragmentation pattern to compound 1 with m/z = 355 as most abundant fragment ion. Thus, this

279

peak could be tentatively assigned to silychristin B, a diastereomer of silychristin (1).29 The analyte

280

peak eluting between 1 and 2 (tR = 6.20 min) showed a significant different behaviour to compound 1

281

in mass spectrometry with a negative ESI base peak at m/z = 383 and – in contrast to the other

282

analytes – a favoured ionisation in the positive ESI modus (m/z = 385). Hence this analyte is most

283

likely not a flavonolignan derivative. The HPLC-DAD assay was validated according to the ICH-

284

guidelines.31 Linearity was confirmed from 10 -1000 µg/mL for 1 and 4 and from 25-2000 µg/mL for

285

2, 3, 5 and 6, correlation coefficients were higher than 0.999 in all cases (Table 3). LODs ranged from

286

1.0 µg/mL to 6.0 µg/mL and LOQs from 2.0 µg/mL to 19.0 µg/mL respectively (S/N > 20). The results

287

of accuracy experiments were comparable to qNMR data with recovery rates from 91.8 – 110.0 %

288

(Table 4). Intra-day variations less than 7.4% and within days variability did not exceed 5.9% (Table

289

5).

290 291

Sample analysis

292

The validated uHPLC-DAD and qNMR methods were subsequently applied for the quantitation of

293

nine batches of AC extracts and ten batches of PL extracts (Tables 6 and 7). AC 01 – 09 showed

294

relatively small variations in the contents of compounds 1 – 6, with AC 02 showing a slightly higher

295

amount of silybins (3,4) and AC 07 showing a higher content in isosilybin derivatives (5,6) and

296

silychristin (1). The ratio of 3 to 4 was relatively constant in AC 01 – 09 and was determined as 1 : 1.7

297

(± 3.3%) by HPLC-DAD, the ratio of 6 to 5 was 1 : 2.8 (± 0.5%). In all cases the total content of

298

flavonolignans was 59.8 ± 2.4% of the total extract. The content of all other flavonolignans, except 5

299

and 6 was significantly decreased. PL extracts derived from plant raw material showed distinctive

300

higher concentration variations and – as to be expected – overall lower relative amounts of 1 – 6

301

(silymarin complex content 9.4 ± 2.4%). Whereas the ratio of 3 to 4 in PL 01 – 10 was with 1 : 1.6 (±

ACS Paragon Plus Environment

Page 10 of 28

Page 11 of 28

Journal of Agricultural and Food Chemistry

302

12.1%) comparable to the AC materials, the ratio of 6 to 5 was found to be significantly different with

303

1 : 1.4 (± 4.3%).

304 305

The comparison of uHPLC and qNMR data by statistical means did prove that both techniques are

306

equivalent regarding the quantitative assessment of S. marianum flavonolignan congeners present in

307

the “silymarin” compound complex. Passing-Bablok correlation analysis of qNMR (Table 6, analytes 1,

308

2, 3/4, 5/6) and uHPLC (Table 7, analytes 1, 2 and sums of 3+4 and 5+6) derived analyte

309

concentrations in AC and PL samples (separate analysis due to different concentration units) did

310

result in ideal correlation equations (Figure 4) with intercepts statistically not different from zero (AC:

311

intercept = 2.91, 95% CI = -1.02 to 18.66; PL: intercept = -4.61, 95% CI = -12.75 to 1.35) and slopes

312

statistically not different from one (AC: slope = 0.99, 95% CI = 0.98 to 1.01; PL: slope = 1.02, 95% CI =

313

1.00 to 1.07). Calculating the rank correlation coefficient (Speaman’s rho) confirmed the visually

314

impressive correlation with rho > 0.99 for both analyses (Figure 4). Relative Bland-Altman plot based

315

method comparison analysis (Figure 4) did confirm the results of the Passing-Bablok correlation

316

analysis and further unveiled that neither AC nor PL samples analysis did show any significant

317

concentration dependent bias. This does prove, that minor matrix constituents neighbouring the

318

NMR signals of compounds 1 and 5+6 (Figure 2) do not significantly perturbate the qNMR based

319

analyte quantification. The 2S confidence intervals of the AC and PL Bland-Altman plots (2S = 5.1%

320

and 7.8% respectively) were in good agreement with the inter-day RSD data gathered in the

321

validation process of the assays (Table 5) and the RSD data obtained from repeated analysis of the

322

samples (Table 6 and Table 7). No signs of additional sources of imprecision (e.g. bias related effects

323

from matrix interferences) were detected. If however, a compound-by-compound quantification is

324

desired, the presented uHPLC assay must be seen superior to the qNMR method due to partial signal

325

overlap in the latter one.

326 327

This presentation aims to prove, that qNMR and uHPLC analysis, although in principal orthogonal

328

instrumental analysis technologies, can be interchangeably used to quantitate secondary metabolites

329

from raw material (fruit extracts = PL) and enriched industry based extracts (acetone dry extracts =

330

AC). In this particular case study sample preparation protocols were identical for qNMR and uHPLC,

331

extracts solved in deuterated NMR solvents can be easily applied to chromatographic assays. Key

332

figures of merit of the assay validation – linear range, LOQ / LOD, repeatability were found to be

333

more the less identical for both assays – this is remarkably different from a recent report, were LC-

334

MS/MS analyses showed a distinctively higher RSD than qNMR.11 Both assays were applied to

335

nineteen sample batches – nine AC samples and ten PL samples. Identical quantitative results were

336

obtained, no statistical differences were found between the methods. From an economic point of

ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

337

view, both approaches are comparable, too when used in routine analysis. Whilst NMR equipment is

338

usually more expensive than chromatography instrumentation, costs per sample are lower in NMR

339

due to reduced solvent use and less need for consumables. Also measurement times were

340

comparable, although not been fully optimized both techniques are capable of running several dozen

341

samples per day. With all the care needed judging such a limited investigation the obtained results

342

provide confidence that – proper assay design and assay validation provided – the concept to use

343

(automatable) qNMR and uHPLC interchangeably for quantitative analysis of primary and secondary

344

metabolites from rather crude mixtures, e.g. from fermentation processes, from extraction plants, or

345

within raw material entry control will inspire further investigations in this topic.

346 347

Acknowledgements

348

The authors acknowledge the substantial support by Bionorica SE, Neumarkt, Germany and Bionorica

349

Research GmbH, Innsbruck, Austria.

350

ACS Paragon Plus Environment

Page 12 of 28

Page 13 of 28

351

Journal of Agricultural and Food Chemistry

References

352 353 354 355 356 357 358 359 360 361 362 363

(1) Seger, C.; Sturm S. Analytical aspects of plant metabolite profiling platforms: current standings and future aims. J. Proteome Res. 2007, 6, 480-497. (2) Holmes, E.; Tang, H.; Wang, Y.; Seger C. The assessment of plant metabolite profiles by NMRbased methodologies. Planta Med. 2006, 72, 771-785. (3) Pauli, G. F.; Gödecke, T.; Jaki, B.U.; Lankin, D. C. Quantitative 1H NMR. Development and potential of an analytical method: an update. J. Nat. Prod. 2012, 75, 834-851. (4) Simmler, C.; Napolitano, J. G.; McAlpine, J. B.; Chen, S. N.; Pauli, G. F. Universal quantitative NMR analysis of complex natural samples. Curr. Opin. Biotechnol. 2014, 25 ,51-59. (5) Pauli, G. F.; Jaki, B. U.; Lankin, D. C. Quantitative 1H NMR: development and potential of a method for natural products analysis. J. Nat. Prod. 2005, 68, 133-149. (6) Maggio, R. M.; Calvo, N. L.; Vignaduzzo, S. E.; Kaufman, T. S. Pharmaceutical impurities and

364

degradation products: uses and applications of NMR techniques. J. Pharm. Biomed. Anal. 2014,

365

101, 102-122.

366 367

(7) Liu, N. Q.; Choi, Y. H; Verpoorte, R.; van der Kooy, F. Comparative quantitative analysis of artemisinin by chromatography and qNMR. Phytochem. Anal. 2010, 21, 451-456.

368

(8) Pieri, V.; Sturm, S.; Seger, C; Franz, C.; Stuppner, H. 1H NMR-based metabolic profiling and target

369

analysis: a combined approach for the quality control of Thymus vulgaris. Metabolomics 2012, 8,

370

335-346.

371 372 373

(9) Pieri, V.; Stuppner, H. Quantification of cynaropicrin in artichoke leaf extracts by ¹H NMR spectroscopy. Planta Med, 2011, 77, 1756-1758. (10) Pieri, V.; Belancic, A.; Morales, S.; Stuppner H. Identification and quantification of major steviol

374

glycosides in Stevia rebaudiana purified extracts by 1H NMR spectroscopy. J. Agric. Food Chem.

375

2011, 59, 4378-4384

376

(11) Napolitano, J. G.; Gödecke, T.; Lankin, D. C.; Jaki, B. U.; McAlpine, J. B.; Chen, S. N.; Pauli, G. F.

377

Orthogonal analytical methods for botanical standardization: determination of green tea

378

catechins by qNMR and LC-MS/MS. J. Pharm. Biomed. Anal. 2014, 93, 59-67.

379 380 381 382 383

(12) Flora, K.; Hahn, M.; Rosen, H.; Benner, K. Milk thistle (Silybum marianum) for the therapy of liver disease. Am. J. Gastroenterol. 1998, 93, 139-143. (13) Morazzoni, P.; Bombardelli, E. Silybum marianum (Carduus marianus). Fitoterapia 1995, 66, 342. (14) Madrid, E.; Corchete, P. Silymarin secretion and its elicitation by methyl jasmonate in cell

384

cultures of Silybum marianum is mediated by phospholipase D-phosphatidic acid. J. Exp. Bot.

385

2010, 61, 747-754.

ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

386 387 388 389

(15) Fraschini, F.; Demartini, G.; Esposti, D. Pharmacology of Silymarin. Clin. Drug. Invest. 2002, 22, 51-65. (16) Abenavoli, L.; Capasso, R.; Milic, N.; Capasso F. Milk thistle in liver diseases: past, present, future. Phytother. Res. 2010, 24, 1423–1432.

390

(17) AbouZid, S. Silymarin, natural flavonolignans from milk thistle. In: Phytochemicals - A Global

391

Perspective of Their Role in Nutrition and Health; Rao, V. Ed.; InTech: Rijeka, Croatia, 2012; pp

392

255-272.

393

(18) Burczynski, F. J.; Yan, J.; Gong, Y.; Nguyen, D.; Wang, G.; Burczynski, S. D.; Smith, H. J.; Gong, Y.

394

The hepatoprotective effect of diltiazem and silymarin. Nat. Prod. Chem. Res. 2013, 1, 111.

395

(19) Kroll, D. J.; Shaw, H. S.; Oberlies, N. H. Milk thistle nomenclature: Why it matters in cancer

396 397 398 399

research and pharmacokinetic studies. Integr. Cancer Ther. 2007, 6, 110–119. (20) Kuki, A.; Nagy, L.; Deak, G.; Nagy, M., Zsuga, M.; Keki, S. Identification of Silymarin Constituents: An Improved HPLC―MS Method. Chromatographia 2012, 75, 175-180. (21) Saleh, I. A.; Vinatoru, M.; Mason, T. J.; Abdel-Azim, N. S.; Aboutabl, E. A.; Hammouda, F. M.

400

Ultrasonic-assisted extraction and conventional extraction of Silymarin from Silybum marianum

401

seeds; a comparison. Res. J. Pharm. Biol. Chem. Sci. 2015, 6, 709-717.

402 403 404

(22) Eklund, L.; Simon, J. P.; Ballenger, J. High Performance Liquid Chromatography of flavonolignans in commercial milk thistle supplements. Bios 2009, 80, 164-169. (23) Lee, J. I.; Narayan, M.; Barrett, J. S. Analysis and comparison of active constituents in commercial

405

standardized silymarin extracts by liquid chromatography-electrospray ionization mass

406

spectrometry. J. Chromatogr. B 2007, 845, 95-103.

407 408

(24) Kvasnicka, F.; Bíba, B.; Sevcík, R.; Voldrich, M.; Krátká, J. Analysis of the active components of silymarin. J. Chromatogr A. 2003, 990, 239-245.

409

(25) Omar, A.; Hadad, G.; Badr J. First detailed quantification of silymarin components in the leaves of

410

Silybum marianum cultivated in Egypt during different growth stages. Acta Chromtogr. 2012, 24,

411

463-474.

412 413 414

(26) Wianowska, D.; Wiśniewski, M. Simplified procedure of silymarin extraction from Silybum marianum L. Gaertner. J. Chromatogr. Sci. 2015, 53, 366-372. (27) Wang, K.; Zhang, H.; Shen, L.; Du, Q.; Li, J. Rapid separation and characterization of active

415

flavonolignans of Silybum marianum by ultra-performance liquid chromatography coupled with

416

electrospray tandem mass spectrometry. J. Pharm. Biomed. Anal. 2010, 53, 1053-1057.

417

(28) Lee, J. I.; Hsu, B. H.; Wu, D.; Barrett, J. S. Separation and characterization of silybin, isosilybin,

418

silydianin and silychristin in milk thistle extract by liquid chromatography-electrospray tandem

419

mass spectrometry. J. Chromatogr. A 2006, 1116, 57-68.

ACS Paragon Plus Environment

Page 14 of 28

Page 15 of 28

420

Journal of Agricultural and Food Chemistry

(29) Shibano, M.; Lin, A. S.; Itokawa, H.; Lee, K. H. Separation and characterization of active

421

flavonolignans of Silybum marianum by liquid chromatography connected with hybrid ion-trap

422

and time-of-flight mass spectrometry (LC-MS/IT-TOF). J. Nat. Prod. 2007, 70, 1424-1428.

423

(30) Brinda, B. J.; Zhu, H. J.; Markowitz, J. S. A sensitive LC-MS/MS assay for the simultaneous analysis

424

of the major active components of silymarin in human plasma. J. Chromatogr. B 2012, 902, 1-9.

425

(31) N.N. (2006) ICH Harmonized Tripartite Guideline Validation of Analytical Procedures: Text and

426

Methodology Q2(R1); http://www.ich.org/products/guidelines/quality/article/quality-

427

guidelines.html (accessed August 23, 2015)

428

(32) Napolitano, J. G.; Lankin, D. C.; Graf, T. N.; Friesen, J. B.; Chen, S. N.; McAlpine J. B.; Oberlies, N.

429

H; Pauli, G.F. HiFSA Fingerprinting applied to isomers with near-identical NMR spectra: The

430

Silybin/Isosilybin case. J. Org. Chem. 2013, 78, 2827-2839.

431 432 433

(33) Bilia, A.R.; Bergonzi, M.C.; Gallori, S.; Mazzi, G.; Vincieri, F. F. J. Pharm. Biomed. Anal. 2002, 30, 613-624. (34) Passing, H.; Bablok, W. A new biometrical procedure for testing the equality of measurements

434

from two different analytical methods. Application of linear regression procedures for method

435

comparison studies in Clinical Chemistry, Part I. J. Clin. Chem. Clin. Biochem. 1983, 21, 709-720.

436

(35) Dewitte, K.; Fierens, C.; Stöckl, D.; Thienpont, L. M. Application of the Bland-Altman plot for

437

interpretation of method-comparison studies: a critical investigation of its practice. Clin. Chem.

438

2002, 48, 799-801.

439

ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

440

Figures Legends

441 442

Figure 1: Structures of major S. marianum flavonolignans (1 - 6). 1H NMR resonances of highlighted

443

protons were used for qNMR based analyte quantitation.

444 445

Figure 2: Representative 1H-NMR spectra of selected S. marianum AC and PL extracts recorded under

446

qNMR conditions. Highlighted inserts show NMR peaks (top in sample AC03, bottom in PL01) used

447

for analyte quantitation, numbers correspond to analytes 1 – 6.

448 449

Figure 3: Representative S. marianum uHPLC-DAD chromatograms of AC and PL extracts recorded at

450

286 nm. Numbers correspond to analytes 1 – 6.

451 452

Figure 4: Passing-Bablok correlation (left-hand column) and Bland-Altman plot (right-hand column)

453

analysis of AC (top row) and PL (bottom row) samples.

454

ACS Paragon Plus Environment

Page 16 of 28

Page 17 of 28

Journal of Agricultural and Food Chemistry

Tables

Table 1: 1H-NMR signal information (methanol-d4, 600 MHz) for NMR resonances used within the qNMR experiments for silymarin flavonolignans quantitation.

Compound

δH (ppm)

multiplicity, JHH (Hz)

T1 (sec)

1

5.55

d, 6.28

1.5

2

2.89

br s, -

2.0

3

7.10

d, 2.02

2.1

4

7.08

d, 2.02

2.1

5

7.15

d, 2.08

2.2

6

7.13

d, 2.08

2.2

ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

Page 18 of 28

Table 2: qNMR based calibration function parameters for 1-6, including regression equations, correlation coefficients (R2), linearity range, LOD and LOQ.

Linearity range

LOD

(μg/mL)

(μg/mL)

0.9986

25-2000

2.9

8.8

y = 0.1476x - 0.0016

0.9998

10-2000

3.4

10.4

3

y = 0.1882x - 0.0002

0.9999

10-2000

0.6

1.8

4

y = 0.1878x - 0.0005

0.9994

25-2000

0.5

1.6

5

y = 0.1848x - 0.0004

1.0000

25-2000

2.3

7.0

6

y = 0.1626x + 0.0046

0.9997

10-2000

1.4

4.3

2

Y (peak area) = kcA+d

R

1

y = 0.1570x + 0.0189

2

Compound

LOQ (μg/mL)

ACS Paragon Plus Environment

Page 19 of 28

Journal of Agricultural and Food Chemistry

Table 3: uHPLC-DAD based calibration function parameters for 1-6, including regression equations, correlation coefficients (R2), linearity range, LOD and LOQ.

Ret. Time Compound (min)

Linearity

2

Y (peak area) = kcA+ d

R

LOD

LOQ

range (μg/mL) (μg/mL) (μg/mL)

1

5.86

y= 2374.1x – 8.4163

0.9993

10-2000

1.0

3.0

2

6.60

y= 1886.7x – 22.4990

0.9997

25-2000

4.0

12.0

3

8.38

y= 2667.6x – 26.5870

0.9994

25-2000

6.0

19.0

4

8.74

y= 2528.0x – 5.2708

0.9992

10-2000

1.0

3.0

5

10.01

y= 2710.3x – 60.6340

0.9994

25-2000

1.0

2.0

6

10.38

y= 2598.4x – 63.9100

0.9993

25-2000

1.0

3.0

ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

Page 20 of 28

Table 4: qNMR and uHPLC-DAD assay accuracy for compounds 2, 3, and 4. Inter-day recovery values (n = 9) expressed in percentage amount added (mean ± RSD).

Compound

NMR Recovery

Compound for

uHPLC Recovery

for NMR

(% ± RSD)

uHPLC

(% ± RSD)

Amount added (µg) AC 500.0 250.0

AC 104.9 ± 2.7

2

91.2 ± 6.7

98.9 ± 3.1 2

100.7 ± 6.8

100.0

90.7 ± 5.1

103.1 ± 6.1

500.0

98.4 ± 0.7

97.2 ± 2.1

250.0

3

100.0

106.4 ± 2.2 101.8 ± 0.6

PL 500.0 250.0

3

103.6 ± 1.5 PL

95.3 ± 2.2 2

98.4 ± 2.0

91.2 ± 5.4

101.3 ± 3.6 2

94.7 ± 4.5

100.0

103.6 ± 3.3

102.1 ± 4.2

500.0

103.1 ± 0.9

106.5 ± 3.0

250.0 100.0

4

105.5 ± 1.0

4

100.8 ± 1.9

ACS Paragon Plus Environment

91.8 ± 2.4 110.0 ± 0.8

Page 21 of 28

Journal of Agricultural and Food Chemistry

Table 5: qNMR and uHPLC-DAD assay precision for compounds 2, 3, and 4. Intra-day (n = 3 on each day) and inter-day (n = 9) precision for both types of extracts.

RSD (%) Compound

Day 1 Day 2 Day 3

Day 1-3

NMR 2 (AC)

7.7

3.4

6.6

8.6

3 (AC)

1.4

0.9

2.0

1.4

2 (PL)

3.5

3.5

5.2

4.0

4 (PL)

2.3

1.4

2.9

2.3

uHPLC 2 (AC)

3.3

6.0

7.4

5.3

3 (AC)

3.8

1.4

1.8

2.9

2 (PL)

3.3

5.2

2.2

5.9

4 (PL)

2.5

2.9

2.8

3.0

ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

Page 22 of 28

Table 6: Quantitative qNMR results for compounds 1 – 6 in S. marianum samples AC 01-09 and PL 0110 (RSD in parenthesis, n=3).

Samples

1

2

3+4

5+6

AC (µg / 10 mg AC material) AC 01

1388.5 (2.1)

43.1 (9.6)

3718.5 (1.5)

676.7 (1.0)

AC 02

1458.7 (0.67) 41.2 (6.8)

3854.7 (0.5)

659.3 (0.7)

AC 03

1593.8 (0.9)

46.0 (5.8)

3311.2 (0.6)

735.2 (1.7)

AC 04

1577.5 (0.9)

40.2 (6.0)

3410.4 (0.4)

729.0 (1.5)

AC 05

1666.4 (1.1)

41.0 (5.8)

3583.9 (0.8)

749.8 (3.3)

AC 06

1683.2 (1.0)

37.2 (1.7)

3619.8 (0.5)

756.5 (1.6)

AC 07

1869.9 (0.9)

39.6 (6.3)

3721.6 (0.9)

843.0 (2.5)

AC 08

1703.5 (0.9)

42.6 (5.5)

3411.3 (1.2)

778.1 (4.6)

AC 09

1752.4 (0.2)

47.5 (6.1)

3284.8 (1.8)

788.5 (0.6)

PL (µg / 100 mg PL) PL 01

372.3 (4.4)

34.0 (4.3)

860.7 (2.8)

221.5 (7.1)

PL 02

106.0 (6.5)

224.3 (6.3)

221.6 (8.2)

169.5 (5.3)

PL 03

114.3 (8.9)

169.0 (9.4)

241.1 (5.9)

181.4 (1.8)

PL 04

158.6 (9.0)

249.9 (4.2)

306.9 (1.0)

207.5 (3.2)

PL 05

144.0 (8.6)

260.0 (4.9)

311.0 (2.7)

201.9 (8.7)

PL 06

128.1 (7.1)

287.0 (6.6)

211.3 (4.6)

214.0 (3.6)

PL 07

151.0 (8.9)

301.2 (5.5)

301.4 (5.2)

201.2 (5.6)

PL 08

104.0 (2.3)

217.0 (3.7)

247.8 (6.7)

182.8 (7.2)

PL 09

148.0 (6.9)

273.2 (7.5)

281.8 (3.5)

196.4 (6.7)

PL 10

137.0 (6.5)

321.7 (6.1)

280.9 (5.5)

200.3 (7.8)

ACS Paragon Plus Environment

Page 23 of 28

Journal of Agricultural and Food Chemistry

Table 7: Quantitative uHPLC-DAD results for compounds 1-6 in S. marianum samples AC 01-09 and PL 01-10 (RSD in parenthesis, n=3) Sample

1

2

AC 01

1387.2 (1.7)

43.5 (6.9)

AC 02

1421.4 (4.2)

AC 03

3

4

3+4

5

6

5+6

AC (µg / 10 mg AC material) 1463.9 (1.9) 2293.1 (1.8) 3757.0

466.2 (1.5)

172.7 (1.8)

638.9

41.7 (7.5)

1527.1 (3.8)

2379.7 (3.9)

3906.8

477.7 (4.2)

183.8 (2.8)

661.5

1576.8 (0.8)

48.3 (3.5)

1275.8 (1.4)

2120.6 (1.2)

3396.4

528.0 (0.6)

176.7 (1.8)

704.7

AC 04

1566.5 (2.8)

41.6 (4.2)

1298.0 (2.9)

2140.1(3.0)

3438.1

527.2 (2.9)

181.8 (3.6)

709.0

AC 05

1644.0 (2.5)

40.6 (6.6)

1332.0 (2.9)

2207.0 (3.0)

3539.0

550.2 (2.7)

189.1 (3.7)

739.3

AC 06

1670.1 (5.1)

38.6 (6.2)

1389.6 (5.0)

2290.8 (5.1)

3680.4

561.8 (4.8)

193.2 (4.9)

754.8

AC 07

1868.7 (7.7)

41.6 (6.7)

1400.2 (7.4)

2378.6 (7.5)

3778.8

607.5 (7.1)

214.4 (5.4)

821.9

AC 08

1719.5 (2.1)

41.3 (3.6)

1279.1 (1.9)

2174.4 (2.0)

3453.5

559.9 (1.5)

196.4 (1.8)

756.4

AC 09

1710.1 (2.2)

48.6 (3.2)

1251.9 (2.2)

2141.4 (2.2)

3393.3

557.6 (2.1)

194.7 (2.2)

752.3

PL 01

382.1 (1.44)

30.5 (9.7)

PL (µg / 100 mg plant material) 325.8 (1.3) 536.9 (1.9) 862.8

153.3 (0.8)

70.4 (0.8)

223.7

PL 02

113.5 (7.89)

228.1 (5.4)

82.2 (5.6)

135.1 (6.2)

217.3

96.6 (4.2)

77.6 (2.8)

174.1

PL 03

118.8 (6.18)

162.1 (7.9)

98.0 (7.0)

139.0 (8.3)

237.0

105.0 (3.5)

84.0 (4.5)

189.0

PL 04

158.3 (1.91)

253.3 (5.9)

108.2 (8.6)

191.2 (1.5)

299.4

116.4 (0.6)

89.2 (0.5)

205.6

PL 05

146.1 (5.75)

243.5 (7.5)

103.3 (5.0)

190.8 (4.0)

294.1

123.5 (3.0)

89.9 (2.5)

213.4

PL 06

120.9 (8.12)

290.3 (5.4)

72.7 (8.4)

129.4 (8.2)

202.1

125.1 (4.2)

96.2 (2.9)

221.3

PL 07

157.5 (9.09)

300.7 (3.7)

107.2 (7.8)

184.6 (8.0)

291.8

120.7 (7.2)

87.5 (6.5)

208.2

PL 08

102.2 (2.69)

211.0 (8.7)

116.2 (6.3)

136.7 (3.9)

252.9

101.0 (6.2)

77.3 (6.6)

178.3

PL 09

141.0 (8.44)

264.2 (5.2)

106.0 (7.4)

170.0 (8.0)

276.0

109.0 (7.6)

86.0 (8.4)

195.0

PL 10

142.8 (8.51)

316.1 (9.7)

103.2 (9.0)

169.7 (8.6)

272.9

117.1 (4.3)

84.4 (4.8)

201.5

ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

Page 24 of 28

Figures

OH O HO

H

O

HO

OH OH O silydianin (2)

silychristin (1) H H

H O

O

CH2OH

H OCH3

O OH

H

HO

O

O

CH2OH OCH3

O

OH

OH

OH O

H

OH

OH O

silybin A (3)

silybin B (4) OH

H H HO

OCH3

H

O

OH OH O

HO

OH

HO

OCH3 CH2OH

O

O

O

O

O OH

H

OCH3 CH2OH

HO

O

O

H

OH O

OH

H

O OH

H

OH O isosilybin A (5)

isosilybin B (6)

Figure 1

ACS Paragon Plus Environment

OCH3 CH2OH

Page 25 of 28

Journal of Agricultural and Food Chemistry

3,4 2

1 5,6

7.14

7.12

7.10

7.08

5.58

5.56

5.54

5.52

2.92

2.90

2.88

2.86

AC03 IS

PL01

8

6

4

2

0

[ppm] 3,4

1

2

5,6

7.14

7.12

7.10

7.08

5.58

5.56

5.54

5.52

Figure 2

ACS Paragon Plus Environment

2.92

2.90

2.88

2.86

Journal of Agricultural and Food Chemistry

Figure 3

ACS Paragon Plus Environment

Page 26 of 28

Page 27 of 28

Journal of Agricultural and Food Chemistry

4000

(uHPLC-qNMR)/Mean of methods (%)

6

AC – samples

Analyte concentration qNMR (µg / 10 mg extract)

y = 0.99 x + 2.91; rho = 0.996

3000

2000

1000

AC - samples

2

0

Mean = - 0.1 % -2

-4

-2SD = -5.2% -6

0 0

1000

2000

3000

4000

0

1000

Analyte concentration uHPLC (µg / 10 mg extract)

2000

3000

4000

Mean of methods (µg / 10 mg extract)

(uHPLC-qNMR)/Mean of methods (%)

10

1000

PL – samples y = 1.02 x – 4.61; rho = 0.994

Analyte concentration qNMR (µg / 100 mg plant)

+2SD = 5.0%

4

800 600 400 200

PL - samples +2SD = 7.5%

5

0

Mean = - 0.3 %

-5 -2SD = -8.1% -10

0 0

200

400

600

800

1000

0

200

Analyte concentration uHPLC (µg / 100 mg plant)

Figure 4

ACS Paragon Plus Environment

400

600

Mean of methods (µg / 100 mg plant material)

800

1000

Journal of Agricultural and Food Chemistry

”TOC” graphic

ACS Paragon Plus Environment

Page 28 of 28