Ultrahigh-Performance Liquid ChromatographyâHigh-Resolution

Subscriber access provided by UNIV OF NEBRASKA - LINCOLN

Article

UHPLC-HR-QTOF-MS based metabolomics reveals key differences between Brachiaria decumbens and B. brizantha, two similar pastures with different toxicity Andy J. Pérez, Syeda M. Hussain, #ukasz Pecio, Mariusz Kowalczyk, Valdo Rodrigues, and Anna Stochmal J. Agric. Food Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jafc.6b01296 • Publication Date (Web): 18 May 2016 Downloaded from http://pubs.acs.org on May 21, 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 30

Journal of Agricultural and Food Chemistry

UHPLC-HR-QTOF-MS

based

metabolomics

reveals

key

differences between Brachiaria decumbens and B. brizantha, two similar pastures with different toxicity

Andy J. Pérez,*,† Syeda M. Hussain,‡ Łukasz Pecio,† Mariusz Kowalczyk,† Valdo Rodrigues‡ and Anna Stochmal†

†

Department of Biochemistry and Crop Quality, Institute of Soil Science and Plant Cultivation,

State Research Institute, ul. Czartoryskich 8, 24-100, Puławy, Poland. ‡

Department of Plant Sciences, College of Animal Sciences and Food Engineering, University

of Sao Paulo, Pirassununga – SP, 13635-900, Brazil.

* Corresponding author. E-mail: [email protected]. Phone: +48 81 4786 886.

1 ACS Paragon Plus Environment


1

ABSTRACT

2

Several species of Brachiaria (Poaceae) currently cover extensive

3

grazing areas in Brazil, providing valuable source of feed for a large cattle

4

population. However, numerous cases of toxicity outbreaks in livestock have

5

raised concerns on safety of using these plants, especially B. decumbens. In

6

this study, chemometric analysis of UHPLC-HR-QTOF-MS data has for the first

7

time uncovered qualitative and quantitative differences between metabolomes

8

of toxic B. decumbens and non-toxic B. brizantha. The steroidal saponin

9

protoneodioscin was established as the main biomarker for B. decumbens when

10

compared to B. brizantha, and therefore the key explanation for their

11

phytochemical differentiation. Quantification of protodioscin in both plants

12

showed no significant differences, consequently the idea that this compound is

13

solely responsible for toxicity outbreaks must be discarded. Instead, we propose

14

that the added occurrence of its stereoisomer, protoneodioscin, in B.

15

decumbens, can be considered as the probable cause of these events.

16

Interestingly, the greatest concentrations of saponins for both species were

17

reached during winter (B. decumbens = 53.6 ± 5.1 mgg-1 D.W.; B. brizantha =

18

25.0 ± 1.9 mgg-1 D:W.) and spring (B. decumbens = 49.4 ± 5.0 mgg-1 D.W.; B.

19

brizantha = 27.9 ± 1.4 mgg-1 D:W.), although in the case of B. decumbens

20

these values do not vary significantly among seasons.

21 22

KEYWORDS: Brachiaria decumbens, Brachiaria brizantha; UHPLC-MS

23

metabolomics; Multivariate data analysis; Saponin quantification.

24


Page 2 of 30

Page 3 of 30


25

INTRODUCTION

26

With the world’s biggest cattle population of over 200 million heads,

27

Brazil is currently one of the largest beef producing countries in the world.1 This

28

production is largely dependent on pasture areas due to easy availability and

29

low costs. The 85% of Brazil’s pasture areas is covered by various species of

30

Brachiaria,2 of which B. decumbens and B. brizantha are the two most

31

prominent species.

32

Along with holding large cattle population, Brazil is also facing poisoning

33

cases by plants,3,4 described in literature as one of the three main reasons

34

causing farm animal’s death.5 Brachiaria spp., especially B. decumbens, are on

35

the top of the list of such plants,4 which have been reported as the cause of

36

hepatotoxic photosensitization outbreaks among goats, sheep and cattle.3,6

37

Initially, Brachiaria spp. toxicity was associated to sporidesmin mycotoxin

38

produced by Pithomyces chartarum fungus; but later investigations suggested

39

that steroidal saponins, mainly protodioscin, contained in this plants are the root

40

of the problem by induction of crystal formation in the biliary system.7,8–10 A

41

mechanism has been proposed in which saponins are hydrolyzed in the

42

digestive tract, producing epi-smilagenin and epi-sarsapogenin. Detoxification of

43

these compounds by conjugation with glucuronic acid yields glucuronides that

44

bind calcium ions to form insoluble salt deposited in form of crystals.3 This

45

causes photosensitization in an indirect manner, by damaging either

46

hepatocytes and/or bile ducts thus disrupting the liver’s ability to excrete

47

phytoporphyrin (phylloerythrin: a chlorophyll post digestion metabolite) into the

48

gastrointestinal tract via the biliary system.11 Then, photosensitization occurs

49

due to accumulation of phytoporphyrin in livestock liver and skin, which upon



50

exposure to sunlight results in excitation of this photoactive molecule, causing

51

liver injury, jaundice, skin problems, anorexia, nervousness and death in some

52

cases.11,3

53

Majority of the poisoning outbreaks reported in Brazil have been caused

54

by B. decumbens, while intoxications involving other species such as B.

55

brizantha, are less frequent.3 For this reason, the replacement of B. decumbens

56

by the less toxic B. brizantha, have been implemented through the last 20

57

years; resulting in a decrease of outbreaks.3 The concentration of protodioscin

58

in these two species in different stages of growth have been measured by many

59

authors; however the results are contradictory, showing not significant

60

differences and even very low or undetectable levels in some cases of

61

outbreak.3,12 This calls into question not only the attribution of Brachiaria spp.

62

toxicity to protodioscin, but also the reliability of measuring methods previously

63

employed. The determination of variation between the metabolome (all

64

metabolites, i.e., small size molecules, intermediates or end-products of

65

metabolic processes present in biological system)13 of B. decumbens and B.

66

brizantha samples, is therefore needed as an overview that could help to

67

identify components potentially responsible for the toxicity.

68

Metabolomics technology has emerged as a suitable research strategy

69

for the profiling of endogenous plant metabolites;14 and it has been successfully

70

applied in the determination of metabolites who contribute most to the

71

distinction between the metabolome of genetically closed species.15,16 In the

72

present study, we applied nontargeted reverse phase UHPLC-HR-QTOF-MS

73

metabolomics approach to investigate the variation in the metabolite profiles

74

between B. decumbens and B. brizantha samples harvested during different


Page 4 of 30

Page 5 of 30


75

seasons and stages of growth. Multivariate analysis techniques, including

76

principal component analysis (PCA) and partial lest square discriminant

77

analysis (PLS-DA), were applied to the data mining. Tandem mass

78

spectrometry

79

(https://metlin.scripps.edu/),17 authentic standard compounds, and the literature

80

were used for the identification of phytochemicals which most contribute to the

81

separation of these two species. In addition, the main identified biomarker for

82

distinguishing groups was isolated, characterized, and quantified in all samples.

83

MATERIALS AND METHODS

(MS/MS),

metabolite

database

of

METLIN

84

Chemicals. Acetonitrile hypergrade for LC-MS and Methanol HPLC

85

grade were purchased from Merck (Darmstadt, Germany). Water was purified

86

in-house with a Milli-Q water purification system (Millipore Co.). Formic acid

87

MS-grade, eluent additive for LC-MS was obtained from Sigma-Aldrich.

88

Plant Material. Authenticated seeds of Brachiaria decumbens stapf (cv.

89

Basilisk) and Brachiaria brizantha (Hochst. ex A. Rich stapf; cv. Xaraés) were

90

provided by Matsuda (Rebeirao Preto, Brazil); and planted in a greenhouse at

91

College of Animal Science and Food Engineering (Pirassununga – USP, Brazil),

92

located at 21° 59' N, 47° 25' W and 635 m altitude. The climate is classified as

93

CWA with average annual precipitation of 1.238 m and relative humidity of 73%.

94

The daylight length and sunshine hours averages were of 12.8 h and 160.5 h

95

for summer (January), 11.2 h and 174.5 h for autumn (April), 11.0 h and 177.0 h

96

for winter (July), and 12.6 h and 158.3 h for spring (October), respectively. The

97

recorded highest and lowest temperatures were of 39.2 °C and 20.5 °C for

98

summer; 31.3 °C and 13.6 °C for autumn; 27.8 °C and 9.9 °C for winter, and

99

33.0 °C and 14.3 °C for spring, respectively. Plant aerial parts were harvested in



100

triplicate at two different heights (B. decumbens – 10 and 20 cm and B.

101

brizantha – 15 and 30 cm) in spring, summer, autumn and winter, exactly before

102

sunrise, under CRBD and were morphologically separated. Plant materials were

103

cut down into pieces of ≤3 inches length, and after drying under 40 °C were

104

ground on a Wiley Mini-Mill at 16 mesh size and stored in black plastic bottles

105

for further uses.

106

Sample Preparation. Each sample was independently processed (24

107

total herb samples per species). Extraction was performed on a Dionex ASE

108

200 extractor (Dionex Corp., Sunnyvale, CA, USA). A portion of 150 mg of the

109

dried and ground herb samples were mixed with diatomaceous earth (Dionex

110

ASE Prep DE, Dionex Corp., Sunnyvale, CA, USA) and placed in a 5 mL

111

stainless steel extraction cells. A cellulose filter (Dionex Corp.) together with 1 g

112

of LiChroprep RP-18 (40 – 63 µm) (Merck, Darmstadt, Germany) were placed at

113

the bottom of the extraction cells to facilitate chlorophyll removal.

114

The cells were filled with extraction solvent (80% methanol in water, v/v),

115

pressurized at 1500 psi, and heated at 60 °C for 5 min to ensure that samples

116

reached thermal equilibrium. Samples were then extracted by three static cycles

117

of 5 min each at the experimental temperature and pressure. After extraction,

118

cells were rinsed with fresh solvent (60% of the cell volume) and purged with a

119

flow of nitrogen for 100 s. The extracts (25 mL each) were collected into the 60

120

mL glass vials, evaporated to dryness using rotary evaporator at 40 °C,

121

suspended in 1 mL of 20% methanol, and stored at -20 °C until required.

122

Before LC-MS analysis, each extract was sonicated for 10 min, followed

123

by centrifugation at 13000 rpm for 20 min and 22 °C using a Sigma 3-16KL

124

Refrigerated Centrifuge (Sigma Laborzentrifugen GmbH, Osterode am Harz,


Page 6 of 30

Page 7 of 30


125

Germany). Aliquots of the supernatants (5 µL each) were taken and brought to

126

1 mL with 20% acetonitrile in water (v/v), and subsequently analyzed by

127

UHPLC-HR-QTOF-MS with two technical replications (96 total runs).

128

UHPLC-HR-QTOF-MS analyses. UHPLC analysis was performed on a

129

Dionex UltiMate 3000RS Ultra High Performance Liquid Chromatograpfic

130

system with a Charged Aerosol Detector (Thermo Scientific, Dionex, Germany),

131

interfaced with a high resolution quadrupole time-of-flight mass spectrometer

132

(HR/Q-TOF/MS, Impact II, Bruker Daltonik GmbH, Bremen, Germany).

133

Chromatographic separation of Brachiaria metabolome was performed on an

134

Acquity UPLC BEH C18 column (100 × 2.1 mm, 1.7 µm, Waters, Manchester,

135

UK) maintained at 30 °C. The mobile phase consisted of solvent A (0.1% formic

136

acid in Milli-Q water, v/v) and solvent B (0.1% formic acid in acetonitrile, v/v) at

137

a flow rate of 0.4 mLmin-1. The gradient elution was as follow: 2% B from 0 to 1

138

min, then concavely increased to 20% B in 20 min, followed by concave

139

increase to 28% B in 8 min and another increase to 55% B in 5 min. The

140

column was eluted with this concentration of solvent B for 2 min more and then

141

was re-equilibrated with 2% B for 5 min. Samples were kept at 15 °C in the

142

autosampler. The injection volume was 5.0 µL. The mass spectrometer was

143

operated in the negative electrospray ionization mode after confirmation of low

144

sensitivity and poor resolution in the positive mode. The following parameters

145

were used: capillary voltage was set at 2.8 kV; nebulizer 0.7 bar; dry gas 6.0

146

Lmin-1; dry temperature 200 °C. The mass scan range was set as 150 – 2000

147

m/z. MS/MS spectra were acquired in a data-dependent manner, whereby ions

148

(maximum 2) from each scan were subjected to collision induced fragmentation

149

if their absolute intensity exceeded 1800 counts. Depending on ion’s m/z,



150

variable collision energy in range from 15 to 35 eV was used. Internal

151

calibration was achieved with 10 mM sodium formate solution introduced to the

152

ion source via a 20 µL loop at the beginning/end of each analysis using a 6-port

153

valve. Calibration was carried out using high precision calibration mode (HPC).

154

Data were collected and processed by the software DataAnalysis 4.3 (Bruker

155

Daltonik GmbH, Germany).

156

Data Processing and Analysis. ProfileAnalysis software (version 2.1,

157

Bruker Daltonik GmbH, Germany) was used to pre-process the raw UHPLC-

158

QTOF-MS data. ProfileAnalysis parameters were set as follow: advanced

159

bucket generation with retention time range 1.0 – 32.0 min, mass range 150 –

160

1200 m/z, normalization to sum of peaks, background subtraction, and time

161

alignment. LC-MS analyses were processed with the Find Molecular Futures

162

(FMF) function to create compounds (molecular features) with signal-to-noise

163

threshold of 3 for peak detection. Generated bucket table consisting of m/z –

164

retention time pairs and respective compound intensity was exported and

165

uploaded to MetaboAnalyst,18 which is an open bioinformatics website providing

166

an analytical pipeline for high-throughput metabolomics studies. The univariate

167

analysis volcano plot, a common method used for exploratory data analysis was

168

performed. This provides a preliminary overview of features that are potentially

169

significant for separation of the two groups. The multivariate principal

170

component analysis (PCA) and partial lest-squares discriminant analysis (PLS-

171

DA) were performed after row-wise normalization based on sum of peak areas

172

and Pareto scaling, to investigate the overall variation in the metabolome of

173

Brachiaria species. In addition, intra-species variations for both species were


Page 8 of 30

Page 9 of 30


174

also investigated using separate PCA models and dendrograms for each

175

species.

176

Isolation and Structural Characterization of Saponin Standards.

177

Extracts of both plants, obtained as described above, were loaded onto a 12 cm

178

× 1.5 cm column packed with Sephadex LH-20 (Sigma-Aldrich, Steinheim,

179

Germany) and eluted with isocratic methanol/water (8:2, v/v) mobile phase. Two

180

milliliter fractions were collected and checked by TLC on Silica gel RP-18 F254S

181

plates (Merck, Darmstadt, Germany), developed with acetonitrile/water/formic

182

acid (4:6:0.5, v/v), sprayed with Liebermann-Burchard reagent and heated at

183

130 °C. A major fraction was obtained, which yielded the pure compounds 19

184

(protoneodioscin)19 and 20 (protodioscin)19,20 after a purification procedure on a

185

semi-preparative HPLC chromatographic system equipped with a Gilson 321

186

pump, a Gilson GX-271 liquid handler, a Gilson Prep ELSTM II detector and a

187

semi-preparative reversed phase column Atlantis Prep T3 (250 mm × 10 mm

188

i.d., 5µm, Waters, Milford, MA). The separation was carried out isocratically at

189

30 °C using 3 mLmin-1 flow of 25% acetonitrile (cont. 0.2% formic acid). The

190

purity and structures of isolated compounds were confirmed by mass

191

spectrometry and NMR spectroscopy. 1D and 2D NMR spectra (1H, 13C, HSQC,

192

HMBC, 1H-1H COSY DQF, TROESY) were recorded on a Bruker Avance III HD

193

AscendTM-500 spectrometer equipped with 5 mm 1H {109Ag-31P} broadband

194

inverse (BBI) probe, in pyridine-d5/deuterated water (95:5, v/v) at 30 °C (see

195

Supporting Information Figures S9 to S18). Exact mass and MS/MS

196

fragmentation patterns were determined on a HR/Q-TOF/MS (Impact II, Bruker

197

Daltonik GmbH, Bremen, Germany) (see Supporting Information Figures S19

198

and S20).



199

Calibration Curves of Standard Compounds. The pure saponins

200

prepared as described above were taken to prepare 0.1 mgmL-1 stock

201

solutions in 50% acetonitrile, which were subsequently stored at -20 °C.

202

Standard working solutions used for calibration were prepared by diluting the

203

stock solutions with 20% acetonitrile to the desired concentrations. Each

204

calibration curve was constructed by running standard at five different

205

concentrations in triplicate.

206

UPLC Quantitative Analysis. Brachiaria samples were analyzed using a

207

Waters ACQUITY UPLCTM system coupled to Waters TQ Detector (Waters

208

Corp.) in selected ion monitoring (SIM) mode, and operated in positive

209

electrospray ion mode, set to m/z 1071.5 [M+Na]+ for both compound 19 and

210

20. The following instrumental parameters were used for ESI-MS analysis:

211

capillary voltage, 3.1 kV; cone voltage, 110 V; desolvation gas, N2 800 L/h;

212

cone gas, N2 100 L/h; source temp. 140 °C, desolvation temp. 350 °C, dwell

213

time 300 ms. Waters MassLynx software v.4.1 was used for acquisition and

214

data processing.

215

Samples were separated on a BEH C18 column (100 × 2.1 mm, 1.7 µm,

216

Waters, Manchester, UK), which was maintained at 25 °C. The flow rate was

217

adjusted to 0.4 mLmin-1. The mobile phase consisted of isocratic 21%

218

acetonitrile in Milli-Q water (v/v) with 0.1% formic acid. Samples were kept at 8

219

°C in the autosampler. The injection volume was of 2.0 µL (partial loop with

220

needle overfill mode). The separation was completed in 35 min.

221

Statistical analyses of the data were performed using GraphPad Prism

222

5.0. The variables were tested for normality and fitted to a Gaussian distribution.

223

Two-way ANOVA analyses were performed to test how the concentrations of


Page 10 of 30

Page 11 of 30


224

saponins were affected by species and seasons, as well as one-way ANOVA to

225

investigate the variation within cutting-heights. Statistical significance was

226

declared at p < 0.05

227

RESULTS AND DISCUSSION

228

As part of an agricultural trial developed in Brazil, aerial parts of B.

229

decumbens and B. brizantha were harvested in triplicate at two different heights

230

in spring, summer, autumn and winter. After drying, herb samples were finely

231

ground and subjected to automated extraction with 80% methanol. The

232

separation of Brachiaria extracts were carried out on an UHPLC system

233

interfaced with a HR/Q-TOF/MS within 40 min. Visual examination of their

234

typical base peak ion chromatograms obtained in negative ESI mode displayed

235

clear differences (Figure 1), which resulted in the detection of 813 features

236

(metabolite ions) after peak alignment.

237

Differentiation of Brachiaria Metabolome. In order to explore the

238

phytochemical variation in the UHPLC-MS data, the generated chromatograms

239

were used for a chemometric analysis since they provide a comprehensive

240

chemistry overview of the two species. Volcano plot (Supporting Information,

241

Figure S1), an univariate analysis method, was used to obtain a first rough

242

ranking of potentially significant features for the differentiation between the two

243

species under study. It is a scatter-plot for the two group data, combining fold

244

change (FC) and t-tests. On the y-axis is plotted the negative log of the p-

245

values, so that data points with low p-values (highly significant) appear toward

246

the top of the plot. The x-axis is the log of the FC between the two groups B.

247

brizantha/B. decumbens. In this way those points that are found toward the top

248

of the plot that are also far to either the left- or right-hand side, represent



249

features displaying large magnitude fold changes as well as high statistical

250

significance. Among features above the set threshold (Table S1), stand out

251

three C-glycosyl flavones (8, 10, and 13), an O-glycosyl flavone (16), two

252

steroidal saponins (19 and 21), and two unidentified compounds (7 and 11),

253

specific for B. decumbens, as well as an O-glycosyl flavonol (15), two O-

254

glycosyl flavonolignans (17 and 18), and a C-glycosyl flavone (3) correlated to

255

B. brizantha, as the most-meaningful (Figure S1).

256

The unsupervised method principal component analysis (PCA) was

257

applied to discover inherent group patterns in the data and to examine whether

258

the metabolites selected by volcano plot are also detected as being significant.

259

This method is called “unsupervised” because is performed without data

260

labeling with class membership. PCA is a powerful method to perform the

261

dimension reduction of a dataset containing hundreds of metabolites, finding

262

only few combinations of them that best explain the total variation in the original

263

dataset. In the Figure 2A the PCA scores plot is presented showing the

264

distribution of B. brizantha and B. decumbens samples along PC1 and PC2.

265

The first two principal components (PCs) accounted for 47.4% and 11.5% of the

266

variation in the spectral data, respectively. The PCA scores plot shows a visible

267

separation along PC1 of two distinct clusters of samples that are species

268

specific. In addition, an intra-species variation within B. brizantha samples along

269

PC2 was also observed, which was further investigated. The corresponding

270

loadings plot (Figure 2B), displays the features that are responsible for the

271

separating groups. Those data points that are further from the origin toward the

272

left-hand side than most other points in the plot, predominate in B. brizantha

273

while those toward the right-hand side are highly correlated to B. decumbens. In


Page 12 of 30

Page 13 of 30


274

this way, it can be observed that the steroidal saponin 19, the two C-glycosyl

275

flavones 12 and 14, and compound 7 predominate in B. decumbens, while the

276

two C-glycosyl flavones 6 and 4, the O-glycosyl flavonol 15, and the phenolic

277

acid 1 mostly occur in B. brizantha. All these compounds, except 14 and 4,

278

were also selected by volcano plot within the top fifty features (see Table S1).

279

Although the steroidal saponin 20 seems not to be a strong contributor for the

280

interspecies separation, it appears highly significant for the variation among B.

281

brizantha samples. In addition, the highlighted metabolites by volcano plot 8,

282

10, 17, 18 and 21 (Figure S1) were also detected in PCA loadings plot (Figure

283

2B), however they contribute less to discrimination between the two groups.

284

The observed intra-species variation for B. brizantha along PC2 was

285

further investigated by creating a separate PCA model and dendrogram. In the

286

Figure S4 the PCA scores plot, loadings plot, and the corresponding

287

dendrogram are presented. The scores plot displayed separation in PC1 of

288

spring samples along with those harvested in winter at the shorter cutting-

289

height, from samples collected in autumn, summer, and those from winter at the

290

higher cutting-height. Consistent with that observed in Figure 2B, the loadings

291

plot also showed the steroidal saponin 20 as the most significant variable for

292

such separation (Figure S4). These results are in agreement with the

293

hierarchical cluster analysis (HCA) where samples were grouped into two major

294

classes with the largest Euclidean distance value using ward clustering

295

algorithm (Figure S4). One of these classes consists of spring samples together

296

with winter samples collected at the shorter cutting-height, while the other is

297

composed by the rest of samples. In the same way, B. decumbens was also

298

investigated for intra-species variations. Its PCA scores plot and dendrogram



299

(Figure S5) displayed a comparable clustering pattern to B. brizantha, with

300

samples collected in spring and winter clearly separated from those harvested

301

in summer and autumn. Again, the steroidal saponins 19 and 20 are the

302

variables responsible for such variation, as it can be found in the loadings plot

303

(Figure S5).

304

To further investigate the clustering pattern in the whole dataset and the

305

ability of significant metabolites detected by PCA to discriminate between B.

306

brizantha and B. decumbens, a supervised partial least squares-discriminant

307

analysis (PLS-DA) approach was applied. It uses multivariate regression

308

techniques to extract via linear combination of original variables (X) the

309

information that can predict the group membership (Y). The 3-dimensional PLS-

310

DA scores plot between PC1 (47.4%), PC2 (10.5%), and PC3 (4.1%) shows a

311

perfect separation between both species along PC1 (Figure 3A). Consistent

312

with PCA, the top contributor to this separation highly correlated to B.

313

decumbens is the steroidal saponin 19, although the C-glycosyl flavones 12 and

314

14, and compound 7 are also significant, as shown the PLS-DA loadings plot

315

(Figure 3B). While, those associated to B. brizantha includes the C-glycosyl

316

favone 4 as the most significant, along with other C-glycosyl flavone (6), a

317

steroidal saponin (20), an O-glycosyl flavonol (15), and a phenolic acid (1).

318

These metabolites were also the most-meaningful contributors to PLS-DA

319

component 1 based on variables important in projection (VIP) (Figure 3C). VIP

320

is a weighted sum of squares of the PLS loadings taking into account the

321

amount of explained Y-variation in each dimension. The steroidal saponin 19

322

and the C-glycosyl favone 4 had the top VIP scores, which in turn means that

323

they can satisfactorily explain most of the variation between both species.


Page 14 of 30

Page 15 of 30


324

In order to assess the statistical significance of the group discrimination

325

obtained by PLS-DA model, a permutation test was performed. In each

326

permutation, a PLS-DA model was built between the data (X) and the permuted

327

group labels (Y) using the optimal number of components determined by cross

328

validation for the model based on the original group assignment. The separation

329

distance is based on the ratio of the between sum of the squares and the within

330

sum of squares (B/W-ratio) for the group assignment prediction of each model

331

(Figure 3D). The observed test statistic was not part of the distribution based on

332

the permutation group assignments and the p-value was < 0.05, thus

333

demonstrating that the PLS-DA model was reliable.

334

The area under the receiver operating characteristic (ROC) curve (AUC)

335

was also used to evaluate the false-positive rate (1-specificity) and true-positive

336

rate (sensitivity) of individual metabolites for distinguishing B. decumbens and

337

B. brizantha. The AUC value is a measure of the ability of a given metabolite to

338

correctly classify the population as belonging to B. decumbens or B. brizantha.

339

The closer the AUC value is to 1.0, the better is the examined metabolite in

340

discriminating the two populations. Figure 4 shows the ROC curves for the top

341

two discriminating metabolites, with AUC = 1.0 for the saponin 19 and AUC =

342

0.877 for the C-glycosyl flavone 4.

343

These findings suggest that the steroidal saponin 19 (protoneodioscin) is

344

the major source of variation or a biomarker for B. decumbens when compared

345

to B. brizantha, which in turn is described here for the first time in these species.

346

Paradoxically, its stereoisomer 20 (protodioscin), which has been linked with the

347

outbreaks causes,7,

348

indicates that toxicity in Brachiaria, at least, should not be directly attributed to

8–10

seems to be closely related to B. brizantha. This



Page 16 of 30

349

this compound as previously thought, but perhaps to protoneodioscin (19)

350

instead. To confirm this, we carried out quantification of these two steroidal

351

saponins among samples from both species and performed corresponding

352

statistical analyses.

353

Quantification

of

Protoneodioscin

and

Protodioscin.

The

354

quantification of saponins 19 and 20 were carried out using a five points

355

calibration curves, which showed excellent fitting to a second order polynomial

356

(quadratic) equation (Figure S6). Analyses were performed in a UPLC-TQ-MS

357

in positive ion mode, using selected ion monitoring (SIM) of the cationized

358

molecule at m/z 1071.5 [M+Na]+ and retention times of 17.70 min and 18.74 min

359

for 19 and 20, respectively. All obtained values (mean ± SD) are quoted in

360

milligrams of compound per grams of dried weight of plant material (mgg-1

361

D.W.) and are presented in Supporting Information Table S2.

362

For comparison, each obtained value for the content of these saponins in

363

B. decumbens and B. brizantha were presented in dot graphs, where data

364

points dispersion is well visible (Figure 5). The concentration of protoneodioscin

365

(19) in B. decumbens (Figure 5A) was higher than 20 mgg-1 D.W. in all

366

seasons, reaching even greater levels in some cases, like in winter (31.07 ± 2.9

367

mgg-1 D.W.). These values were however found to be not significantly different

368

among seasons (p > 0.05) (Figure S7). By contrast, the content of this

369

compound was considerably different in B. brizantha (p < 0.05), with values five

370

times lower than in B. decumbens, and never reaching more than 4 mg·g-1 D.W.

371

Although important variations of the protodioscin (20) content within seasons

372

were found in B. brizantha, especially in the summer when the lowest amount

373

was registered (8.53 ± 0.5 mgg-1 D.W.), the values were not significantly


Page 17 of 30


374

different from those of B. decumbens (Figure 5B). This puts into question that

375

only protodioscin (20) can exert the toxicity in Brachiaria, particularly when

376

much higher concentrations of this compound were found in the less toxic

377

species (B. brizantha) in spring (24.83 ± 1.4 mgg-1 D.W.) and winter (25.47 ±

378

3.3 mgg-1 D.W.). Hence, the added value of protoneodioscin for the total

379

saponin concentration in B. decumbens could be crucial. Obviously, the sum of

380

these two saponins in B. decumbens were considerably higher than in B.

381

brizantha (Figure 5C), ranging from 4.1% to 5.3% of the dry weight (Table S2),

382

while in B. brizantha they were only between 1.0% and 2.9%.

383

The variation in concentration of saponins 19, 20, and their sum between

384

different cutting-heights within seasons were also investigated by one-way

385

ANOVA (Table S2). In B. decumbens the saponins concentrations were not

386

significantly different either within seasons or plant heights (Figure S7). B.

387

brizantha however showed different properties (Figure S8), especially for the

388

content of 20 that significantly changes within summer, being lower also when

389

plants are younger (15 cm height).

390

Despite that metabolomics study can hardly give hints regarding

391

bioactivity of specific secondary metabolite, there are previous investigations

392

that may validate our findings. Lajis et al (1993) reported the isolation of epi-

393

sarsasapogenin and epi-smilagenin from the rumen content of sheep

394

intoxicated by B. decumbens.21 The crystals deposited in the biliary system of

395

intoxicated livestock which are products of a detoxification, were reported as

396

insoluble salts of epi-sarsasapogenin and epi-smilagenin glucuronates.22 It is

397

important to note that these two spirostanol sapogenins differ only in their

398

configuration around carbon 25, just as the difference between protoneodioscin



Page 18 of 30

399

(19) and protodioscin (20). Epi-sarsasapogenin and epi-smilagenin are

400

considered to be the metabolized products of saponins 19 and 20, respectively.

401

Metabolites

Identity

Assignment.

To

achieve

unambiguous

402

identification of compounds 19 and 20, they were isolated and then elucidated

403

based on a combination of NMR, HRMS, and MS/MS techniques (Figures S9 to

404

S20). Compound 19 had NMR spectroscopic characteristics identical to those of

405

protoneodioscin, a furostanol-type steroidal saponin (Figure 6).19 To our

406

knowledge, this is the first time protoneodioscin was detected in Brachiaria

407

genus. Compound 20 showed NMR data almost superimposable to that of 19,

408

except for small differences in the chemical shifts of the opened ring F

409

(Supporting Information Table S3). It was finally identified as protodioscin, also

410

a furostanol-type steroidal saponin.19,20 The structure of these two saponins

411

basically differ only in the configuration around carbon 25. This was deduced as

412

25S for 19 and 25R for 20, according to the Agrawal’s rule, which is based on

413

the difference in the chemical shifts between germinal protons of the

414

glycosyloxy methylene H2-26 (∆ab = δa - δb), establishing the configuration as

415

25R if ∆ab < 0.48 or 25S if ∆ab > 0.57.23 The fragmentation patterns observed in

416

MS/MS spectra of 19 and 20 (Figures S19 and S20) were identical, displaying

417

neutral losses of two deoxyhexoses and two hexoses (Table 1).

418

Remaining compounds (Table 1) were tentatively identified based on

419

tandem mass spectrometry (MS/MS), metabolite database of METLIN,17

420

authentic standards, and data available in the literature. In this way, the two

421

phenolic acid isomers 1 and 2 were assigned as 3-O-caffeoylquinic acid (3-

422

CQA) and 4-O-caffeoylquinic acid (4-CQA), respectively.24 The ten C-glycosyl

423

flavones were identified as orientin 2ʹʹ-O-glucoside (3),25 carlinoside (4),26


Page 19 of 30


424

schaftoside (5),25 orientin 7-O-rhamnoside (6),27 apigenin 6,8-di-C-arabinoside

425

(8),28

426

diosmetin-8-C-rhamnosyl-7-O-glucoside

427

deoxyhexoside (12), apigenin 6-C-deoxyhexosyl-8-C-pentoside (13), and

428

cassiaoccidentalin B (14).26 To our knowledge, compounds 12 and 13 have not

429

been reported before from Brachiaria or other natural source, therefore they

430

represent presumably new natural compounds. The O-glycosyl flavonoids 15

431

and 16 were identified as being ombuin-3-O-rutinoside31 and tricin 7-O-

432

glucoside,32 respectively. Based on the similarities of MS/MS fragmentation

433

patterns with related phytochemicals, the two glycosyl flavonolignan isomers 17

434

and 18 were assigned as tricin 4ʹ-O-(β-guaiacylglyceryl) ether 7-O-(O-

435

deoxyhexosyl)-glucoside,32 which presumably are also new natural compounds.

436

Compound 21 was identified as acetyl protodioscin33 while compounds 7 and 11

437

remain unidentified.

2ʹ,4ʹ,5,7-tetrahydroxy-flavone-8-C-arabinosyl-7-O-glucoside (10),30

luteolin

(9),29

6-C-pentosyl-8-C-

438 439

Supporting Information: Tables with quantitative and NMR data for

440

protoneodioscin (19) and protodioscin (20), volcano plot, PCA scree plot and

441

biplot, PCA and HCA for intra-species variation, copy of NMR spectra including

442

1

H,

13

C, HMBC, COSY DQF, HSQC, TROESY, and HR-QTOF-MS/MS for

443

compounds 19 and 20.

444

Funding: This work was funded by the Statutory Activities of the Institute of Soil

445

Science and Plant Cultivation – State Research Institute and by Twas-CNPq

446

Postgraduate Fellowship Programme (Pocess No. 190178/2012-4).



447

REFERENCES

448 449

(1)

FAOSTAT database. Food and Agriculture Organization of the United Nation http://faostat3.fao.org/browse/Q/QA/E, (2014).

450 451 452

(2)

Vigna, B. B. Z.; Jungmann, L.; Francisco, P. M.; Zucchi, M. I.; do Valle, C. B.; de Souza, A. P. Genetic diversity and population structure of the Brachiaria brizantha germplasm. Trop. Plant Biol. 2011, 4, 157–169.

453 454 455

(3)

Riet-Correa, B.; Castro, M. B.; Lemos, R. a.; Riet-Correa, G.; Mustafa, V.; RietCorrea, F. Brachiaria spp. poisoning of ruminants in Brazil. Pesqui. Veterinária Bras. 2011, 31, 183–192.

456 457 458

(4)

Furlan, F. H.; Colodel, E. M.; Lemos, R. A. A.; Castro, M. B.; Mendoça, F. S.; Riet-Correa, F. Poisonous plant affecting cattle in Central-Western Brazil. Int. J. Poisonous Plant Res. 2012, 2, 1–13.

459 460 461

(5)

Tokaria, C. H.; Brito, M. F.; Barbosa, J. D.; Peixoto, P. V.; Döbereiner, J. Plantas toxicas do Brasil: Para as animais de producção, 2nd ed.; Helianthus, R. J., 2012.

462 463 464

(6)

Driemeier, D.; Colodel, E. M.; Seitz, A. L.; Barros, S. S.; Cruz, C. E. F. Study of experimentally induced lesions in sheep by grazing Brachiaria decumbens. Toxicon 2002, 40, 1027–1031.

465 466 467

(7)

Cruz, C. E. F.; Driemeier, D.; Pires, V. S.; Schenkel, E. P. Experimentally induced cholangiohepatopathy by dosing sheep with fractionated extracts from Brachiaria decumbens. J. Vet. Diagn. Invest. 2001, 13, 170–172.

468 469 470

(8)

Saturnino, K. C.; Mariani, T. M.; Barbosa-Ferreira, M.; Brum, K. B.; dos Santos Fernandes, C. E.; Lemos, R. A. A. Intoxicação experimental por Brachiaria decumbens em ovinos confinados. Pesqui. Vet. Bras. 2010, 30, 195–202.

471 472 473

(9)

Driemeier, D.; Döbereiner, J.; Peixoto, P. V.; Brito, M. F. Relação entre macrófagos espumosos (“foam cells”) no fígado de bovinos e ingestão de Brachiaria spp no Brasil. Pesqui. Vet. Bras. 1999, 19, 79–83.

474 475 476

(10)

Lemos, R. A. A.; Salvador S.; Nakazato L. Photosensitization and crystalassociation cholangiohepatopathy in cattle grazing Brachiaria decumbens in Brazil. Vet. Hum. Toxicol. 1997, 39, 376–377.

477 478 479

(11)

Quinn, J. C.; Kessell, A.; Weston, L. a. Secondary plant products causing photosensitization in grazing herbivores: Their structure, activity and regulation. Int. J. Mol. Sci. 2014, 15, 1441–1465.

480 481 482 483 484

(12)

Gracindo, C. V.; Louvandini, H.; Riet-Correa, F.; Barbosa-Ferreira, M.; de Castro, M. B. Performance of sheep grazing in pastures of Brachiaria decumbens, Brachiaria brizantha, Panicum maximum, and Andropogon gayanus with different protodioscin concentrations. Trop. Anim. Health Prod. 2014, 46, 733–737.

485 486 487

(13)

Rathahao-Paris, E.; Alves, S.; Junot, C.; Tabet, J.-C. High resolution mass spectrometry for structural identification of metabolites in metabolomics. Metabolomics 2016, 12, 1–15.

488 489

(14)

Allwood, J. W.; Vos, R. C. H. De; Moing, A.; Deborde, C.; Erban, A.; Kopka, J.; Goodacre, R.; Hall, R. D. Plant Metabolomics and Its Potential for Systems 20 ACS Paragon Plus Environment

Page 20 of 30

Page 21 of 30


Biology Research: Backgraound concepts, technology, and methodology, 1st ed.; Elsevier Inc., 2011; Vol. 500.

490 491 492 493 494 495

(15)

Xiong, A.; Yang, L.; Ji, L.; Wang, Z.; Yang, X.; Chen, Y.; Wang, X.; Wang, C.; Wang, Z. UPLC-MS based metabolomics study on Senecio scandens and S. vulgaris: an approach for the differentiation of two Senecio herbs with similar morphology but different toxicity. Metabolomics 2012, 8, 614–623.

496 497 498 499

(16)

Viljoen, A. M.; Zhao, J.; Sandasi, M.; Chen, W.; Khan, I. A. Phytochemical distinction between Pelargonium sidoides (“Umckaloabo”) and P. reniforme through 1H-NMR and UHPLC–MS metabolomic profiling. Metabolomics 2015, 11, 594–602.

500 501 502

(17)

Smith, C. A.; O’Maille, G.; Want, E. J.; Qin, C.; Trauger, S. A.; Brandon, T. R.; Custodio, D. E.; Abagyan, R.; Siuzdak, G. METLIN: A metabolite mass spectral database. Ther Drug Monit 2005, 27, 747–751.

503 504

(18)

Xia, J.; Sinelnikov, I. V.; Han, B.; Wishart, D. S. MetaboAnalyst 3.0-making metabolomics more meaningful. Nucleic Acids Res. 2015, 43, W251–W257.

505 506 507

(19)

Hu, K.; Dong, A.; Yao, X.; Kobayashi, H.; Iwasaki, S. Antineoplastic agents; II. Four furostanol glycosides of Dioscorea collettii var. hypoglauca. Planta Med. 1997, 63, 161–165.

508 509

(20)

Huang, H. L.; Liu R. H.; Shao, F. Structural determination of two new steroidal saponins from Smilax china. Magn. Reson. Chem. 2009, 47, 741–745.

510 511 512 513

(21)

Lajis, N. H.; Abdullah, A. S. H.; Salim, S. J. S.; Bremner, J. B.; Khan, M. N. Episarsasapogenin and epi-smilagenin: two sapogenins isolated from the rumen content of sheep intoxicated by Brachiaria decumbens. Steroids 1993, 58, 387389.

514 515 516

(22)

Flåøyen, A. Do steroidal saponins have a role in hepatogenous photosensitization diseases of sheep? In Saponins used in food and agriculture, 1st edition; Waller, G. R., Yamasaki, K.; Plenum Press: New York, 1996; pp-395.

517 518

(23)

Agrawal, P. K. Assigning stereodiversity of the 27-Me group of furostane-type steroidal saponins via NMR chemical shifts. Steroids 2005, 70, 715–724.

519 520 521

(24)

Clifford, M.; Johnston, K.; Knigh, S.; Kuhnert, N. A hierarchical scheme for LCMSn identification of chlorogenic acid. J. Agric. Food Chem. 2003, 51, 2900– 2911.

522 523 524 525

(25)

Ferreres, F.; Silva, B. M.; Andrade, P. B.; Seabra, R. M.; Ferreira, M. A. Approach to the study of C-glycosyl flavones by ion trap HPLC-PADESI/MS/MS: Application to seeds of quince (Cydonia oblonga). Phytochem. Anal. 2003, 14, 352–359.

526 527 528

(26)

Costa, G.; Ferreira, J. P.; Vitorino, C.; Pina, M. E.; Sousa, J. J.; Figueiredo, I. V.; Batista, M. T. Polyphenols from Cymbopogon citratus leaves as topical antiinflammatory agents. J. Ethnopharmacol. 2016, 178, 222–228.

529 530

(27)

Ibrahim, R. K.; Shaw, M. Phenolic constituents of the oil flax (linum usitatissimum). Phytochemistry 1970, 9, 1855–1858.

531 532 533

(28)

Singh, A.; Kumar, S.; Bajpai, V.; Reddy, T. J.; Rameshkumar, K. B.; Kumar, B. Structural characterization of flavonoid C - and O -glycosides in an extract of Adhatoda vasica leaves by liquid chromatography with quadrupole time-of-flight 21 ACS Paragon Plus Environment


mass spectrometry. Rapid Commun. Mass Spectrom. 2015, 29, 1095–1106.

534 535 536 537

(29)

Ibrahim, L. F.; Marzouk, M. M.; Hussein, S. R.; Kawashty, S. A.; Mahmoud, K.; Saleh, N. A. M. Flavonoid constituents and biological screening of Astragalus bombycinus Boiss. Nat. Prod. Res. 2013, 27, 386–393.

538 539 540 541

(30)

Colombo, R.; Yariwake, J. H.; Ferreira, E.; Ndjoko, K.; Hostettmann, K. LCMs/Ms analysis of sugarcane extracts and differentiation of monosaccharides moieties of flavone C-glycosides. J. Liq. Chromatogr. Relat. Technol. 2013, 36, 239–248.

542 543 544

(31)

Simirgiotis, M. J. Antioxidant capacity and HPLC-DAD-MS profiling of chilean peumo (Cryptocarya alba) fruits and comparison with german peumo (Crataegus monogyna) from Southern Chile. Molecules 2013, 18, 2061–2080.

545 546 547 548

(32)

Yang, Z.; Nakabayashi, R.; Okazaki, Y.; Mori, T.; Takamatsu, S.; Kitanaka, S.; Kikuchi, J.; Saito, K. Toward better annotation in plant metabolomics: isolation and structure elucidation of 36 specialized metabolites from Oryza sativa (rice) by using MS/MS and NMR analyses. Metabolomics 2014, 10, 543–555.

549 550

(33)

Ivanova, A.; Mikhova, B.; Klaiber, I.; Dinchev, D.; Kostova, I. Steroidal saponins from Smilax excelsa rhizomes. Nat. Prod. Res. 2009, 23, 916–924.

551


Page 22 of 30

Page 23 of 30


FIGURE CAPTIONS Figure 1. Base peak chromatograms in negative ion mode on a C-18 column for B. decumbens and B. brizantha extracts showing the most important metabolites for their differentiation. Figure 2. (A) PCA scores plot based on UHPLC-MS data showing separation of B. brizantha (red filled triangles) and B. decumbens (green filled squares), together with their respective 95% confidence regions. The explained variances are shown in brackets. (B) The corresponding loadings scatter plot showing the compounds that are correlated to separation in scores plot. Figure 3. PLS-DA of metabolites between B. brizantha and B. decumbens. (A) 3-dimensional scores plot using the three first components, accounting for 47.4, 10.5, and 4.1% of the total variance. (B) loadings plot showing variables responsible for discrimination in scores plot. (C) variables important in projection (VIP) scores of 15 top contributors to PLS-DA component 1. (D) PLS-DA model validation by permutation tests based on separation distance (B/W-ratio), with a p-value based on permutation as p=0.0165 (33/2000). Figure 4. Receiver operating characteristic (ROC) curve for the top two discriminating metabolites. For each of them, the left panel shows the area under the ROC curve (AUC), true positive and false positive rates, and confidence interval, the right panel shows the relative concentration of metabolite in B. brizantha and B. decumbens. Figure 5. Aligned dot graphs showing the concentrations of protoneodioscin (19), protodioscin (20), and their sum among B. decumbens and B. brizantha samples; mean with SEM is shown per each season. Two-way ANOVA analyses for the species and seasonal effects indicated that: (A) species, extremely significant (p < 0.0001); seasons, very significant (p = 0.0043). (B) species, not significant (p = 0.8433); seasons, extremely significant (p < 0.0001). (C) species, extremely significant (p < 0.0001); seasons, extremely significant (p < 0.0001). Figure 6. Structures of saponins 19 and 20 isolated from Brachiaria.



Page 24 of 30

Table 1. Tentative Identification of Most Significant Metabolites for Distinguishing Groups and Occurrence in B. decumbens (Bd) and B. brizantha (Bb).

No.

Rt (min)

Meas. mass (neg)

error (ppm)

Formula

MS2 diagnostic ions, m/z (intensity %)

Identification

Bd

Bb

+ +

+ +

-

+

carlinoside

+

+

schaftoside

+

+

orientin 7-Orhamnoside apigenin 6,8-di-Carabinoside 2ʹ,4ʹ,5,7-tetrahydroxyflavone-8-C-arabinosyl7-O-glucoside diosmetin-8-Crhamnosyl-7-Oglucoside luteolin 6-C-pentosyl-8C-deoxyhexoside apigenin 6-Cdeoxyhexosyl-8-Cpentoside cassiaoccidentalin B

-

+

+

-

-

+

+

-

+

-

+

-

+

+

Phenolic acids 1 2

5.74 9.31

353.0868 [M-H]353.0874 [M-H]

2.9 1.3

C16H18O9 C16H18O9

3

11.81

609.1447 [M-H]-

2.2

C27H30O16

4

13.05

-

579.1357 [M-H]

-0.2

C26H28O15

5

14.25

563.1404 [M-H]

-

0.4

C26H28O14

-

191 (100), 179 (41), 161 (7) 191 (100), 179 (58), 173 (57), 161 (12)

3-CQA 4-CQA

C- and C,O-glycosyl flavones 447 (8), 357 (28), 327 (100), 313 (46), 298 (21) 489 (5), 459 (12), 429 (29), 399 (95), 369 (100) 473 (16), 443 (24), 413 (13), 383 (71), 353 (100) 473 (18), 447 (0.8), 357 (20), 327 (36), 298 (100) 515 (15), 473 (29), 443 (34), 413 (18), 383 (100), 353 (95) 489 (25), 417 (6), 357 (22), 327 (46), 298 (100)

orientin 2ʹʹ-O-glucoside

6

14.35

593.1513 [M-H]

-0.1

C27H30O15

8

15.80

533.1302 [M-H]-

-0.3

C25H26O13

9

16.38

-

579.1351 [M-H]

0.7

C26H28O15

10

17.34

607.1673 [M-H]-

-0.8

C28H32O15

12*

17.56

563.1411 [M-H]-

-0.8

C26H28O14

13*

18.61

547.1456 [M-H]

0.2

C26H28O13

14

19.52

575.1408 [M-H]-

-0.3

C27H28O14

429 (4), 411 (10), 385 (9), 367 (23), 325 (100), 298 (62), 285 (24)

7 11

15.70 17.52

401.1821 [M-H]533.1314 [M-H]-

-0.9 -2.5

C19H30O9 C25H26O13

401 (100), 221 (39), 195 (32), 177 (38) O-glycosyl flavonol

unidentified unidentified

+ +

-

15

19.33

637.1770 [M-H]-

0.7

C29H33O16

607 (7), 491 (6), 461 (5), 329 (100), 313 (84), 299 (57), 271 (16) O-glycosyl flavone

ombuin-3-O-rutinoside

-

+

16

20.30

491.1202 [M-H]-

-1.5

C23H24O12

459 (6), 323 (13), 315 (38), 175 (83), 152 (100)

tricin 7-O-glucoside

+

-

17*

20.42

833.2511 [M-H]

-

-0.2

C39H46O20

637 (100), 525 (6), 477 (11), 329 (80)

tricin 4ʹ-O-(βguaiacylglyceryl) ether 7-O-(deoxyhexosyl)glucoside (Isomer 1)

-

+

18*

20.95

833.2516 [M-H]-

-0.8

C39H46O20

637 (100), 477 (4), 329 (81)

tricin 4ʹ-O-(βguaiacylglyceryl) ether 7-O-(deoxyhexosyl)glucoside (Isomer 2)

-

+

-

533 (4), 503 (6), 475 (14), 445 (10), 429 (44), 401 (8), 371 (13), 341 (54), 327 (100), 312 (30) 503 (5), 473 (15), 429 (26), 399 (100), 369 (78) 503 (4), 473 (25), 443 (19), 413 (17), 383 (100), 353 (89)

Flavonolignans

Steroidal saponins 19

23.61

1093.5421 [M+HCOOH-H]-

1.4

C51H84O22

901 (10), 755 (39), 593 (26), 575 (38), 431 (100)

protoneodioscin

+

+

20

23.80

1093.5417 [M+HCOOH-H]-

1.8

C51H84O22

901 (8), 755 (28), 593 (20), 575 (15), 431 (100)

protodioscin

+

+

21

30.50

1135.5505 [M+HCOOH-H]-

3.2

C53H86O23

901 (15), 755 (27), 593 (35), 575 (17), 431 (100)

acetyl-protodioscin

+

-

* presumably new compounds.


Page 25 of 30


Figure 1



Figure 2


Page 26 of 30

Page 27 of 30


Figure 3



Figure 4


Page 28 of 30

Page 29 of 30


Figure 5

Figure 6



TOC Graphic


Page 30 of 30

Ultrahigh-Performance Liquid ChromatographyâHigh-Resolution

Recommend Documents

Ultrahigh-Performance Liquid ChromatographyâHigh-Resolution