The construction of prediction models for the TRPV1-stimulating

Taichi Yoshitomi†, Naohiro Oshima§, Yuto Goto†, Shunsuke Nakamori‡, ... Faculty of Pharmaceutical Sciences, Tokyo University of Science, 2641 Y...
2 downloads 0 Views 917KB Size
Subscriber access provided by University of Newcastle, Australia

Article

The construction of prediction models for the TRPV1stimulating activity of ginger and processed ginger based on LC-HRMS data and PLS regression analyses Taichi Yoshitomi, Naohiro Oshima, Yuto Goto, Shunsuke Nakamori, Daigo Wakana, Naoko Anjiki, Koji Sugimura, Noriaki Kawano, Hiroyuki Fuchino, Osamu Iida, Toshiko Kagawa, Hideto Jinno, Nobuo Kawahara, Yoshinori Kobayashi, and Takuro Maruyama J. Agric. Food Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jafc.7b00577 • Publication Date (Web): 11 Apr 2017 Downloaded from http://pubs.acs.org on April 16, 2017

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 36

Journal of Agricultural and Food Chemistry

The construction of prediction models for the TRPV1-stimulating activity of ginger and processed ginger based on LC-HRMS data and PLS regression analyses Taichi Yoshitomi†, Naohiro Oshima§, Yuto Goto†, Shunsuke Nakamori‡, Daigo Wakana¶, Naoko Anjiki#, Koji Sugimura#, Noriaki Kawano#, Hiroyuki Fuchino#, Osamu Iida#, Toshiko Kagawa┴, Hideto Jinno∆, Nobuo Kawahara#, Yoshinori Kobayashi‡, Takuro Maruyama†,*

† Division of Pharmacognosy, Phytochemistry and Narcotics, National Institute of

Health Sciences, 1-18-1 Kamiyoga, Setagaya-ku, Tokyo 158-8501, Japan § Faculty of Pharmaceutical Sciences, Tokyo University of Science, 2641 Yamazaki, Noda, Chiba 278-8510, Japan ‡ School of Pharmacy Sciences, Kitasato University, 5-9-1 Shirogane, Minato-ku, Tokyo, 108-8641, Japan ¶ Faculty of Pharmaceutical Sciences, Hoshi University, 2-4-41 Ebara, Shinagawa-ku, Tokyo 142-8501, Japan # National Institutes of Biomedical Innovation, Health and Nutrition, 1-2 Hachimandai, Tsukuba, Ibaraki 305-0843, Japan

1

ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

Page 2 of 36

┴ Department of Clinical Pharmacy, Yokohama University of Pharmacy, 601 Matano-cho, Totsuka-ku, Yokohama, Kanagawa 245-0066, Japan ∆ Faculty of Pharmacy, Meijo University, 150 Yagotoyama, Tempaku-ku, Nagoya, Aichi 468-8503, Japan Corresponding author: Dr. Takuro Maruyama, Division of Pharmacogonosy, Phytochemistry and Narcotics, National Institute of Health Sciences, 1-18-1 Kamiyoga, Setagaya-ku, Tokyo 158-8501, Japan. E-mail: [email protected], Phone & FAX: (+81)3-3700-9165

2

ACS Paragon Plus Environment

Page 3 of 36

Journal of Agricultural and Food Chemistry

1

ABSTRACT

2

To construct a model formula to evaluate the thermogenetic effect of ginger (Zingiber

3

officinale Roscoe) from the ingredient information, we established transient receptor

4

potential vanilloid subtype 1 (TRPV1)-stimulating activity prediction models by using

5

a partial least squares projections to latent structures (PLS) regression analysis in

6

which the ingredient data from liquid chromatography/high-resolution mass

7

spectrometry (LC-HRMS) and the stimulating activity values for TRPV1 receptor were

8

used as explanatory and objective variables, respectively. By optimizing the peak

9

extraction condition of the LC-HRMS data and the data preprocessing parameters of

10

the PLS regression analysis, we succeeded in the construction of a TRPV1-stimulating

11

activity prediction model with high precision ability. We then searched for the

12

components responsible for the TRPV1-stimulating activity by analyzing the loading

13

plot and s-plot of the model, and we identified [6]-gingerol (1) and hexahydrocurcumin

14

(3) as TRPV1-stimulating activity components.

15

16

KEYWORDS

17

Zingiber officinale Roscoe, PLS regression analysis, TRPV1, LC-HRMS, [6]-shogaol,

18

[6]-gingerol, hexahydrocurcumin

3

ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

19

Page 4 of 36

INTRODUCTION

20

Ginger (Zingiber officinale Roscoe) is one of the most widely used spices

21

worldwide. Although the tropical Asian region is considered the origin of ginger, the

22

exact area remains unclear.1 Ginger has been used as spice in China and India for over

23

2,500 years.2 Today, ginger is cultivated in temperate regions including India, China,

24

Africa and Australia as a perennial plant, and it is used not only as a spice alone but

25

also in confectionery products and alcoholic beverages.3

26

The rhizome of ginger has also been used heavily in the traditional medicine of

27

East Asia for several hundred years. In the Japanese Pharmacopoeia (17th edition), two

28

crude drugs derived from Z. officinale are listed as "ginger" and "processed ginger".

29

They were defined as the rhizome of Z. officinale for ginger and the rhizome of

30

Z. officinale, after being passed through hot water or being steamed for processed

31

ginger. The crude drugs are known to have antiemetic and thermogenetic actions.

32

The thermogenetic action of ginger in particular has recently attracted much

33

attention, and various health foods containing it are sold in claiming to cure the

34

cold-natured people.4 Ginger's thermogenetic action is strongly suggested to be

35

attributed to the TRPV1 channel, a member of a subfamily of TRP channels.5

36

TRP channels are a type of membrane protein, and the nonselective cation ion

4

ACS Paragon Plus Environment

Page 5 of 36

Journal of Agricultural and Food Chemistry

37

channel composed of seven subfamilies.6 A member of these subfamilies, TRPV1 is

38

known as the receptor activated by heat, acid and pungent ingredients. The activation

39

of TRPV1 channels promotes an increase in the adrenalin concentration in the blood by

40

stimulating the vagus nerve through signal transmission. This adrenalin promotes the

41

expression of uncoupling protein (UCP) in brown adipocytes and results in

42

thermogenesis.7 As another pathway, it was proposed that adrenalin released in the

43

blood acts on β-adrenergic receptors on liver and adipose tissue, which also promotes

44

sthenia of the energy metabolism.8

45

Capsaicin, the pungent ingredient of red peppers, is well known as the active

46

component of TRPV1.9 Ginger also activates TRPV1, and the ginger constituents,

47

[6]-gingerol, [6]-shogaol, and their derivatives are identified as the responsible

48

compounds (Fig. 1).10, 11 These components have a vanilloid moiety as the common

49

structure, and the structure-activity relationship with TRPV1 receptor have been well

50

studied.11-13

51

Ginger is a natural product and has many cultivars. The intensity of ginger's

52

thermogenetic effect could thus differ depending on the genotype, habitat, cultivation

53

procedure and processing. Appropriate quality control is required for the saving of the

54

reproducible activities of health foods derived from ginger, and approaches using

5

ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

Page 6 of 36

55

chemometrics are now frequently used for the evaluation of food and crude drug

56

quality, in concert with the marked advances in analytical instruments and computer.14,

57

15

58

analysis (PCA) and the partial least squares projections to latent structures (PLS)

59

regression analysis. PCA analyses are used mainly to confirm the variance of entire

60

data set and to lead notable data from factor loading. PLS analyses are used for the

61

construction of a prediction model in which the analytical data and interest scores such

62

as the pharmacological activity and the product grade are treated as the explanatory

63

variable and the objective variable, respectively. In recent years, analyses by

64

orthogonal partial least-squares (OPLS; an improvement of the PLS method) are used

65

in various fields including the quality evaluation of foods and crude drugs.16, 17

The most representative multivariate analysis methods are the principal component

66

Multivariate analyses are conducted because they can easily express the quality

67

differences among the foods or crude drugs of interest by analyzing the quantity

68

variance of their components. For example, Pongsuwan and co-workers performed a

69

PLS analysis by using the ingredient data of green tea for the explanatory variable and

70

using a sensory evaluation for the objective variable, and they succeeded in making a

71

highly predictive model of the sensory evaluation.18 Their report provided an

72

evaluation method with high objectivity for subjective estimations such as those

6

ACS Paragon Plus Environment

Page 7 of 36

Journal of Agricultural and Food Chemistry

73

provided by the five senses. Their report also confirmed the usefulness of multivariate

74

analyses.

75

In the present study, we focused on the thermogenetic activity of ginger and

76

constructed an activity prediction model based on the ingredient data. Namely, we

77

measured the TRPV1-stimulating activity and obtained the LC-HRMS data of several

78

ginger and processed ginger samples. We applied these data to a PLS analysis to make

79

the prediction model. We also identified the active compounds by performing a

80

regression analysis of the resultant model.

81

82

MATERIALS AND METHODS

83

Samples

84

The details of the ginger and processed ginger used in this study are shown in

85

Table 1. The samples were collected from crude drug companies in Japan, and voucher

86

samples were deposited in the National Institutes of Biomedical Innovation, Health and

87

Nutrition (Japan). Each sample was pulverized and extracted with 10 volumes of

88

boiled water for 2 h. The extracts were freeze-dried to amorphous powder. The

89

processed ginger for the isolation of hexahydrocurcumin (3) was purchased from

90

Uchida Wakanyaku (Tokyo, Japan).

7

ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

91

Page 8 of 36

General procedure

92

The flash chromatography and the recycle HPLC were performed using an

93

Isolera One system (Biotage, Tokyo, Japan) equipped with a SNAP Ultra flash

94

chromatography cartridge (50 g), and an LC-9201 HPLC system (Japan Analytical

95

Industry, Tokyo, Japan) equipped with JAIGEL-GS310 (21.5 mm ID × 500 mm; Japan

96

Analytical Industry), respectively. The nuclear magnetic resonance (NMR) spectrum

97

was measured by an ECA-600 NMR Spectrometer (JEOL, Tokyo, Japan). The

98

LC-HRMS analysis was performed using a Prominence UFLC (Ultra Fast Liquid

99

Chromatography; Shimadzu, Kyoto, Japan) linked to an Orbitrap XL mass

100

spectrometer with electron-transfer dissociation (Thermo Fisher Scientific, Waltham,

101

MA, USA) equipped with a Kinetex 2.6 µm, C18, 100A, 2.1 × 100 mm column

102

(Phenomenex, Torrance, CA, USA).

103

Liquid chromatography/high-resolution mass spectrometry (LC-HRMS) analysis

104

One milligram of extract powders was dissolved in ultra-pure water (1 mL) and

105

filtered through a polytetrafluoroethylene (PTFE) filter, pore size 0.45 µm. Separately,

106

the reference solution for the alignment of the chromatogram was prepared at 1 mg/mL

107

by mixing all 26 sample solutions. All sample solutions were analyzed by the

108

Prominence UFLC with Orbitrap high-resolution mass spectrometry. The column was

8

ACS Paragon Plus Environment

Page 9 of 36

Journal of Agricultural and Food Chemistry

109

Kinetex 2.6 µm, C18, 100A, 2.1 × 100 mm column (Phenomenex) and was maintained

110

at 40°C. The auto sampler temperature was set to 4°C and the injection volume was 1

111

µL. The mobile phase consisted of 0.1% formic acid (solvent A), and acetonitrile

112

containing 0.1% formic acid (solvent B) was used at a flow rate of 0.3 mL/min under

113

the following gradient condition: 5% B for the initial 5 min; 5%–70% B for 11 min;

114

70%–100% B for 2 min; and 100% B for 4 min. The electrospray ionization mass

115

spectrometry (ESI-MS) parameters were as follows: ion polarity, positive; scan range,

116

m/z 100 to 1000; capillary temperature, 400°C; sheath gas, 50 L/min; AUX gas, 25

117

L/min; capillary voltage, 15 V; MS resolution, 60,000. All samples were analyzed in

118

triplicate.

119

TRPV1-stimulating activity measurement

120

Human TRPV1-HEK293 stable cells were prepared as described in our previous

121

study.13 TRPV1-stimulating activity was measured according to procedures described

122

previously.13 However, in this study, the test solutions were prepared in dimethyl

123

sulfoxide and diluted in HBSS buffer. The TRPV1-stimulating activities of the ginger

124

and processed ginger extracts were measured at 5 ng/mL and are expressed as the

125

relative values for that of 0.2 µM capsaicin by using fluorescence intensity on end

126

point of the reaction. The activities of the purified or authentic compounds were

9

ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

Page 10 of 36

127

measured at 0.039 to 500 µM and are expressed as the EC50. EC50 values were

128

calculated using the GraphPad Prism 6 software program (GraphPad Software, La Jolla,

129

CA, USA). All samples and compounds were measured in sextuplicate.

130

Multivariate statistical analysis

131

We used the software program Progenesis QI (Waters, Milford, MA, USA) to

132

perform the peak alignment of the chromatograms and the peak extraction. For the

133

peak extraction, the automatic mode and the base peak mode were used. The obtained

134

peaks were converted into the data matrix by EZinfo statistical package (Umetrics,

135

Malmö, Sweden). The PLS regression analyses were carried out by using SIMCA-14

136

software (Umetrics). Pareto (Prt) as a pretreatment parameter without any

137

transformations was applied to the PLS-DA. In the TRPV1-stimulating activity

138

prediction models, we selected the UV, pareto (Prt) or Mean-center (Ctr) as

139

pretreatment of data matrix and applied them to the PLS, OPLS and PLS with

140

orthogonal signal correction (OSC) regression analysis. For the transformation of the

141

data matrix, none or Log10 (The explanatory variables with small score are

142

emphasized this transformation) was used. In addition, the number of latent variables

143

was fixed at 3 in all models. The prediction precision of the resultant prediction models

144

was evaluated by 'leave one out' cross-validation. The search for TRPV1-stimulating

10

ACS Paragon Plus Environment

Page 11 of 36

Journal of Agricultural and Food Chemistry

145

activity components was conducted on the basis of loading plot, variable importance in

146

the projection (VIP) and s-plot analyses.

147

148

RESULTS AND DISCUSSION

149

TRPV1-stimulating activity

150

We first measured the in vitro TRPV1-stimulating activity of ginger and

151

processed ginger extracts by using TRPV1-expressing cells (See Supporting Info.).

152

The results are summarized in Table 1. A wide variety of intensity values was observed

153

among the samples. Since the activity of a sample (NIB0790) was less than the

154

quantitative range (Fig. S1), it was excluded from the subsequent multivariate analysis.

155

The TRPV1-stimulating activity of the remaining 25 samples had the mean 51.4%, the

156

maximum 84.8% (NIB1115) and the minimum 19.1% (NIB0789).

157

Quantitative range and reproducibility in the LC-HRMS analysis

158

In the LC-HRMS data acquisition for the multivariate analysis, the quantitative

159

range and repeatability were important factors. We therefore estimated these factors in

160

advance. Representative total ion current (TIC) chromatograms in ginger and

161

processed ginger extracts are shown in Supplementary Figure S2. In the

162

chromatograms, we used the highest peak with m/z 277.1795 of the quasi-molecular

11

ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

Page 12 of 36

163

ion peak (12.9 min; peak A) and the relative lower peak with m/z 357.1689 (10.8 min;

164

peak B) for the estimation. The calibration curves were prepared from the peak area of

165

the mass chromatogram of each peak on the sample solution ranging from 0.005 to 10

166

mg/mL (Fig. S3a,b). The appropriate linearity (R2 >0.99) was obtained in the range of

167

0.005–10 mg/mL for peak A and 0.01–10 mg/mL for peak B. Based on this result, the

168

sample concentration for the LC-HRMS analysis was determined at 1 mg/mL. We also

169

analyzed the same samples in triplicate for the verification of data reproducibility. As a

170

result, there were no significant changes in the retention time of each peak, and high

171

reproducibility was confirmed (Fig. S3c).

172

The advantage of mass spectrometry is that a great deal of molecular weight

173

information is available, with high sensitivity.19 However, due to this property, it is

174

thought that an adequate concentration of sample solution in data acquisition is

175

important to avoid ion saturation and the incorporation of the peaks attributed to the

176

carry-over. In fact, some reports strongly suggest the possibility of misidentifying a

177

peak derived from noise as the specific peak of a sample in a multivariate analysis

178

using mass spectrometry.15, 20 In the present study, the optimal concentration of sample

179

solution and the repeatability of the LC-MS analysis results were assured by the

180

above-described investigation concerning the quantitative range and the repeatability.

12

ACS Paragon Plus Environment

Page 13 of 36

Journal of Agricultural and Food Chemistry

181

Peak alignment of the chromatogram and the peak extraction

182

We aligned the chromatograms obtained from LC-HRMS analysis using

183

Progenesis QI. This is an indispensable process for the multivariate analysis of LC-MS

184

and GC-MS data to fix the gap of the retention time among each chromatogram.17, 21

185

We mixed all 26 sample solutions in equal amounts and used the mixture for the

186

reference solution in the alignment process, which resulted in high concordance

187

(>90%) without manual correction (Fig. S5). We feel that the high concordance was

188

attributable mostly to the adequate sample concentration and the confirmation of

189

repeatability, as well as an effect of the reference solution.

190

We then used two methods to examine the peak extraction. The first method was

191

automatic extraction with the software. This method is performed based on algorithm

192

of progenesis QI, which can choose five phases as the threshold. The second method

193

was extraction based on the peak intensity of 1%, 0.1% and 0.01% for the base peak,

194

where the peaks with the higher intensity than the threshold (1, 0.1 and 0.01% to

195

strongest peak on TIC) was extracted. Consequently, 16,382 peaks were extracted by

196

the automatic extraction method. With the second method, 226 peaks for the 1%

197

condition, 2,302 peaks for the 0.1% condition and 9,321 peaks for the 0.01% condition

198

were extracted.

13

ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

Page 14 of 36

199

We selected the optimal extracted peak number according to the separation

200

degree of the ginger and processed ginger extracts in the PLS-DA. In the automatic

201

extraction, 0.01% and 0.1% conditions, the ginger and processed ginger were classified

202

into each group at the same degree (Figs. S4a, S4b and S4c). However, the separation

203

was insufficient (Fig. S4d) when the data matrix from 1% condition was used.

204

Considering the balance between the prediction precision and the risk of the carry-over,

205

we chose the moderate condition, 0.01% of base peak (Fig. S4b). This result indicates

206

that the greater the number of peaks extracted, the more clearly the separation becomes.

207

However, excessive peak extraction could induce the incorporation of a carry-over

208

peak. We therefore decided that the optimal peak number was 9,321, obtained from the

209

moderate condition, 0.01% of base peak.

210

Regression parameters of the TRPV1-stimulating activity prediction models

211

The TRPV1-stimulating activity value of each sample was added to the

212

above-mentioned data matrix. For the construction of the TRPV1-stimulating activity

213

prediction model, we applied the data matrix to a PLS or OPLS regression analysis in

214

which the LC-MS data and the activity values were treated as explanatory and

215

objective variables, respectively. We also used orthogonal signal correction (OSC) in

216

the PLS analysis to remove the explanatory variables that were not correlated with the

14

ACS Paragon Plus Environment

Page 15 of 36

Journal of Agricultural and Food Chemistry

217

objective variables.22,

23

218

mean-center (Ctr) as pretreatment parameters in the PLS, OPLS and PLC with OSC

219

analyses. For transformation of the explanatory variables, none or Log10 were used.

220

The number of latent variables was fixed at three in all models.

We selected unit variance (UV), pareto (Prt) and the

221

To evaluate all of the models obtained in the combination of each regression

222

formula and parameter, we checked the multiple correlation coefficient (R2) and

223

prediction ability parameter (Q2) of the resultant models and performed the

224

permutation test. R2 and Q2 values show the linearity and the prediction ability of

225

model, respectively. A prediction model with an R2 of >0.65 and a Q2 value of >0.5

226

is regarded as adequate for quantitative prediction.24

227

A permutation test is used to validate the incidence of over-fitting in a prediction

228

model.24 In this test, the provisional prediction models are constructed based on

229

various data matrixes in which objective and explanatory variables are randomly

230

combined many times and then the R2 and Q2 of each provisional model are calculated.

231

The correlation coefficient between the original data matrix and the permuted data

232

matrix versus both R2 and Q2 are plotted in the x- and y-axis, respectively. The

233

y-intercept of the regression line in the plot is used for the estimative index of

234

over-fitting: generally, R2 0.65 and Q2 values >0.5 were achieved in the PLS, OPLS and PLS

264

with OSC analyses. However, the PLS and OPLS analyses had RMSEP (%) values

265

ranging from 0.728 to 57.942 in several samples, whereas the RMSEP (%) of the PLS

266

with OSC analysis was lower mostly than the PLS and OPLS analyses except for three

267

samples (NIB794, NIB1115 and NIB008). The lower RMSEP (%) means that the

268

prediction precision is high. It was thus suggested that the prediction precision of PLS

269

with OSC was highest among the three models created in the present study. In the

270

evaluation of models by the 'leave one out' cross-validation, we felt that it was

17

ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

Page 18 of 36

271

important to estimate the error between the validation and test sets by using the

272

RMSEP as well as R2 and Q2 values reported to date. We also predicted

273

TRPV1-stimulating activity values when eliminating two samples. Representative

274

regression lines are shown in Figure 2a,b. In the cross-validation for the model in

275

which two samples were eliminated from the data matrix, the prediction precision

276

made little difference with that when one sample removed, indicating the high

277

robustness of the prediction models. In addition, the cross-validation for the model in

278

which six samples were randomly eliminated from the data matrix afforded very high

279

prediction precision (Fig. 2c).

280

Based on the above results, we finally selected the parameters of the

281

TRPV1-stimulating activity prediction models as follows: PLS with OSC; scale, Prt;

282

transform, and none. This prediction model will be very useful for the quality control

283

of health foods derived from ginger.

284

The search for TRPV1-stimulating activity components

285

After our successful construction of a TRPV1-stimulating activity prediction

286

model with high precision, we searched for the components that are responsible for

287

TRPV1-stimulating activity by analyzing the loading plot, the variable importance in

288

the projection (VIP), and the s-plot of the model. As a result, two peaks with m/z

18

ACS Paragon Plus Environment

Page 19 of 36

Journal of Agricultural and Food Chemistry

289

277.1795 and m/z 357.1689 were predicted as the active components from all

290

approaches using the loading plot, VIP and s-plot (Fig. 3). In these approaches, the

291

explanatory variables that most significantly contributed to the TRPV1-stimulating

292

activity are plotted at the top or the edge (e.g., m/z 277.1795 of the blue circles and m/z

293

357.1689 of the red circles in Figs. 3a–c). The peak with m/z 277.1795 was estimated

294

to be [6]-gingerol (1) from the exact mass measurements. m/z 277.1795 was regarded

295

as the quasi-molecular ion of [6]-shogaol (2) or as the ion derived from the dehydration

296

of [6]-gingerol (1).

297

The LC-MS analysis of the authentic 1 afforded the consistent retention time

298

and mass spectrum with this peak. It was thus identified as [6]-gingerol (1). This

299

compound was already reported as the active component.12,

300

compound was led as the first candidate of the active compound in each model of the

301

PLS with OSC and OPLS analyses shows that these models were useful for the search

302

of the active components. In the loading plot, VIP and s-plot, the variables around the

303

blue circles and red circles in Figure 3a–c were fragment and adduct ions derived from

304

[6]-gingerol (1). Next, the compound corresponding to the peak with m/z 357.1689 was

305

isolated from processed ginger and was identified as hexahydrocurcumin (3) (Fig. 1)

306

based on the NMR and MS data (See Supporting Info.). We plotted the

28

The fact that this

19

ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

Page 20 of 36

307

TRPV1-stimulating activity of each sample and the peak area of [6]-gingerol (1) and

308

hexahydrocurcumin (3) in the MS chromatogram. Those plots were added as Fig. 6S.

309

Consequently, both 1 and 3 showed the positive correlation to the activity (Figs. S6a

310

and S6b). In addition, this tendency was almost same in the plot for sum of the peak

311

areas of 1 and 3 (Fig. S6c). Therefore, it is thought that 1 and 3 in hot water extract

312

showed the activity in the additive manner. We also measured the TRPV1-stimulating

313

activity of not only 1 and 3 but also [6]-shogaol (2) and curcumin (4) as referential

314

compounds. Each EC50 value was given as follows: 0.525 µM for [6]-shogaol; 4.167

315

µM for [6]-gingerol; 69.91 µM for hexahydrocurcumin; inactive for curcumin.

316

Although [6]-shogaol has potent TRPV1-stimulating activity among them,12 its content

317

in the hot water extract was much smaller than [6]-gingerol and hexahydrocurcumin.

318

We think that [6]-shogaol was not picked up as the active compound due to its low

319

content and not so much contribute to the activity.

320

In conclusion, we succeeded in the construction of TRPV1-stimulating activity

321

prediction models with high precision by optimization using the estimative indexes for

322

the PLS analysis such as the R2, Q2 and permutation test. In our search for the

323

components that contribute to TRPV1-stimulating activity based on the loading plot

324

and s-plot for the resultant model, [6]-gingerol (1) and hexahydrocurcumin (3) were

20

ACS Paragon Plus Environment

Page 21 of 36

Journal of Agricultural and Food Chemistry

325

identified as the active components. Hexahydrocurcumin (3) isolated from ginger also

326

indicated TRPV1-stimulating activity, with the EC50 of 69.91 µM. This is first report

327

about the TRPV1-stimulating activity of hexahydrocurcumin. A series of the

328

procedures mentioned above could contribute to the survey of new pharmacologically

329

active components as well as quality evaluations of natural products.

330

ACKNOWLEDGEMENTS

331

This study was supported by a research grant from the Japan Agency for Medical

332

Research and Development.

333

ABBREVIATIONS USED

334

TRPV1, transient receptor potential vanilloid subtype 1; PLS, partial least squares

335

projections to latent structures; PLS-DA, partial least squares projections to latent

336

structures discrimination analysis; R2, multiple correlation coefficient; Q2, prediction

337

ability parameter; RMSEP, root mean square error of prediction; UV, Unit Variance;

338

Prt, pareto; Ctr, mean-center; LC-HRMS, liquid chromatography/high-resolution mass

339

spectrometry; EC50, half-maximal effective concentration

340

SUPPORTING INFORMATION DESCRIPTION

341

Figure S1; Fluorescence intensity curve for capsaicin.

342

Figure S2; Representative TIC chromatograms obtained from ginger (a) and processed

21

ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

Page 22 of 36

343

ginger extracts (b).

344

Figure S3; The calibration curve from mass chromatograms (a) m/z 277.1795 and (b)

345

m/z 357.1698, and (c) verification of data reproducibility.

346

Figure S4; The separation degree of ginger and processed ginger extracts in the

347

PLS-DA. (a) automatic extraction, (b) 0.01%, (c) 0.1% and (d) 1% conditions.

348

Additionally, the details on chemicals used, the isolation methods and the assignment

349

of hexahydrocurcumin.

350

Figure S5; Peak alignment of the chromatograms for all samples by using reference

351

solution with progenesis QI (broken red line is >90% for concordance rate).

352

Figure S6; The relationship of MS peak area vs TRPV1-stimulating activity. (a)

353

[6]-gingerol, (b) hexahydrocurcumin and (c) the total area of [6]-gingerol and

354

hexahydrocurcumin.

22

ACS Paragon Plus Environment

Page 23 of 36

Journal of Agricultural and Food Chemistry

355

REFERENCES

356

1. Ruchi, B. S.; Deepak K. S.; Sandra, C.; Alvaro, M. V. Gingerols and shogaols:

357

Important nutraceutical principles from ginger. Phytochemistry 2015, 117,

358

554-568.

359

2. Vinay, Kr. S.; Pragya, Y., Narender, T. Recent advances in the synthesis, chemical

360

transformations and pharmacological studies of some important dietary spice's

361

constituents. Chem. Biol. Inter. 2014, 4(2), 66-99.

362

3. Wohlmuth, H. Phytochemistry and pharmacology of plants from the ginger family,

363

Zingiberaceae, ePublications: Lismore, Australia, Southern Cross University, 2008,

364

pp. 16-20.

365

4. Muhammad, S. M.; Yu-Ming, N.; Amy, L. R.; Michael, K.; Arindam, R.;

366

Marie-Pierre, S. Ginger consumption enhances the thermic effect of food and

367

promotes feelings of satiety without affecting metabolic and hormonal parameters

368

in overweight men: A pilot study. Metabolism 2012, 61(10), 1347-1352.

369

5. Arpad, S.; Daniel, N. C.; Charles, A. B.; Samer, R. E. The vanilloid receptor

370

TRPV1: 10 years from channel cloning to antagonist proof-of-concept. Drug

371

Discovery 2007, 6, 357-372.

372

6. David, J. TRP Channels and Pain. Cell and Dev. Biol. 2013, 29, 355-384.

23

ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

373

374

Page 24 of 36

7. Bradford, B. L.; Bruce, M. S. Towards a molecular understanding of adaptive thermogenesis. Nature 2000, 404, 652-660.

375

8. Kawada, T.; Sakabe, S.; Aoki, N.; Watanabe, T.; Higeta, K.; Iwai, K.; Sugimoto, E.

376

Intake of sweeteners and pungent ingredients increases the thermogenin content in

377

brown adipose tissue of rat. J. Agric. Food Chem. 1991, 39, 651-654.

378

9. Tominaga, M., Caterina, MJ., Malmberg, AB., Rosen, TA., Gilbert, H., Skinner, K.,

379

Raumann, BE.; Basbaum, AI.; Julius, D. The cloned capsaicin receptor integrates

380

multiple pain-producing stimuli. Neuron 1998, 21, 531-543.

381

10. Iwasaki, Y.; Morita, A.; Iwasawa, T.; Kobata, K.; Sekiwa, Y.; Morimitsu, Y.;

382

Kubota,

K.;

Watanabe

T.

A

nonpungent

component

of

steamed

383

ginger-[10]-shogaol-increases adrenaline secretion via the activation of TRPV1.

384

Nutr. Neurosci. 2006, 9, 169–178.

385

11. Morera, E.; De Petrocellis, L.; Morera, L.; Moriello, A.S.; Nalli, M.; Di Marzo, V.;

386

Ortar, G. (2012). Synthesis and biological evaluation of [6]-gingerol analogues as

387

transient receptor potential channel TRPV1 and TRPA1 modulators. Bioorga. Med.

388

Chem. Lett. 2012, 22, 1674–1677.

389

12. Morita, A.; Iwasaki, Y.; Kobata, K.; Yokogoshi, H.; Watanabe, T. Newly

390

Synthesized Oleylgingerol and Oleylshogaol Activate TRPV1 Ion Channels. Biosci.

24

ACS Paragon Plus Environment

Page 25 of 36

Journal of Agricultural and Food Chemistry

391

Biotechnol. Biochem. 2007, 71 (9), 2304-2307.

392

13. Ohkawara, S.; Tanaka-Kagawa, T.; Furukawa, Y.; Nishimura, T.; Jinno, H.

393

Activation of the human transient receptor potential vanilloid subtype 1 by

394

essential oils. Biol. Pharm. Bull. 2010, 33(8) 1434-1437.

395

14. Xue, J.; Il-Hwan, O.; Seul-Gi, L.; Hyung-Kyoon, C. The application of

396

metabolomics to processed traditional Chinese medicine. J. Korean Soc. Appl. Bio.

397

Chem. 2013, 56, 475-481.

398

15. Augustin, S.; Lorraine, B.; Oliver, F.; Thomas, H.; Bruce, S. K.; Ben van, O.;

399

Estelle, P.; Elwin, V.; David, W.; Suzan, W. Mass-spectrometry-based

400

metabolomics: limitations and recommendations for future progress with particular

401

focus on nutrition research. Metabolomics 2009, 5, 435-458.

402

16. Tarachiwin, L.; Masako, O.; Fukusaki, E. Quality evaluation and prediction of

403

Citrullus lanatus by 1H-NMR-based metabolomics and multivariate analysis. J.

404

Agric. Food Chem. 2008, 56(14), 5827-5835.

405

17. Fujimura, Y.; Kurihara, K.; Ida, M.; Kosaka, R.; Miura, D.; Wariishi, H.;

406

Maeda-Yamamoto, M.; Nesumi, A.; Saito, T.; Kanda, T.; Yamada, K.; Tachibana, H.

407

Metabolomics-driven nutraceutical evaluation of diverse green tea cultivars. PLOS

408

One 2011, 6(8), e23426.

25

ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

Page 26 of 36

409

18. Pongsuwan, W.; Bamba, T.; Harada, K.; Yonetani, T.; Kobayashi, A.; Fukusaki, E.

410

High-Throughput Technique for Comprehensive Analysis of Japanese Green Tea

411

Quality Assessment Using Ultra-performance Liquid Chromatography with

412

Time-of-Flight Mass Spectrometry (UPLC/TOF MS). J. Agric. Food Chem. 2008,

413

56, 10705-10708.

414

19. Makarov, A.; Denisov, E.; Lange, O.; Horning, S. Dynamic range of mass accuracy

415

in LTQ orbitrap hybrid mass spectrometer. J. Am. Soc. Mass Spectrom. 2006, 17,

416

977-982.

417

418

20. Bin, Z.; Jun, F.; Leepika, T.; Habtom, W. R. Mass-spectrometry-based metabolomics. Mol. Biosyst. 2012, 8(2), 470-481.

419

21. Tsugawa, H.; Tsujimoto, Y.; Arita, M.; Bamba, T.; Fukusaki, E. GC/MS based

420

metabolomics: development of a data mining system for metabolite identification

421

by using soft independent modeling of class analogy (SIMCA). BMC

422

Bioinformatics 2011, 12, 131-143.

423

424

425

426

22. Wold, S.; Antti, H.; Lindgren, F.; Ohman, J. Orthogonal signal correction of near-infrared spectra. Chemometr. Intell. Lab. Syst. 1998, 44, 175-185. 23. Trygg, J.; Holmes, E.; Lundstedt, T. Chemometrics in metabonomics. J. Proteome Res. 2007, 6 469-479.

26

ACS Paragon Plus Environment

Page 27 of 36

Journal of Agricultural and Food Chemistry

427

24. Mohamed, N. T.; Laurence, L. Moyec.; Roland, A.; Corentine, G.; Nadia, B.; Pierre,

428

N.; Douglas, N.; Philippe, S. PLS/OPLS models in metabolomics: the impact of

429

permutation of dataset rows on the K-fold cross-validation quality parameters. Mol.

430

Biosyst. 2015, 11, 13-19.

431

25. Wiklund, S.; Johansson, E.; Sjöström, L.; Mellerowicz, EJ.; Edlund, U.; Shockcor,

432

P.; Gottfries, J.; Moritz, T.; Trygg, J. Visualization of GC/TOF-MS-Based

433

metabolomics data for identification of biochemically interesting compounds using

434

OPLS class models. Anal. Chem. 2008, 80(1), 115-122.

435

26. Van den Berg, R.; Hoefsloot, H.; Westerhuis, J.; Smilde, A.; Van der Werf, M.

436

Centering, scaling, and transformations: improving the biological information

437

content of metabolomics data. BMC Genomics 2006, 7, 142-157.

438

27. Bowei, X.; Haiwei, G.; Hamid, B.; Daniel, R. Statistical analysis and modeling of

439

mass spectrometry-based metabolomics data. Methods Mol. Biol. 2014, 1198,

440

333-353.

441

28. Riera, CE.; Menozzi-Smarrito, C.; Affolter, M.; Michlig, S.; Munari, C.; Robert, F.;

442

Vogel, H.; Simon, SA.; Coutre, J. Compounds from Sichuan and melegueta

443

peppers activate, covalently and non-covalently, TRPA1 and TRPV1 channels. Brit.

444

J. Pharmacol. 2009, 157, 1398-1409.

27

ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

445

446

447

448

Page 28 of 36

29. Eriksson, L. Multi- and megavariate data analysis principles and applications, Umeå, Sweden, Umetrics academy, 2001. 30. Marcia, M. C. F. Encyclopedia of Physical Organic Chemistry, Hoboken, USA, Wiley Online Library, 2017, pp. 2041-2058.

449

FIGURE CAPTIONS

450

Fig. 1. The chemical structures of the components tested for TRPV1-stimulating

451

activity. The TRPV1-stimulating activity of [6]-gingerol (1) and hexahydrocurcumin

452

(3) predicted by the TRPV1-stimulation activity prediction models was measured in

453

vitro. In addition, [6]-shogaol (2) and curcumin (4) were measured as referential

454

compounds.

455

456

Fig. 2. The prediction of two or multiple samples by using the PLS with OSC

457

regression model. The prediction precision for the PLS with OSC regression model

458

was evaluated by the 'leave one out' cross-validation. (a) Two ginger samples (NIB 055

459

and 091), (b) two processed ginger samples (NIB 797 and 802), and (c) multiple

460

samples (NIB 796, 798, 802, 147, 169 and 179) were used as the test set, respectively.

461

The predicted samples by using 'leave out' cross-validation were indicated as the blue

462

dots.

28

ACS Paragon Plus Environment

Page 29 of 36

Journal of Agricultural and Food Chemistry

463

464

Fig. 3. Search of the components responsible for TRPV1 stimulating activity by

465

analyzing

466

transformations: none), (b) VIP (regression formula: PLS, pretreatment: Prt,

467

transformations: none) and (c) s-plot (regression formula: OPLS, pretreatment: Prt,

468

transformations: none) of the models. Two peaks with m/z 277.1795 (blue circle or bar)

469

and m/z 357.1689 (red circle or bar) were predicted as the active components from all

470

approaches.

(a)

loading

plot

(regression

formula:

PLS,

pretreatment:

Prt,

471

472

473

474

475

476

477

478

479

480

29

ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

Page 30 of 36

Tables

Table 1. Details of the samples used and their TRPV1-stimulating activity Sample No. Crude drug name NIB-008 ginger NIB-039 ginger NIB-055 ginger NIB-060 ginger NIB-091 ginger NIB-110 ginger NIB-147 ginger NIB-169 ginger NIB-179 ginger NIB-0787 processed ginger NIB-0788 processed ginger NIB-0789 processed ginger NIB-0790 processed ginger NIB-0791 processed ginger NIB-0792 processed ginger NIB-0793 processed ginger NIB-0794 processed ginger NIB-0795 processed ginger NIB-0796 processed ginger NIB-0797 processed ginger NIB-0798 processed ginger NIB-0799 processed ginger NIB-0800 processed ginger NIB-0801 processed ginger NIB-0802 processed ginger NIB-1115 processed ginger

Place of collection Yunnan Yunnan Yunnan Yunnan Yunnan Yunnan Yunnan Yunnan Yunnan unknown Guangxi Guangxi Zhuang Autonomous Region Sichuan Guangdong Guangdong Guangdong Sichuan Guangdong Guangdong Guangdong Guangdong Guangdong Guangdong Yunnan Guangxi Guangxi Zhuang Autonomous Region

Form whole whole whole pieces slice pieces whole whole whole raw whole pieces whole whole whole whole whole whole whole whole whole whole whole pieces pieces pieces

Collection year 2010 2009 2010 2008 2009 2010 2010 2009 2010 unknown unknown 2013 1990 1996 2002 2002 2002 2006 2009 2009 2010 2011 2012 2013 2013 2012

TRPV1 activity 57.8 54.9 48.9 81.1 57.0 66.4 59.9 58.0 82.0 80.8 73.0 19.1 16.3 26.6 32.4 42.0 20.4 23.2 37.9 32.7 23.0 81.3 25.4 62.0 55.6 84.8

30

ACS Paragon Plus Environment

Page 31 of 36

Journal of Agricultural and Food Chemistry

Table 2. The estimation index results for all of the prediction models Regression formula PLS Pretreatment Transformation Prt None Ctr None UV None Prt Log Ctr Log UV Log Regression formula OPLS Pretreatment Transformation Prt None Ctr None UV None Prt Log Ctr Log UV Log Regression formula PLS with OSC Pretreatment Transformation Prt None Ctr None UV None Prt Log Ctr Log UV Log

Estimative index 2

Permutaiton test

2

R

Q

0.967 0.938 0.977 0.975 0.978 0.974

0.952 0.925 0.957 0.947 0.946 0.946

2

R 0.230 0.058 0.400 0.701 0.726 0.650

2

Q -0.383 -0.287 -0.407 -0.032 0.023 -0.138

Estimative index 2

Permutaiton test

2

R

Q

0.978 0.949 0.988 0.988 0.990 0.987

0.969 0.942 0.979 0.971 0.975 0.973

2

R 0.027 0.111 0.483 0.773 0.803 0.750

2

Q -0.586 -0.343 -0.656 -0.476 -0.450 -0.510

Estimative index 2

2

R

Q

0.995 0.993 0.994 0.990 0.989 0.991

0.990 0.987 0.988 0.920 0.920 0.921

Permutaiton test 2

R 0.325 0.228 0.408 0.922 0.910 0.929

2

Q -0.416 -0.420 -0.434 -0.212 -0.220 -0.130

31

ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

Page 32 of 36

Table 3. The R2, Q2 and RMSEP (%) results for the 'leave one out' cross-validation 2

NIB787 NIB788 NIB789 NIB791 NIB792 NIB793 NIB794 NIB795 NIB796 NIB797 NIB798 NIB799 NIB800 NIB801 NIB802 NIB1115 NIB008 NIB039 NIB055 NIB060 NIB091 NIB110 NIB147 NIB169 NIB179 a

PLS 0.965 0.965 0.965 0.969 0.966 0.970 0.966 0.966 0.970 0.966 0.968 0.968 0.965 0.977 0.972 0.967 0.967 0.968 0.970 0.965 0.970 0.966 0.969 0.969 0.963

R OPLS 0.978 0.977 0.976 0.982 0.978 0.981 0.977 0.978 0.980 0.977 0.978 0.982 0.976 0.985 0.981 0.977 0.979 0.979 0.982 0.978 0.982 0.977 0.978 0.982 0.975

2

PLS with OSC 0.995 0.995 0.995 0.995 0.995 0.996 0.995 0.995 0.996 0.995 0.995 0.994 0.995 0.997 0.995 0.996 0.996 0.995 0.996 0.996 0.995 0.995 0.995 0.995 0.995

PLS 0.945 0.946 0.946 0.951 0.948 0.954 0.949 0.949 0.955 0.950 0.951 0.949 0.948 0.964 0.958 0.948 0.950 0.951 0.954 0.952 0.960 0.950 0.955 0.955 0.947

Q OPLS 0.969 0.969 0.967 0.972 0.970 0.974 0.969 0.970 0.974 0.970 0.971 0.973 0.969 0.980 0.976 0.971 0.974 0.974 0.973 0.971 0.973 0.968 0.970 0.974 0.966

a

PLS with OSC 0.987 0.988 0.988 0.989 0.989 0.990 0.988 0.988 0.990 0.988 0.987 0.985 0.986 0.990 0.988 0.987 0.990 0.989 0.987 0.991 0.991 0.990 0.990 0.990 0.988

PLS 4.443 9.416 34.344 17.219 8.772 25.435 38.576 57.942 18.355 7.166 39.118 18.153 6.285 17.689 17.706 3.925 2.392 7.193 5.275 36.296 18.472 2.911 12.880 10.716 10.587

RMSEP(%) OPLS 6.285 12.836 32.691 13.934 13.556 17.823 35.958 43.331 19.237 2.239 41.486 16.719 3.227 19.659 19.484 3.474 5.806 7.132 4.939 32.934 20.027 0.728 12.215 10.288 12.024

PLS with OSC 1.629 9.888 26.346 6.346 13.303 7.356 39.060 40.920 9.927 3.176 9.458 8.617 2.044 8.541 3.245 4.111 8.205 3.951 0.430 28.890 3.902 0.500 3.153 2.031 5.685

The values of RMSEP were expressed as RMSEP (%) by dividing with TRPV1-stimulating activity.

32

ACS Paragon Plus Environment

Page 33 of 36

Journal of Agricultural and Food Chemistry

Figure 1 Yoshitomi et al.

33

ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

Page 34 of 36

Figure 2 Yoshitomi et al.

34

ACS Paragon Plus Environment

Page 35 of 36

Journal of Agricultural and Food Chemistry

Figure 3 Yoshitomi et al.

35

ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

Page 36 of 36

For Table of Contents Use Only

The construction of prediction models for the TRPV1-stimulating activity of ginger and processed ginger based on LC-HRMS data and PLS regression analyses

36

ACS Paragon Plus Environment