Rapid and Nondestructive Determination of ... - ACS Publications

Feb 7, 2017 - Pearling Fractions of Barley by Near-Infrared (NIR) and Fluorescence ... Different varieties of barley were pearled, and dust fractions ...
1 downloads 0 Views 1MB Size
Subscriber access provided by Washington University | Libraries

Article

Rapid and Non-Destructive Determination of Aleurone Content in Pearling Fractions of Barley by NIR and Fluorescence Spectroscopy Jens Petter Wold, Diego Airado-Rodríguez, Ann Katrin Holtekjolen, Ulla Riikka Maria Holopainen-Mantila, and Stefan Sahlstrom J. Agric. Food Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jafc.6b04954 • Publication Date (Web): 07 Feb 2017 Downloaded from http://pubs.acs.org on February 7, 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 33

Journal of Agricultural and Food Chemistry

1

Rapid and Non-Destructive Determination of Aleurone Content in

2

Pearling Fractions of Barley by NIR and Fluorescence Spectroscopy

3

Jens Petter Wold†*, Diego Airado-Rodríguez†, Ann-Katrin Holtekjølen†, Ulla Holopainen-

4

Mantila‡, Stefan Sahlstrøm†

5 6

† – Nofima AS, Norwegian Institute for Food and Fisheries Research, Tromsø, Norway

7

‡ – VTT Technical Research Centre of Finland, Espoo, Finland

8

Diego Airado ([email protected]); [email protected];

9

[email protected]; Holopainen-Mantila Ulla ([email protected])

10

11

*Corresponding author (Tel: +47 95979749; Fax: +47 64940333; E-mail:

12

[email protected])

13

14

15

16

17

18

19

20 1 ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

21

Abstract

22

The aim of the present work was to develop and evaluate near-infrared (NIR) and

23

fluorescence spectroscopy as rapid and potential on-line methods for determination of the

24

amount of aleurone in pearling dust fractions of barley. Phytic acid was used as marker for the

25

aleurone layer. Different varieties of barley were pearled and dust fractions were

26

progressively taken out. Sample concentration of phytic acid varied in the range 0.5 – 4.1

27

g/100g and highest concentrations were obtained in dust fractions for pearling degrees in the

28

range 15 - 25%. Regression models for phytic acid were developed with the same high

29

correlations for NIR and fluorescence spectroscopy (R2=0.96) and prediction errors of ±0.16 –

30

0.18 g/100g. The models performed well on a test set of pearling fractions from two other

31

barley varieties. The techniques are rapid and non-destructive, which means that they can be

32

used on-line in connection with industrial pearling systems.

33

34

2 ACS Paragon Plus Environment

Page 2 of 33

Page 3 of 33

Journal of Agricultural and Food Chemistry

35

Introduction

36

Currently, barley processing mainly consists of de-hulling, pearling and then milling to a

37

flour. De-hulling removes the hull and also portions of the pericarp and testa. Pearling the de-

38

hulled barley grains further removes the remaining hull, pericarp and testa, aleurone layers

39

and germ.1 De-hulling and pearling are accomplished by abrasion, which subjects the grains

40

to a rotating abrasive surface. Pearled barley typically makes up 60-70% of the total grain

41

weight,2 while the byproduct, the pearling flour (30-40%), is mainly used for animal feeds.

42

This process is difficult to optimize to increase process yield, that is, obtain as much flour as

43

possible with a minimum or a desired amount of fractions from the outer layers. Different

44

properties of the barley grains, for instance different varieties, requires different pearling

45

percentages to achieve high quality flour and a minimum of byproduct. Practice is to adjust

46

the pearling process according to experience or habit. Optimization of pearled barley yield

47

and barley flour quality can, however, be solved by establishment of a rapid online method for

48

monitoring of the pearling dust composition. Such measurements could also be used to detect

49

and quantify particularly nutritional pearling fractions for better utilisation. In this study we

50

focussed on the determination of aleurone content in the pearling dust.

51 52

Barley is known to contain certain health components. The aleurone layer is composed mainly

53

of arabinoxylans (67 – 71%) and beta-glucans (26%) as well as phenolic acids, vitamin E, B-

54

vitamins, protein, triacylglycerol, minerals, phytate and sugars.3 The phenolic compounds

55

such as ferulic acid after uptake in the gut, are transported to the blood and can possibly help

56

to reduce the destructive effect of free radicals.4 Barley is also rich in beta-glucan, a

57

polysaccharide building block of the cell walls in aleurone layer and the endosperm, which is

58

approved by EFSA to reduce the cholesterol level and sugar in blood after a meal.5

59

Holtekjølen et al.6 found that the pearling flour is a potential natural source of phytochemicals 3 ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

60

with antioxidant effects. Chemical characterization of different pearling fractions of barley

61

was done by Blandino et al.,7 who found that the total antioxidant activity (TAA) in the 15-

62

25% pearling fraction was significantly higher than in the other fractions. The inclusion of

63

this fraction in preparation of bread increased TAA and dietary fibers without impairing the

64

physical bread properties. For milling companies it can be important to be able to quantitate

65

the amount of kernel tissues and health promoting components in their milling fractions, for

66

example the content of hull, soluble fibre (β-glucan) and aleurone layer, and in that way create

67

a competitive advantage in terms of diversifying milling and bakery product ranges.

68

Ultimately, the consumers will benefit from a naturally produced healthier product.

69 70

Barron et al.8 isolated barley grain tissues from both hulled and hulless barley for chemical

71

analyses. They found that phytic acid is specifically localized in the aleurone layer and that it

72

can be used as a marker for this tissue among other grain tissues. They also identified the

73

phenolic compound p-Coumaric acid as an efficient marker of the hull and proposed that it

74

could be used to monitor the dehulling process.

75 76

In order to perform continuous product or process control, rapid analytical techniques are

77

required. Traditional chemical methods typically require several hours or days, and such a

78

time span is rarely applicable in modern production. Spectroscopic instrumentation has gained

79

popularity in processing industries during the last 2-3 decades, being rapid, reliable and non-

80

destructive analytical techniques. For online use in food production, near-infrared (NIR)

81

spectroscopy is probably the most widely used spectroscopic technique. The near infrared part

82

of the electromagnetic spectrum comprises overtones and combination bands of fundamental

83

bond vibrations found in the infrared region. The molecular bonds that absorb in the NIR

84

region are mainly -O-H, -C-H, -N-H and -S-H, reflecting molecular structural characteristics

4 ACS Paragon Plus Environment

Page 4 of 33

Page 5 of 33

Journal of Agricultural and Food Chemistry

85

of proteins, carbohydrates, lipids and water. NIR is established in the milling industry for at-

86

line and on-line determination of typically moisture, protein, fat and ash in flour and whole

87

grains. Promising results for determination of beta-glucan has been obtained.9 It is also

88

reported that phytic acid has a distinct absorption spectrum in the NIR region and that it can

89

be determined in green gram seeds and cotton seeds by NIR spectroscopy.10,11 NIR is

90

therefore a promising method for continuous monitoring of the pearling fractions, where the

91

amount of e.g. protein, beta-glucan and phytic acid will vary systematically through the

92

different kernel layers.7

93 94

It is well known that different tissues of the grains have characteristic fluorescent properties.

95

The strong bluish fluorescence from the aleurone layer comes mainly from ferulic acid.12

96

Jensen et al.13 showed that endosperm emits fluorescence in the UV typical for tryptophan,

97

and that pericarp has a distinct green-yellow fluorescence peaking at about 520 nm with

98

unknown origin. They also demonstrated that it is possible to quantitate the amount of

99

pericarp, aleurone, and endosperm in wheat milling fractions by the use of autofluorescence

100

spectra and multivariate calibration. The spectroscopic measurements were performed on 20

101

mg flour samples suspended in glycerol and water. Symons and Dexter14 used fluorescence

102

imaging in pilot mill streams to monitor aleurone and pericarp as an estimator of flour

103

refinement. Pericarp fluorescence correlated especially well with flour refinement and

104

indicated promise for on-line quality evaluation in mills.

105 106

The main aim of the present work was to develop and evaluate a rapid and preferably on-line

107

method for determination of the amount of aleurone layer in pearling fractions of barley. Such

108

a method could be used to control the pearling process toward maximum yield as well as

109

extract fractions with high aleurone content for use in certain products. We have tested and

5 ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

110

compared two rapid spectroscopic techniques, which both have potential for on-line use: NIR

111

and fluorescence spectroscopy. Different varieties of barley were pearled and dust fractions

112

were systematically taken out over time. As reference marker for the aleurone layer we used

113

phytic acid. Concentrations of different phenolic acids were also determined. Multivariate

114

regression as well as spectroscopic curve resolution were used to analyse the data and make

115

calibration models.

116 117 118

Materials and Methods

119

Materials

120

For the investigations, three separate sample sets of pearling dust were produced. The sample

121

sets were made as follows:

122

SampleSet1: A J.E. Barley Huller (Glenn.A. Burdick, Minneapolis, MN) pearler was used to

123

make 12 fractions each of two different barley varieties, 'Marigold' and 'Olve'. The pearler

124

consists of a wire brush that presses small particles through a sieve. A maximum of 20 grams

125

of grain could be pearled at the time. The dust fractions were made by collecting pearling dust

126

systematically according to time of pearling. The fractions were collected for 0-30, 30-60, 60-

127

90, and so on up to 330-360 s. A total of 24 samples were produced.

128

SampleSet2: A Testing Mill TM05 (Satake, Stafford, TX) with abrasive stone #30 was used to

129

make 9 fractions each from two parallels of 200 g of 'Marigold' kernels, and then 10 fractions

130

each of two 200 g parallels of 'Olve' kernels. A total of 38 samples were produced.

131

Progressive pearling fractions were taken out after approximate pearling degrees 9, 14, 18, 22,

132

26, 30, 33, 38, 42 and 49%.

6 ACS Paragon Plus Environment

Page 6 of 33

Page 7 of 33

Journal of Agricultural and Food Chemistry

133

SampleSet3: The same Satake Testing Mill TM05 was used to make 8 fractions each from two

134

parallels of 200 g of 'Pihl' kernels, and then 8 fractions each of two 200 g parallels of 'Pirona'

135

kernels. 'Pihl' and 'Pirona' are hulless barley varieties. A total of 32 samples were produced.

136

Progressive pearling fractions were taken out after approximate pearling degrees 5, 10, 15, 21,

137

26, 31, 36 and 40%.

138

Spectroscopic measurements

139

NIR spectra in reflection mode were obtained directly on intact dust samples with a DA7200

140

spectrometer (Perten Instruments, Hägersten, Sweden). Samples were disposed in a circular

141

sampling cup and measured area was approximately 30 cm2. Sampling time was set to 6 sec

142

and spectra were registered by covering every fifth wavelength in the region 950-1650 nm.

143

Each sample was measured in triplicate and the average spectrum was used for further

144

analysis.

145

Fluorescence spectra were also collected from the intact pearling dust samples. Samples were

146

placed in black circular sample cups, which exposed a flat circular surface with a diameter of

147

5 cm for the measurements. Measurements were carried out with a Fluoromax-4

148

spectrofluorometer (Horiba Scientific, Kyoto, Japan) in front face mode via a FL-300/FM4-

149

3000 bifurcated fiber-optic probe (Horiba Scientific). Distance between the probe head and

150

the sample was about 5 cm and created a circular measurement area of 40 mm diameter.

151

Probe and sample were covered by a black shield to avoid ambient straylight. Three different

152

excitation wavelengths were employed to register emission spectra of each sample. Samples

153

were sequentially illuminated with 470 nm, 350 nm and 280 nm excitation light and the

154

corresponding fluorescence emission spectra were registered in the ranges 490-650 nm, 400-

155

600 nm, and 300-540 nm, respectively. The three assayed excitation wavelengths were

156

systematically applied in decreasing order to minimize potential sample alteration by the UV

7 ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

157

excitation. The selected excitation wavelengths should enable detection of pericarp, aleurone

158

and endosperm, respectively.13,14

159

Determination of phytic acid

160

Phytic acid was used as a marker for the aleurone layer and was measured for all

161

samples. It was determined by means of a K-PHYT phytic acid (phytate)/total

162

phosphorus assay kit (Megazyme, Chicago, IL). This method involves acid extraction

163

of inositol phosphates followed by treatment with a phytase that is specific for phytic

164

acid (IP6) and the lower myo-inositol phosphate forms (i.e. IP2, IP3, IP4, IP5).

165

Subsequent treatment with alkaline phosphatase ensures the release of the final

166

phosphate from myo-inositol phosphate (IP1) which is relatively resistant to the action

167

of phytase. The total phosphate released is measured using a modified colourimetric

168

method and given as g of phosphorus/100 g of sample material.

169 170

Determination of phenolic acids

171

Total phenolic acids (ferulic acid, caffeic acid, p-coumaric acid and sinapic acid) were

172

determined for four series of pearling fractions, one from each of 'Olve', 'Marigold',

173

'Pihl' and 'Pirona'. Extraction of total phenolic acids was performed as described

174

earlier.15 Identification and quantitation of phenolic acids were carried out using a

175

UltiMate 3000 rapid separation high-performance liquid chromatography (HPLC)

176

system with a 3000 RS diode array detector (Dionex, Sunnyvale, CA) and an Acquity

177

UPLC BEH C8, 1.7 µm, 2.1 × 150 mm column (Waters, Milford, MA). A gradient

178

elution program was used with 1% acetic acid in Milli-Q water as eluent A and 1%

179

acetic acid in acetonitrile as eluent B. The injection volume was 10 µL, and the flow

180

rate was 0.450 mL/min, with a column temperature of 50 °C. The gradient elution

181

program was as follows: 5% B (0 min), 5% B (0.8 min), 10% B (1.2 min), 10% B (1.9 8 ACS Paragon Plus Environment

Page 8 of 33

Page 9 of 33

Journal of Agricultural and Food Chemistry

182

min), 15% B (2.4 min), 15% B (3.7 min), 21% B (4.0 min), 21% B (5.2 min), 27% B

183

(5.7 min), 50% B (8.0 min), 100% B (9.0 min), 100% B (12.5 min), 5% B (15.0 min),

184

5% B (20.0 min). Initial conditions were used for 10 min equilibration before a new

185

analysis was started. The identification and quantitation of phenolic acids were carried

186

out using external standards (caffeic, p-coumaric, ferulic, and sinapic acid) and UV

187

spectrum characteristics with the Chromeleon chromatography information

188

management System (Dionex).16

189 190

Microscopy of pearled barley grains

191

Selected grains of cultivars 'Marigold', 'Olve', 'Pihl' and 'Pirona' representing various pearling

192

times were subjected to microscopy to evaluate the effect of pearling on the aleurone layer. A

193

quick staining method was developed for this examination of aleurone layer in pearled grains.

194

Grains were cut in half cross-sectionally in the middle of the grain. Cross-cut surfaces of

195

individual grains were stained with 0.1% (w/v) acid fuchsin (BDH Chemicals Ltd., Poole,

196

UK) in 1% acetic acid for 1 min and rinsed twice with water. Stained grains were then dried

197

in dark for 30 min. The cross-cut surfaces were imaged using excitation wavelength (330-

198

380 nm) and emission from 420 nm with an AxioImager M.2 microscope (Carl Zeiss GmbH,

199

Göttingen, Germany). Under these lightning conditions, aleurone cell walls were detected as

200

bright blue due to autofluorescence of phenolic substances present in cell walls while starchy

201

endosperm appeared bright red. Micrographs were obtained using a Zeiss Axiocam 506 CCD

202

colour camera (Zeiss) and the Zen imaging software (Zeiss). In addition to epifluorescence

203

microscopy, the grains were imaged using a SteREO Discovery.V8 stereomicroscope

204

equipped with Achromat S 0.5x objective (Carl Zeiss MicroImaging GmbH) and an Olympus

205

DP-25 single chip colour CCD camera (Olympus Life Science Europa GmbH, Hamburg,

9 ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

206

Germany). Approximately 18 – 20 grains/sample were examined and the representative

207

images were chosen for publication.

208

209

Data analysis

210

Pretreatment of spectroscopic data

211

NIR spectra are affected by differences in sample scattering. Differences in scattering can

212

occur due to the use of different pearling systems. To reduce the effect of such phenomena the

213

NIR spectra were normalized by standard normal variate (SNV)17 prior to calibration.

214

The fluorescence spectra were normalised by unit area normalisation, meaning that the area

215

under each spectrum equals one. This was done separately for emission spectra for the

216

different excitations. This normalisation removes the main intensity variations and emphasises

217

the shapes of the spectra.

218

Calibration

219

Partial least squares regression (PLSR)18 was used to make calibrations between spectroscopic

220

data and the reference values of phytic acid. Separate and common models for samples in

221

SampeSet1 and SampleSet2 were made. SampleSet3 was used as a test set.

222

The optimal number of PLSR factors of the calibration models was determined by full cross

223 224

validation. The predicted value for each sample, ‫ݕ‬ො௜ , was compared with the reference value, yi (nominal phytic acid concentration). The squared multivariate correlation coefficient (R2),

225

which describes the relationship between the measured and the predicted values of the y

226

variable, and the prediction error of the regression model, expressed as root-mean-square

10 ACS Paragon Plus Environment

Page 10 of 33

Page 11 of 33

Journal of Agricultural and Food Chemistry

227

error of either cross-validation (RMSECV), were used to evaluate the model performance.

228

RMSECV is defined as: 1 ܴ‫ = ܸܥܧܵܯ‬ඩ ෍(‫ݕ‬௜ − ‫ݕ‬ො௜ )ଶ ܰ ே

௜ୀଵ

229

where i denotes the samples from i to N.

230

The obtained calibration models were used to predict phytic acid concentration in the test set

231

samples. Model performance was reported as the root-mean-square error of prediction

232 233

(RMSEP), calculated exactly as RMSECV (‫ݕ‬ො௜ being result of prediction and not cross-

validation), and the multivariate correlation coefficient (R2). PLSR and spectroscopic pre-

234

processing were performed with the software The Unscrambler, ver. 9.8 (Camo AS, Oslo,

235

Norway).

236

Multivariate curve resolution

237

Fluorescence spectra lend themselves to curve resolution. Multivariate curve resolution

238

(MCR), enables extraction of potential pure spectroscopic components originating from the

239

different inherent fluorophores and their respective concentrations. The obtained

240

concentrations for MCR-components were compared with reference analyses. A non-

241

negativity constraint was applied in the spectra and concentration modes in order to obtain a

242

realistic solution. The algorithm for MCR has previously been described in detail.19,20 MCR

243

was carried out with the software The Unscrambler, ver. 9.8 (Camo AS)

244

245

Results and discussion

246

Detection of the effect of pearling on grain layers by microscopy 11 ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

247

The effect of the pearling on barley grain outer layers was verified by microscopy. For this,

248

grains of four barley varieties studied with varying degree of pearling were examined. In

249

Figure 1A-E, the appearance of native and pearled barley kernels of 'Olve' and the cross-cut

250

surfaces of corresponding kernels stained with Acid Fuchsin are presented. Stereomicroscopy

251

images showed that pearling first removed the hull and proceeded then gradually towards the

252

centre parts of the grain. However, the ends of the kernels seemed to be more readily pearled

253

than the sides.

254

The images of the stained cross-cut surfaces revealed that aleurone appearing blue due to

255

autofluorescence was first removed from the ventral side of the grain, excluding the ventral

256

furrow (Figure 1B) and then from the dorsal side (Figure 1C). This was followed by the

257

removal of the aleurone in the flanks of the kernels. Aleurone layer located in the ventral

258

furrow was not removed even with the pearling degree of 39% (Figure 1E). In general, it was

259

noticed that pearling of hulled and hulless barley kernels followed similar pattern although

260

aleurone layer of hulless varieties was removed at slightly lower degree of pearling than that

261

of hulled varieties (data not shown).

262

Phytic acid in pearling fractions

263

Figure 2 shows the concentration of phytic acid in the obtained dust fractions in the three

264

sample sets. It was expected that the concentration of phytic acid should increase in the dust

265

fractions from the beginning of the pearling and then decrease after passing the aleurone

266

layer. In the first trial (SampleSet1) the concentration of phytic acid vs degree of pearling

267

showed a quite flat appearance, which indicated that the aleurone layer was not concentrated

268

in a few dust fractions. It became obvious that the pearling did not remove all the aleurone,

269

since a clear decrease with pearling time was not observed. It turned out that the pearling

270

system used was not very efficient. For the second trial (SampleSet2), a different pearling

271

system was used, and in this case, it was observed that the concentration of phytic acid did 12 ACS Paragon Plus Environment

Page 12 of 33

Page 13 of 33

Journal of Agricultural and Food Chemistry

272

indeed decrease with longer pearling times. This difference indicated that the second pearling

273

system was more efficient than the first one. The aleurone was distributed more or less across

274

all fractions, but highest concentrations were obtained for pearling yields in the 13-25%

275

range. This result corresponds with Blandino et al.7 who found that the 15-25% pearling

276

fractions of barley had the highest antioxidant activity, which most likely was due to the

277

antioxidants in the aleurone layer. It should also be mentioned that the measured

278

concentrations of phytic acid were higher for the variety 'Olve' than for 'Marigold'. In the third

279

trial (SampleSet3) on hull-less barley, there was an immediate high concentration of phytic

280

acid due to very thin layers of pericarp and testa. A gradual decrease comparable with

281

SampleSet2 then occurred. The level of phytic acid in the two varieties 'Pihl' and 'Pirona' were

282

quite similar.

283

Phenolic acids in pearling fractions

284

Concentrations of ferulic acid, caffeic acid, p-coumaric acid and sinapic acid in one pearling

285

series of 'Olve' and 'Marigold' from SampleSet2 are shown in Figure 3. The evolution of the

286

concentrations of caffeic and sinapic acid with pearling degree followed a pattern similar to

287

the concentration of phytic acid, although the maximum concentrations were reached at

288

slightly lower pearling degrees than phytic acid. Ferulic and p-coumaric acid followed

289

different patterns, which are related to their distribution along the different layers of the grain.

290

The concentration of ferulic acid continuously decreased at an approximately constant rate for

291

pearling degrees between 10-50%. This corresponds to Ndolo et al.21 who found that the

292

concentration of ferulic acid in the pericarp and aleurone in two barley varieties were high,

293

but much lower in the endosperm. Thus, the more endosperm entering the pearling fraction,

294

the concentration of ferulic acid will decrease. Even though ferulic acid is responsible for the

295

strong bluish fluorescence in the aleurone layer, the concentrations are also high in the

296

pericarp tissue, which does not exhibit this strong blue fluorescence. Regarding p-coumaric 13 ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

297

acid, its maximum concentration in dust is obtained at the beginning of the pearling process,

298

indicating that highest concentrations are found in the pericarp. These profiles suggest that the

299

concentrations of p-coumaric and ferulic acid are not the best markers for the aleurone layer.

300

This is in agreement with results obtained after isolation of barley grain tissues from both

301

hulled and hulless barley for chemical analyses of biomarkers.8

302

Spectroscopic measurements

303

Figure 4 shows typical NIR spectra from dust fractions ('Olve' from SampleSet2). For

304

unprocessed absorption spectra, the highest offset was obtained for first fraction, and

305

then the offset steadily decreased with increasing pearling time. This systematic

306

variation in offset is probably due to a corresponding variation in light scattering in the

307

samples. Colour of the samples did also vary systematically from dark to white with

308

pearling time, however, colour should not affect NIR spectra much in this spectral

309

region. For scatter corrected spectra the systematic offset disappeared, but clear

310

systematic changes with pearling time could be observed around 1100 nm and 1200 nm.

311

Figure 5 shows typical emission fluorescence spectra from dust fractions at the three assayed

312

excitation wavelengths. The emission spectra for excitation at 280 nm showed a main

313

maximum centred at 320 nm and a small shoulder around 438 nm. The emission spectra for

314

excitation at 360 nm was less intense and had a single maximum centred at 435 nm, slightly

315

broader than the main maximum of the former one. Lastly, the emission spectra for λex 470

316

nm, in some cases presents a maximum at 544 nm and in other cases it presents a quite flat

317

aspect with no maxima. For each sample set, the series of spectra registered at λex 280 nm,

318

showed a systematic increase from first to last fraction. For λex 360 nm, there was an increase

319

at start, and then a decrease; while for λex 470 nm there appeared to be a quite steady decrease

14 ACS Paragon Plus Environment

Page 14 of 33

Page 15 of 33

Journal of Agricultural and Food Chemistry

320

with pearling time. For regression analysis as well as MCR we used all three emission spectra

321

concatenated as shown in Figure 5.

322

Calibration results

323

Table 1 summarises calibration results for NIR spectroscopy for SampleSet 1 and

324

SampleSet2. It is clear that quite good regression models could be established for phytic acid

325

in the pearling fractions. The two sample sets behaved differently, better models were

326

obtained for SampleSet2. The two sample sets were produced by different pearling machines,

327

and this might be the reason for why different results were obtained, and also why the two

328

sample sets were not well modelled together. Attempts to remove the main spectral variation

329

due to differences in light scattering properties did not significantly improve the model where

330

the two datasets were combined. The best model was obtained for SampleSet2 after SNV

331

correction of the spectra. This model required one PLS factor more than the model based on

332

raw absorption spectra, and this suggests that SNV enables modelling of more subtle spectral

333

changes. An R2 of 0.96 and a prediction error (RMSECV) of 0.16% is acceptable for the kind

334

of process application we are aiming at. Based on the regression coefficients (not shown) it is

335

not easy to identify the main chemical sources of information for the model, however, several

336

are probably contributing. Phytic acid has a distinct signature in the NIR region.10 The amount

337

of bran, protein and beta-glucan will vary systematically in these fractions, and NIR is

338

sensitive to all of them.9,22

339

The results for fluorescence spectroscopy are quite similar to those of NIR (Table 1). Better

340

results for SampleSet2 than for SampleSet1 were also obtained now, and the combination of

341

the two data sets lead to slightly more complex models. The unit area normalisation did not

342

give notably better calibration models, and for the combined data set the performance was

343

worse than for raw fluorescence spectra. This suggests that the actual fluorescence intensity is

15 ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

344

important for the modelling. Common for NIR and fluorescence was that models for DataSet2

345

required more components than did DataSet1. This indicates that SampleSet2 contained some

346

spectral variation that was not present in SampleSet1. Since poorer models for phytic acid

347

were obtained for SampleSet1 for both NIR and fluorescence, there might have been some

348

mismatch between the spectroscopic data and the reference values. From Figure 2 is can be

349

seen that phytic acid in SampleSet1 varied up and down in an unexpected way as function of

350

pearling degree.

351

It is reasonable that fluorescence can be used to indirectly measure the amount of phytic acid

352

since it is a marker for the aleurone layer. As pointed out above, pericarp, aleurone and

353

endosperm have distinct and different fluorescent properties.13 Different phenolic acids in the

354

aleurone are fluorescent, for instance, ferulic acid, caffeic acid, sinapic acid and p-coumaric

355

acid (Figure 3).23 Phytic acid is, however, to our knowledge not fluorescent.

356

The chemical foundation for the calibrations (both NIR and fluorescence) is not completely

357

clear. It was therefore important to validate these models on independent samples. Table 2

358

summarises how the prediction models based on Sampleset2 performed on SampleSet3,

359

which were made of two other barley varieties. The models for Sampleset2 were used since

360

SampleSet2 and 3 were produced by the same pearling system. It can be seen that except for

361

the model based on raw fluorescence data, the correlations between the predicted and

362

measured values of phytic acid were high (0.95-0.97). For all the models there were unwanted

363

issues with offsets or biases compared to the reference values, but Figure 6 shows that the

364

predicted values were quite accurate and useful. All the predicted values were in the right

365

concentration range and they followed the same trend with pearling degree as did measured

366

phytic acid. On samples from 'Pihl', the NIR predicted values were very close to the reference

367

values, while for those from 'Pirona' the predicted values were systematically below. The

368

fluorescence model performed at about the same level as NIR. 16 ACS Paragon Plus Environment

Page 16 of 33

Page 17 of 33

Journal of Agricultural and Food Chemistry

369

There are differences between varieties of barley, and for the optical measurements described

370

here, this can probably result in systematic offsets as seen for the NIR predictions for 'Pirona'.

371

That means that for process purposes, it might be necessary to follow the progression of the

372

pearling dust during pearling rather than to rely on absolute values.

373

The model based on raw fluorescence spectra did not perform well on the test samples, while

374

the normalised spectra worked well. This indicates that the shape of the spectra are more

375

important than the actual intensities. Pre-processing of fluorescence spectra from front face

376

measurements of intact bio-samples for quantitative modelling is not a well explored topic,

377

and should be more thoroughly elucidated.

378

Multivariate curve resolution

379

MCR was attempted only on fluorescence data, since these spectra apparently consisted of

380

rather few spectral components. Normalised fluorescence spectra from SampleSet2 and

381

SampleSet3 were together subjected to MCR and three dominant spectral components were

382

found (Figure 7). Based on these components their relative concentrations could be estimated,

383

shown in Figure 8 for a set of pearling fractions for each of the varieties 'Olve', 'Marigold',

384

'Pihl' and 'Pirona'. One component shows a maximum at 335 nm for excitation 280 nm, which

385

do probably originate from tryptophan with possible contributions from tyrosine and

386

phenylalanine. The estimated concentrations for this component increases steadily with

387

pearling time and indicates the amount of endosperm in the fractions. The second component

388

has a maximum emission around 430 nm for excitation at 280 nm and at about 445 nm for

389

excitation at 350 nm. These are typical properties for phenolic acids. The concentration of this

390

component tends to follow a quite similar course to phytic acid and the aleurone layer (Figure

391

2). The third spectral component has a broad maximum around 550 nm for excitation 470 nm,

17 ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

392

and it decreased quite rapidly with pearling time. The identification of this component is not

393

clear, however it has previously been detected in pericarp.13

394

The estimated concentrations for the second component correlated well with the

395

concentrations of phytic acid, ferulic acid and caffeic acid (0.95, 0.95 and 0.98, respectively)

396

for the 'Pihl' and 'Pirona' samples. For SampleSet2, 'Olve' and 'Marigold', the corresponding

397

correlations were lower (0.64, 0.88, 0.73, respectively) mainly due to mismatch with the first

398

pearling fraction in every pearling series. If these were removed from the data set, the

399

correlations increased notably (0.90. 0.90 and 0.93, respectively). It is clear that the intensity

400

of fluorescence from the outer pericarp does not correlate well with the amount of phenolic

401

acids. Even though the fluorescent ferulic acid is present in the pericarp, the fluorescence is

402

very low. A reason might me that the fluorescence is quenched or reabsorbed by other

403

molecules in this tissue.

404

Overall, the results illustrate that it is possible to decompose rather complex fluorescence

405

spectra from intact pearling fractions into approximate concentrations of aleurone, endosperm

406

and pericarp. This approach is excellent for interpretation of the spectroscopic data, and might

407

also be used for continuous and semi-quantitative process monitoring.

408

The results from this work indicate that it is possible to estimate the amount of phytic acid,

409

and therefore the amount of aleurone, in pearling dust fractions from barley by both NIR and

410

fluorescence spectroscopy. The techniques are rapid and no sample pre-treatment is required,

411

which means that they can be used on-line in connection with industrial pearling systems.

412

NIR is well established in the cereal industry and suitable equipment is available for on-line

413

use on flour streams. A system for fluorescence spectroscopy would in principle be easy to

414

make, but might not presently be ready available off the shelf. In this work we used

415

concatenated emission spectra for three different excitations, but it might be sufficient to use

18 ACS Paragon Plus Environment

Page 18 of 33

Page 19 of 33

Journal of Agricultural and Food Chemistry

416

two or even just one excitation. Based on modelling the 280 nm excitation was the most

417

informative, then 350 nm, and 470 nm the least informative.

418

There are systematic differences between barley varieties. For example, 'Olve' seems to

419

contain more phytic acid than 'Marigold' (Figure 2b), and 'Pirona' might contain more than

420

'Pihl'. It means that an aleurone peak in dust fractions of 'Olve' will have a higher phytic acid

421

content than the corresponding fractions from 'Marigold'. The level of phytic acid in the

422

grains at hand must probably be monitored over time in the process before process

423

optimisation can be performed.

424

An on-line system can be used to adjust the pearling time for each batch to optimise the

425

amount and flour quality and to minimise the amount of pearling dust. One can also envision

426

specially designed pearling systems where it can be possible to follow the pearling process

427

and automatically extract fractions with high concentrations of aleurone and at the same time

428

control the process towards optimal utilisation of the raw material.

429

430

Acknowledgements

431

This work was partially funded by the EU-project BarleyBoost and the Norwegian

432

Agricultural Food Research Foundation through the projects Food Imaging (project

433

number 225347/F40) and Food for Health - Dietary fibre and glycaemic carbohydrates

434

(225352/F40). Hanne Zobel is thanked for skilled technical assistance.

435

436 437

19 ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

438

References

439

1.

Izydorczyk, M.S.; Symmons, S.J.; Dexter, J.E. Fractionation of wheat and barley. In

440

Whole-grain foods in health and desease. Marquart, L.M., Slavin, J.L., Fulcher, R.G.,

441

Eds.: American Association of Cereal Chemists, St. Paul, MN, 2002. 2. Jadhav, S.J.; Lutz, S.E.; Ghorpade, V.M.; Salunkhe, D.K. Barley: chemistry and

442

value-added processing. Crit. Rev. Food Sci. Nutr. 1998, 38, 123-171.

443 444

3.

Hemery, Y.; Rouau, X.; Lullien-Pellerin, V.; Barron, C.; Abecassis, J. Dry processes to

445

develop wheat fractions and products with enhanced nutritional quality. J. Cereal Sci.

446

2007, 46, 327-347.

447

4.

Hemery, Y.; Lullien-Pellerin, V.; Rouau, X.; Abecassis, J.; Samson, M.F.; Åman, P.; von

448

Reding, W.; Spoerndli, C.; Barron, C. Biochemical markers: Efficient tools for the

449

assessment of wheat grain tissue proportions in milling fractions. J. Cereal Sci. 2009, 49,

450

55-64.

451

5.

EFSA. 2011. Scientific Opinion on the substantiation of health claims related to beta-

452

glucans from oats and barley and maintenance of normal blood LDL-cholesterol

453

concentrations (ID 1236, 1299), increase in satiety leading to a reduction in energy intake

454

(ID 851, 852), reduction of post-prandial glycaemic responses (ID 821, 824), and

455

“digestive function” (ID 850) pursuant to Article 13(1) of Regulation (EC) No

456

1924/20061. EFSA J. 2011, 9, 2207.

457

6. Holtekjølen, A.K.; Sahlstrøm, S.; Knutsen, S.H. Phenolic contents and antioxidant

458

activities in covered whole-grain flours of Norwegian barley varieties and in fractions

459

obtained after pearling. Acta Agric. Scand., Sect. B. 2011, 61, 67-74.

460

7.

Blandino, M.; Locatelli, M.; Sovrani, V.; Coïsson J.D.; Rolle, L.; Travaglia, F.; Giacosa,

461

S.; Bordiga, M.; Scarpino, V.; Reyneri, A.; Arlorio, M. Progressive pearling of barley

462

kernel: chemical characterization of pearling fractions and effect of their inclusion on the 20 ACS Paragon Plus Environment

Page 20 of 33

Page 21 of 33

Journal of Agricultural and Food Chemistry

463

nutritional and technological properties of wheat bread. J. Agric. Food Chem. 2015, 63,

464

5875–5884.

465

8.

Barron, C.; Holopainen-Mantila, U; Sahlstrøm, S.; Holtekjølen, A.K.; Lullien-Pellerin, V.

466

Assessment of biochemical markers identified in wheat for monitoring barley grain

467

tissue. J. Cereal Sci. In press. Accepted 09.01.2017.

468

9.

Schmidt, J.; Gergely, S.; Schönlechner, R.; Grausgruber, H.; Tömösközi, S; Salgó, A.;

469

Berghofer E. Comparison of different types of NIR instruments in ability to measure β-

470

glucan content in naked barley. Cereal Chem. 2009, 86, 398–404.

471

10. Pande, R.; Mishra, H.N. Fourier transform near-infrared spectroscopy for rapid and

472

simple determination of phytic acid content in green gram seeds (Vigna radiata). Food

473

Chem. 2014, 172, 880–884.

474

11. Parrish, F.W.; Madacsi, J.P.; Phillippy, B.Q.; Wilfred, A.G.; Buco, S.M. Determination

475

of phytic acid in cottonseed by near-infrared reflectance spectroscopy. J. Agric. Food

476

Chem. 1990, 38, 407-409

477

12. Fulcher, R.G.; O’Brien, T.P.; Lee, J.W. Studies on the aleurone layer 1. Conventional and

478

fluorescence microscopy of the cell wall with emphasis on phenol-carbohydrate

479

complexes in wheat. Aust. J. Biol. Sci. 1972, 25, 23-34.

480

13. Jensen, S.Å.; Munck, L.; Martens, H. The botanical constituents of wheat and wheat

481

milling fractions. I. Quantification by autofluorescence. Cereal Chem. 1982, 59, 477-484.

482

14. Symons, S.J.; Dexter, J.E. Computer analysis of fluorescence for the measurement of

483

flour refinement as determined by flour ash content, flour grade color, and tristimulus

484

color measurements. Cereal Chem. 1991, 68, 454-460.

485

15. Hole, A.S.; Grimmer, S.; Naterstad, K.; Jensen, M.R.; Paur, I.; Johansen, S.G.; Balstad,

486

T.R.; Blomhoff, R.; Sahlstrom, S. Activation and inhibition of nuclear factor Kappa B

21 ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

487

activity by cereal extracts: Role of dietary phenolic acids. J. Agric. Food Chem. 2009, 57,

488

9481-9488.

489

16. Hole, A.S.; Rud, I.; Grimmer, S.; Sigl, S.; Narvhus, J.; Sahlstrom, S. Improved

490

bioavailability of dietary phenolic acids in whole grain barley and oat groat following

491

fermentation with probiotic Lactobacillus acidophilus, Lactobacillus johnsonii, and

492

Lactobacillus reuteri. J. Agric. Food Chem. 2012, 60, 6369-6375.

493 494

495 496 497 498 499

17. Barnes, R.J.; Dhanoa, M.S.; Lister, S.J. Standard normal variate transformation and detrending of near-infrared diffuse reflectance spectra. Appl. Spectrosc. 1989, 43, 772-777. 18. Martens, H.; Naes, T. Multivariate Calibration; Wiley: Chichester, U.K., 1989. 19. Tauler, R. Multivariate curve resolution applied to second order data. Chemom. Intell. Lab. Syst. 1995, 30, 133–146. 20. Tauler, R.; Smilde, A.; Kowalski, B. Selectivity, local rank, 3-way data analysis and ambiguity in multivariate curve resolution. J. Chemom. 1995, 9, 31-58.

500

21. Ndolo, V.U; Beta, T.; Fulcher, R.G. Ferulic acid fluorescence intensity profiles and

501

concentration measured by HPLC in pigmented and non-pigmented cereals. Food Res.

502

Int. 2013, 52, 109-118.

503

22. Kays, S.E.; Shimizu, N.; Barton, F.E.; Ohtsubo, K. Near-infrared transmission and

504

reflectance spectroscopy for the determination of dietary fiber in barley cultivars. Crop

505

Sci. 2005, 45, 2307-2311.

506

23. Lichtenthaler, H.K.; Schweiger, J. Cell wall bound ferulic acid, the major substance of

507

the blue-green fluorescence emission of plants. J. Plant Physiol. 1998, 152, 272-282.

508 509

510

22 ACS Paragon Plus Environment

Page 22 of 33

Page 23 of 33

Journal of Agricultural and Food Chemistry

511

Figure captions

512 513

Figure 1 Stereomicrographs (upper panel) and epifluorescence micrographs (lower panel) of

514

grains (cv. 'Olve') with different degree of pearling: (A) Native, unpearled grain; grains with

515

pearling degree of 14% (B), 27% (C), 32% (D) and 39% (E). In epifluorescence images,

516

aleurone layer appears bright blue surrounding starchy endosperm stained red. Scale bars

517

represent 5 mm and 500 µm in the lower and upper panel images, respectively.

518 519

Figure 2 Concentration of phytic acid in pearling fractions from (A) 'Olve' (red) and

520

'Marigold' (blue) (SampleSet1), (B) 'Olve' (red) and 'Marigold' (blue) (SampleSet2), and

521

(C) 'Pihl' (blue) and 'Pirona' (red) (SampleSet3).

522 523

Figure 3 Concentration of (A) Ferulic acid, (B) p-Coumaric acid, (C) caffeic acid, and

524

(D) Sinapic acid in pearling fractions from 'Olve' (red) and 'Marigold' (blue) from

525

SampleSet2.

526 527

Figure 4 NIR spectra from dust fractions after pearling. Absorption spectra (left), SNV

528

corrected spectra (right). Line colors indicate different pearling degrees.

529 530

Figure 5 Fluorescence spectra from dust fractions after pearling. Line colors indicate different

531

pearling degrees.

532 533

Figure 6 Predicted phytic acid in 32 pearling fractions of barley varieties 'Pihl' (upper panel)

534

and 'Pirona' (lower panel). Referenced phytic acid (black line), predicted by fluorescence

535

(blue line), predicted by NIR (red line).

536 537

Figure 7 Estimated spectral components from fluorescence spectra by MCR. Component 1

538

(green), component 2 (red), component 3 (blue).

539 540

Figure 8 Estimated concentrations of MCR components based on fluorescence spectra from

541

different barley varieties, A) 'Marigold', B) 'Olve', C) 'Pihl', D) 'Pirona'. Component 1 (green),

542

component 2 (red), component 3 (blue).

543

23 ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

Page 24 of 33

544

Tables

545

Table 1 Results for Regression Models Based on NIR, Cross-Validated Models. Pearling data set Spectral data # PLSR comp. R2 RMSECV SampleSet1 NIR Absorption 2 0.86 0.23 NIR SNV 4 0.85 0.24 SampleSet2 NIR Absorption 5 0.94 0.20 NIR SNV 6 0.96 0.16 SampleSet1 + NIR Absorption 7 0.87 0.28 SampleSet2 NIR SNV 8 0.89 0.26 SampleSet1 Fluorescence 2 0.85 0.23 Fluorescence norm 3 0.85 0.24 SampleSet2 Fluorescence 5 0.95 0.18 Fluorescence norm 4 0.96 0.18 Trial 1+3 Fluorescence 5 0.92 0.22 Fluorescence norm 5 0.87 0.28

546

547

Table 2 Results for Prediction of Test Set Pearling data set Spectral data NIR Absorption SNV Fluorescence Raw fluorescence Unit area norm

R2 0.95 0.97 0.49 0.97

548 549

550

551

552

24 ACS Paragon Plus Environment

RMSEP 1.47 0.35 1.14 0.36

Offset -1.04 0.17 2.20 0.65

Page 25 of 33

Journal of Agricultural and Food Chemistry

553

FIGURES

A

B

C

D

Figure 1

Figure 1

25 ACS Paragon Plus Environment

E

Journal of Agricultural and Food Chemistry

3.5

A

3

g/100g

2.5 2 1.5 1 0.5 0 5

15

25

35

45

55

15

25

35

45

55

25

35

45

55

3.5

B

3.0

g/100g

2.5 2.0 1.5 1.0 0.5 0.0 5 4.5

C

4 3.5

g/100g

3 2.5 2 1.5 1 0.5 0 5

15

Pearling degree (% w/w)

Figure 2

26 ACS Paragon Plus Environment

Page 26 of 33

Page 27 of 33

Journal of Agricultural and Food Chemistry

A

B

C

D

Figure 3

27 ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry

Page 28 of 33

B

A

Figure 4

28 ACS Paragon Plus Environment

Page 29 of 33

Journal of Agricultural and Food Chemistry

Figure 5

250000

F lu o res cen ce in ten s ity

600000

40000

λex= 280 nm

λex= 360 nm

0

0

300

400

500

10 % (w/w) 14 % (w/w) 18 % (w/w) 22 % (w/w) 26 % (w/w) 30 % (w/w) 33 % (w/w) 38 % (w/w)

λex= 470 nm

0

400

450

500

550

480

520

Emission wavelength (nm)

Figure 5

29 ACS Paragon Plus Environment

560

600

640

680

4

4

3.5

3.5

3

3

2.5

2.5

g/100g

g/100g

Journal of Agricultural and Food Chemistry

2

2

1.5

1.5

1

1

0.5

0.5

0

0 5.0

15.0

25.0

35.0

45.0

5.0

Pearling degree (%)

15.0

25.0

35.0

45.0

Pearling degree (%)

4.5

4.5

4

4

3.5

3.5

3

3

g/100g

g/100g

Page 30 of 33

2.5 2

2.5 2

1.5

1.5

1

1

0.5

0.5

0

0 5.0

15.0

25.0

35.0

45.0

5.0

Pearling degree (%)

15.0

25.0

35.0

Pearling degree (%)

Figure 6 554

30 ACS Paragon Plus Environment

45.0

Page 31 of 33

Journal of Agricultural and Food Chemistry

0.1

0.08

0.1

0.07 0.08

0.08 0.06 0.05

0.06

0.06

0.04 0.04

0.04

0.03 0.02

0.02

0.02 0.01

0 300

350

400

450

500

Wavelength (nm)

0 400

0 450

500

550

600

Wavelength (nm)

Figure 7

31 ACS Paragon Plus Environment

500

550

600

Wavelength (nm)

650

MCR estimated concentrations

Journal of Agricultural and Food Chemistry

1.8

B

1.6

1.4

1.4

1.2

1.2

1

1

0.8

0.8

0.6

0.6

0.4

0.4

0.2

0.2

0

0 5

MCR estimated concentrations

1.8

A

1.6

Page 32 of 33

1.8

15

25

35

45

55

5 1.8

C

1.6

1.4

1.2

1.2

1

1

0.8

0.8

0.6

0.6

0.4

0.4

0.2

0.2

0

25

35

45

55

D

1.6

1.4

15

0 5

15

25

35

45

5

Pearling degree (% w/w)

15

25

35

Pearling degree (% w/w)

Figure 8

32 ACS Paragon Plus Environment

45

Page 33 of 33

Journal of Agricultural and Food Chemistry

TOC Graphic

33 ACS Paragon Plus Environment