Detection of Peanut Adulteration in Food Samples by Nuclear Magnetic Resonance Spectroscopy

Article abstract of DOI:10.1021/acs.jafc.0c01999

The addition of cheap and also readily available raw materials, such as peanut powder, to visually and chemically similar matrices is a common problem in the food industry. When peanuts are used as an adulterant, there is an additional risk of potential health hazard to consumers as a result of allergy-induced anaphylaxis. In this study, different series of peanut admixtures to visually similar food products, such as powdered hazelnuts, almonds, and walnuts, were prepared and analyzed by ¹H nuclear magnetic resonance (NMR) spectroscopy. For identification, an isolated signal at 3.05 ppm in the ¹H NMR spectrum of polar peanut extract was used as an indicator of peanut adulteration. The chemical marker was identified as (2S,4R)-N-methyl-4-hydroxy-l-proline by resynthesis of the compound and used as an internal standard. The signal-to-noise ratio and the integral of the signal of the marker can both be used to detect peanut impurities. Overall, an approximate limit of detection of 4% admixtures of peanut in various food products was determined using a 400 MHz spectrometer. With regard to food fraud, we present a viable screening method for detection of economic-relevant peanut adulteration.

Full text of DOI:10.1021/acs.jafc.0c01999

Subscriber access provided by Uppsala universitetsbibliotek
Omics Technologies Applied to Agriculture and Food
Detection of Peanut Adulteration in Food Samples by NMR Spectroscopy
Caroline Schmitt, Tim Bastek, Alina Stelzer, Tobias Schneider, Markus Fischer, and Thomas Hackl
J. Agric. Food Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jafc.0c01999 • Publication Date (Web): 27 May 2020
Downloaded from pubs.acs.org on May 29, 2020
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 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 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.
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 38
Journal of Agricultural and Food Chemistry
Detection of Peanut Adulteration in Food Samples by
NMR Spectroscopy
Caroline Schmitt1, Tim Bastek1, Alina Stelzer1,2, Tobias Schneider1, Markus Fischer2 and Thomas
Hackl1,2*
1Institute of Organic Chemistry, University of Hamburg, Martin-Luther-King-Platz 6, 20146 Hamburg,
Germany
2HAMBURG SCHOOL OF FOOD SCIENCE - Institute of Food Chemistry, University of Hamburg,
Grindelallee 117, 20146 Hamburg, Germany
1
ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry
Page 2 of 38
1
2
ABSTRACT
The addition of cheap and also readily available raw materials, such as peanut powder, to visually and
chemically similar matrices is a common problem in the food industry. When peanuts are used as
adulterant, there is an additional risk of potential health hazard to consumers due to allergy-induced
anaphylaxis. In this study, different series of peanut admixtures to visually similar food products, such as
3
4
5
1
6
powdered hazelnuts, almonds and walnuts, were prepared and analyzed by H NMR spectroscopy. For
7
identification, an isolated signal at 3.05 ppm in the 1H NMR spectrum of polar peanut extract was used
as an indicator of peanut adulteration. The chemical marker was identified as (2S,4R)-N-methyl-4-
hydroxy-L-proline by resynthesis of the compound and use as internal standard. The signal-to-noise ratio
and the integral of the signal of the marker can both be used to detect peanut impurities. Overall, an
approximate limit of detection of 4% admixtures of peanut in various food products was determined using
a 400 MHz spectrometer. In regard to food fraud, we present a viable screening method for detection of
economic relevant peanut adulteration.
8
9
10
11
12
13
14
15
Keywords: adulteration, peanut, NMR spectroscopy, metabolomics
2
ACS Paragon Plus Environment

Page 3 of 38
Journal of Agricultural and Food Chemistry
16
17
INTRODUCTION
The adulteration of food products to achieve higher profits is a common problem in the food market.1
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
High price differences are an important factor and a potential incentive for food fraud and do not
exclusively concern adulterating with the same product, but can also lead to chemically and visually
similar admixtures in food.2 With the increasing global trade of food and the resulting uncertainties, it is
necessary to have analytical methods at hand in order to be able to detect adulterations as quickly and
easily as possible, but also validly.3 The most common analytical methods used in recent years include
e.g. chromatography and spectroscopy based techniques.4,5 For example, near-infrared reflectance (NIR)
spectroscopy was used to identify purple sweet potato, white sweet potato and their adulterated samples.5
Furthermore, gas chromatography combined with multivariate data analysis was used to obtain fatty acid
fingerprints of almond powder samples adulterated with apricot kernels.3 There are many studies towards
detection of adulterations of different oils, which are often affected by food fraud, whereby a relatively
expensive vegetable oil, such as olive oil, is adulterated with low-cost and low-quality oil (e.g. sunflower
oil).6–8
Nuclear magnetic resonance (NMR) spectroscopy as a non-destructive and relatively rapid method
could be an alternative to detect adulteration in food products.9 Furthermore, NMR spectroscopy shows
several advantages such as rich spectral information, the possibility of structure elucidation and fast
acquisition times.10 A relatively new but increasingly important field is the analysis of mixtures of
different foods using NMR spectroscopy. For example, 1H NMR spectroscopy was applied to identify the
presence of Arabica and Robusta species in coffee.10,11 Arabica is affected by food fraud, because Robusta
with its lower price is used as adulterant.12 Differences in chemical composition were examined leading
to the diterpene 16-O-methylcafestol (16-OMC) as a chemical marker in Robusta beans.10,11 Furthermore,
NMR spectroscopy can be used in combination with chemometric algorithms, as Parker et al. have shown
for the detection of olive oil adulteration with hazelnut oil using a 60 MHz spectrometer.13 An analytical
3
ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry
Page 4 of 38
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
protocol for the detection of adulterations of olive oil with hazelnut oil down to 10% using a 600 MHz
spectrometer is presented by Mannina et al.8 A new approach in assessment of mixtures by spectral
superposition was shown by Bachmann et al. for detection of admixtures of hazelnuts from different
geographical origin.14
For economic reasons it is important to identify food fraud with peanut and verify the quality of affected
products.15 The detection of adulteration of relatively expensive food products, such as almond (Prunus
dulcis MILL.), with peanut (Arachis hypogaea) is difficult due to similarity in chemical composition and
beyond the low price and high availability is an incentive for food fraud.15 Additionally, undeclared
admixtures can be life-threatening for peanut allergy sufferers even in smallest concentrations.16 Peanuts
are usually detected by their allergens using immunological techniques, such as enzyme-linked
immunosorbent assay (ELISA), DNA based techniques, such as polymerase chain reaction (PCR) and
MS based techniques.16–21 Food fraud also concerns hazelnuts (Corylus avellana) and walnuts (Juglans
regia), which also have a chemical composition similar to that of peanuts and which are used mainly in
confectionary and bakery industry as powder, making it even more difficult to distinguish between
them.15,22,23 This problem can be seen in the report of the international investigation operation OPSON VI
organized by Europol/INTERPOL which showed that one batch of 1.3 tonnes of roasted chopped
hazelnuts were adulterated with 8% peanuts.24 Using NMR spectroscopy with fast acquisition times has
advantages because of qualitative and quantitative detection of adulterant and is therefore a promising
alternative method to other already available approaches.10
In this study, a novel technique for detecting peanut adulteration was developed using
NMR spectroscopy. We analyzed three peanut blends of food products with economic relevance, that are
visually and chemically similar when powdered, such as almond, hazelnut and walnut. It was found that
the spectrum of the polar extract of peanut at 3.05 ppm shows a universal marker signal associated to
(2S,4R)-N-methyl-4-hydroxy-L-proline (1) that can be used to identify peanut additives in these nuts. In
4
ACS Paragon Plus Environment

Page 5 of 38
Journal of Agricultural and Food Chemistry
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
order to assess the applicability as a screening method for qualitative detection in regard to food fraud
linear regression models were developed using the calculated signal-to-noise ratio of the marker’s signal
and the known peanut content of each sample in the mixture series. By measuring twelve different peanut
samples the biological variance of the marker was estimated. To validate the models obtained and verify
the limitations due to natural biological variance, we prepared nine peanut mixtures with matrices from
different countries of origins and crop years and measured them by NMR spectroscopy.
MATERIALS AND METHODS
Reagents and chemicals
Deuterated solvents chloroform-d (99.8%) and methanol-d4 (99.8%) were purchased from EURISOTOP
(Saint-Aubin Cedex, France). Deuteriumoxide (99.9%), acetonitrile-d3 (99.5%) and 3-
(trimethylsilyl)propionic-2,2,3,3-d4 acid sodium salt (99.0%) were purchased from DEUTERO (Kastellaun,
Germany). Sodium azide (99.5%), potassium phosphate monobasic anhydrous (>99%), potassium
phosphate dibasic anhydrous (>98%), trans-4-hydroxy-L-proline (≥99%), N-methyl-cis-4-hydroxy-L-
proline (≥98%) and sodium borohydride were purchased from SIGMA ALDRICH (Merck KGaA, Darmstadt,
Germany). Formaldehyde (37%) was purchased from GRÜSSING GmbH (Filsum, Germany). Ultrapure
water for reversed-phase chromatography was purified by a SARTORIUS arium® pro apparatus (Sartopore
0.2 µm, UV) and acetonitrile (HPLC grade) was purchased from VWR.
Chemical Analysis and Chromatography
Mass spectra were recorded on an AGILENT 6224 ESI-TOF instrument. The optical rotation was
measured with a P8000 polarimeter from A. Krüss Optronic GmbH. For automated reversed-phase flash
column chromatography a Sepacore® flash system from BÜCHI in combination with Flash RS 40 C18ec
column from MACHEREY-NAGEL was used. Method A: 0–10 min. H2O (100%), flow rate 40 mL min-1,
5
ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry
Page 6 of 38
88
89
fraction size 15 mL; Method B: 0–10 min. CH3CN (0–50%), 10–12 (50–100%), 12–15 (100%), flow rate
40 mL min-1, fraction size 10 mL or 5 mL. For preparative HPLC an Infinity system from AGILENT
TECHNOLOGIES with a VP 250/10 Nucleodur C18 Pyramid column from MACHEREY-NAGEL was used.
Method C: 0–10 min. CH3CN (5%), 10–20 (5–95%), 20–25 (95%), flow rate 4 mL min-1, fraction size:
1.33 mL.
90
91
92
93
94
Sample Preparation
95
The peanut samples (roasted, in shell) were purchased in local grocery stores in Hamburg (Germany)
as detailed in the supplementary material table S1. The shell and skin were removed, 100 g of a sample
were shock-frozen with liquid nitrogen and ground with 150 g dry ice using a Grindomix GM 300 knife
mill equipped with a stainless-steel grinding container and a full metal knife (RETSCH, Haan, Germany).
The powdered sample was freeze-dried for 48 hours and stored at -20 °C. Hazelnut samples were provided
by producers and importers from a previous project.22 Walnut and almond samples were purchased from
different suppliers as detailed in the supplementary material table S1. Every nut sample was shock-frozen
with liquid nitrogen and then the shell was removed. After the shock freezing was repeated, 100 g of the
kernels were mixed with 150 g of dry ice and ground using a Grindomix GM 300 knife mill equipped
with a stainless-steel grinding container and a full metal knife. The powdered sample was freeze-dried for
48 hours and stored at -20 °C.
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
Extractions
Extraction A: The series of mixtures were prepared by weighing the lyophilizate of peanut and a
powdered sample of a food product (almond/hazelnut/walnut) with a total of 300 mg in a corresponding
mixing ratio (Table 1). Then, two steel balls (Ø = 2 mm), 400 μL methanol-d4, 500 μL chloroform-d and
600 μL deuterated phosphate buffer (50 mM) were added to the lyophilizate mixture and extracted in a
6
ACS Paragon Plus Environment

Page 7 of 38
Journal of Agricultural and Food Chemistry
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
ball mill for three minutes at 3.1 m/s. Each sample was centrifuged at 14.000 rcf and 4 °C for ten minutes.
Then, 100 μL of the supernatant was diluted with 600 μL phosphate buffer. Finally, 600 μL of the diluted
extract were transferred to a 5 mm NMR tube. The extraction procedure was carried out as a triplicate for
each sample.
Extraction B: For the methanol extract of the food products 300 mg lyophilizate was mixed with
1000 μL methanol-d4 and two steel balls (Ø = 2 mm) were added. The mixture was extracted in a ball mill
for three minutes at 3.1 m/s. It was centrifuged at 14.000 rcf and 4 °C for ten minutes. Finally, 600 μL of
the extract were transferred to a 5 mm NMR tube.
Extraction C: For the methanol/acetonitrile extract of the food products 300 mg lyophilizate was mixed
with 500 μL methanol-d4 and 500 μL acetonitrile-d3 and two steel balls (Ø = 2 mm) were added. The
mixture was extracted in a ball mill for three minutes at 3.1 m/s and it was centrifuged at 14.000 rcf (4 °C)
for ten minutes. Then, 350 μL of the supernatant was diluted with 350 μL methanol-d4. Finally, 600 μL
of the diluted extract were transferred to a 5 mm NMR tube.
Extraction D: For the acetonitrile extract of the food products 300 mg lyophilizate was mixed with
1000 μL acetonitrile-d3 and two steel balls (Ø = 2 mm) were added. The mixture was extracted in a ball
mill for three minutes at 3.1 m/s. It was centrifuged at 14.000 rcf and 4 °C for ten minutes. Finally, 600 μL
of the extract were transferred to a 5 mm NMR tube.
Isolation of marker metabolite 1: A suspension of 6 g peanut lyophilizate in 60 mL methanol was stirred
for one hour at room temperature. The solvent was removed under reduced pressure at 40 °C and the
process was repeated. Then 60 mL water, 50 mL chloroform and 40 mL methanol were added to the
7
ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry
Page 8 of 38
136
137
138
139
140
141
142
143
144
145
146
147
residue and the suspension was stirred for 90 minutes at room temperature. It was centrifuged at 14.000 rcf
for 30 minutes. The aqueous phase was separated, the methanol was removed under reduced pressure at
40 °C and the water by lyophilization. The residue was dissolved in 15 mL water and filtered through
Amicon® Ultra centrifugal filters (cut-off: 3 kDa) at 14.000 rcf for 15 minutes at room temperature. The
filtrate was lyophilized, and the residue was separated by automated reversed-phase flash column
1
chromatography using method A. The individual fractions were examined by H NMR spectroscopy for
the signal of the marker at 3.05 ppm and the relevant fractions were combined. The lyophilized residue
of the combined fractions was analyzed a second time by automated reversed-phase flash column
chromatography using method B. This procedure was repeated using method B with a smaller fraction
size of 5 mL. After collection of the fractions showing the signal of the marker, 112 mg of lyophilized
residue was separated by high-performance liquid chromatography using method C. The optical rotation
of the fraction obtained showing the marker metabolite was [α]20= -7.1° (c 0.49, H2O).
D
148
149
150
151
152
153
154
155
156
157
158
159
NMR acquisition
1H NMR spectra were acquired on a Bruker Avance III HD 400 MHz spectrometer (Bruker Biospin,
Rheinstetten, Germany) operating at 400.13 MHz. The noesygppr1d pulse sequence was used for
acquisition of water suppressed 1H NMR spectra applying the digitization mode baseopt. For each sample,
two spectra with different numbers of scans (NS = 8, NS = 128) were recorded at 300 K with
65536 complex data points and a spectral width of 8417.5 Hz. The RG was set to 64 and the transmitter
frequency offset to 1932.6 Hz.
1H NMR spectra from polar extract of peanut for identification of the chemical marker were acquired
on a Bruker Avance III HD 600 MHz spectrometer (Bruker Biospin, Rheinstetten, Germany) operating at
1
600.13 MHz. The zgesgp pulse sequence was used for acquisition of water suppressed H NMR spectra
of the fractionated polar peanut extract. The spectra were recorded at 298 K with NS = 128,
8
ACS Paragon Plus Environment

Page 9 of 38
Journal of Agricultural and Food Chemistry
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
65536 complex data points and a spectral width of 9615.4 Hz. The RG was set to 64 and the transmitter
frequency offset to 2821.8 Hz. For quantitative methods the relaxation delay D1 was set to 60 seconds
and the transmitter frequency offset to 2898.63 Hz.
NMR data processing and analysis
The FIDs were Fourier transformed with a line broadening factor of 0.3 Hz. The spectra obtained were
calibrated to the TMSP signal and processed by automatic phase and baseline correction with Topspin 3.5
(Bruker Biospins, Rheinstetten, Germany). For the series of mixtures, the signal-to-noise ratio (signal
region: 3.0394 ppm to 3.0575 ppm, noise region: -1.000 ppm to -2.000 ppm) was determined
automatically by using the implemented sino command. The obtained values were automatically extracted
from the spectra data by using a Python script (supplementary material S9). The integral of the singlet at
3.05 ppm was determined with the same signal-region automatically for all spectra using the auto-
integration modus of Topspin 3.5. The relative integrals were used to estimate of the biological variance
of the marker metabolite by comparing them. The linear regressions were performed using the software
Origin 2019 (OriginLab Corporation, Northampton, USA). The calculated mean of the admixture of
peanut was plotted versus the mean signal-to-noise ratio.
Synthesis of N-methyl-trans-4-hydroxy-L-proline (1)
The method of Sondengam et al. was used to synthesize the metabolite N-methyl-trans-4-hydroxy-L-
proline (1).25 First, 208 mg (1.59 mmol) trans-4-hydroxy-L-proline were suspended in 4 mL methanol
and 3 mL (0.03 mol) formaldehyde (37%) were added. The reaction mixture was refluxed for two hours.
After cooling the solution to 0 °C, 308 mg (8.14 mmol) sodium borohydride was added and the mixture
stirred for one hour at room temperature. The solvent was removed under reduced pressure and the residue
was dissolved in 2 mL methanol and filtered. After removal of solvent under reduced pressure, the crude
9
ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry
Page 10 of 38
184
185
186
187
188
189
190
191
192
193
product was separated by reversed-phase flush chromatography using method A. 15 mg were used for
further purification with preparative HPLC using method C. The product was obtained after lyophilization
of fraction no. 10 as a colorless solid. 1H-NMR (deuterated phosphate buffer (50 mM, pH = 7), 600 MHz,
300 K): δ [ppm] = 4.66 – 4.62 (m, 1H, H-4), 4.20 (dd, 3JH,H = 7.5, 11.1 Hz, 1H, H-2), 3.96 (dd, 1H, 2JH,H
= 13.0 Hz, 3JH,H = 4.8 Hz, 1H, H-5a/b), 3.20 (ddd, 2JH,H = 13.0 Hz, 4JH,H = 2.0, 2.0 Hz, 1H, H-5a/b), 3.05
(s, 3H, H-6), 2.49 (dddd, 2JH,H = 14.1 Hz, 3JH,H = 7.5 Hz, 4JH,H = 2.0, 2.0 Hz, 1H, H-3a/b), 2.26 (ddd, 2JH,H
13
= 14.1 Hz, 3JH,H = 5.0, 11.1 Hz, 1H, H-3a/b). C-NMR (deuterated phosphate buffer (50 mM, pH = 7),
151 MHz, 300 K): δ [ppm] = 172.9 (C-1), 70.0 (C-2), 69.4 (C-4), 62.6 (C-3), 43.1 (C-6), 38.2 (C-5).
HRMS (ESI+): For ([M+H]+) calculated for C6H11NO3: m/z = 146.0812, found: m/z = 146.0807.
[α]20= -17.2° (c 0.50, H2O).
D
194
195
196
197
198
199
200
201
202
203
204
205
206
207
RESULTS AND DISCUSSION
Peanuts are affected by food fraud because of their high availability and low price compared to other
nuts, which is a challenging problem, especially if there is a chemical and visual similarity with the
adulterated food. In order to detect peanut admixtures in powdered samples, an analytical marker, which
must be individual for peanut extracts, is required. For this, four extractions of peanut sample and
chemically similar raw materials (almond, hazelnut, walnut) with methanol-d4, acetonitrile-d3,
acetonitrile-d3/methanol-d4 (1:1) and deuterated phosphate buffer/methanol-d4/chloroform-d (2.5:2:3),
1
were performed and analyzed by H NMR spectroscopy. The spectra of peanut extracts were compared
with all other spectra (supplementary material, Figure S1 – S3). Of these four extractions, only the
1H NMR spectra obtained by the two polar solvents (A: deuterated phosphate buffer/methanol-d4,
B: methanol-d4) show an individual marker signal (Extraction A: singlet at 3.05 ppm, extraction B: singlet
at 3.02 ppm), which could indicate the admixture of peanut (Figure 1). The spectrum of peanut extract
obtained by extraction A additionally shows three regions (δ = 3.17 – 3.19; 2.45 – 2.53; 2.20 – 2.27) in
10
ACS Paragon Plus Environment

Page 11 of 38
Journal of Agricultural and Food Chemistry
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
which individual signals are observed, but with a lower intensity than the singlet at 3.05 ppm. Further, in
the NMR spectra of hazelnut and almond extract  (methanol-d4) signals with low intensity were detected
in the spectral region of the singlet at 3.02 ppm (supplementary material, Figure S1). Since these signals
can interfere and no such signals were observed in the spectra of extraction method A, the latter was
selected for further analysis. Next, a total of nine different samples of the adulterated raw materials
(hazelnut/almond/walnut; each from different countries of origin), were extracted and analyzed to verify
they did not show any signal in the singlet area. It should be noted that there is a doublet of low intensity
with a similar chemical shift of 3.04 ppm in the spectrum of aqueous walnut extract (Figure 1). In contrast,
the spectra of hazelnut and almond extracts showed no signals in this region.
The stability of the potential marker was monitored at 3.05 ppm for one week at room temperature
under observation of the singlet, with no changes in intensity (supplementary material, Figure S4). Figure
S5 in the SI shows a spectrum of a peanut sample after roasting it for 6 hours at 180 °C. The peanuts
turned mostly black under these conditions, but the metabolite, although partially degraded, was still
detected in considerable amount. In addition, twelve roasted peanut samples of different origins
(supplementary material, Figure S6) were purchased, measured and the presence of the chemical marker
in all samples was confirmed. Therefore, the chemical marker belongs to the signal pattern of peanut
extracts, regardless of the geographical origin. A biological variance of the marker signal was observed
and further quantified. Ten individual kernels of three peanut samples, each from different countries
(Egypt/South Africa/USA), were shock frozen and ground. Then, the powdered peanut kernels were
extracted (Extraction A) and measured by 1H NMR applying 128 scans. The relative integrals obtained
were compared and the standard deviation was calculated for different intensities of the signal. Kernels
from the USA exhibited the lowest standard deviation of 28%, a standard deviation of 37% was calculated
for kernels from South Africa. Based on the total of 30 measured spectra of individual peanut kernels, an
overall deviation of 46% was obtained for the marker singlet (supplementary material, Table S5). In
11
ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry
Page 12 of 38
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
contrast to the variability of single kernels, the biological variance of the twelve peanut samples, prepared
from 100 g kernels each, showed a standard deviation of 14% (Table 2). The lowest absolute
concentration of the metabolite was calculated for peanut sample 12 with 5 mg mL-1 related to TMSP and
the highest was found in sample 11 with 9 mg mL-1. Due to the biological variance, the use of the signal
of the marker for the accurate calculation of peanut admixtures is difficult. Nevertheless, it is suitable for
semi-quantitative and most important, for qualitative detection of peanut impurities. The first and most
important question in an incoming goods inspection will be whether or not the delivery of a product has
been adulterated by the addition of peanuts.
IDENTIFICATION OF THE PEANUT MARKER
The superimposition of individual signals makes the identification of metabolites in complex extraction
mixtures difficult, and 2D NMR experiments can also provide only limited information, since the
spectrum of the polar peanut extract is dominated by carbohydrate signals. The corresponding metabolite
with the singlet at 3.05 ppm in the NMR spectrum was identified after polar extraction and reversed-phase
1
column chromatography including fractionation to reduce the complexity of the spectrum. A H NMR
spectrum (600 MHz) was measured of each fraction and all fractions showing the singlet were collected
and fractionated again. This was repeated three times and the metabolite was finally purified by HPLC in
one fraction (no. 10) with the metabolite as the main component (Figure 5). With the spectral data the
metabolite was identified as N-methyl-4-hydroxy-L-proline. This metabolite was first isolated and
described from peanut flour by Lee et al. in 1965.26 However, it has not been characterized whether it is
the cis- or trans-4-hydroxy-L-proline derivative. The compound was confirmed by comparing the
chemical shifts of the signals with literature values (Haraguchi et al.27). The measured anticlockwise
rotation direction of [α]20= -7.1° (c 0.49, H2O) of the fraction no. 10 with the metabolite as a main
D
255
component indicates L-proline as the basic structure. In the ESI+-MS spectrum the molecular peak was
12
ACS Paragon Plus Environment

Page 13 of 38
Journal of Agricultural and Food Chemistry
256
257
258
259
260
261
262
observed for [M+H]+ at m/z 146.0817. Additionally, two fragment ions were observed at m/z 100.0763
and m/z 82.0659, which had been described by Sciuto et al. for sequential losses of carboxyl and water.28
As a reference, metabolite 1 was synthesized according to the method of Sondengam et al.25 and used as
an internal standard. An increase in intensity of the singlet and the associated signals in the fraction no. 10
was observed (Figure 5). Acquired N-methyl-cis-4-hydroxy-L-proline (2) was also used as an internal
standard to exclude this structure. The optical rotation of metabolite 1 was measured as
[α]20= -17.2° (c 0.50, H2O) and an anticlockwise rotation direction was also indicated. This was
D
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
compared with the rotation direction of the isolated product and literature and the absolute configuration
was confirmed.27,28 In this study it was shown that the metabolite in peanut is N-methyl-trans-4-hydroxy-
L-proline (1). This metabolite has also been isolated from the poisonous plant Ipomoea carnea27 and from
leaf of five species of the leguminous tropical tree Copaifera.29 In addition, metabolite 1 was found in red
algae (Chondria coerulescens)28 and higher plants of the species Croton gubouga.30 Metabolite 1 was
only described in pistachios as a nut-like food product.31
PEANUT ADULTERATION
Three series of mixtures (peanut/almond, peanut/hazelnut, peanut/walnut) were prepared by mixing the
lyophilizates in specific mixing ratios (Table 1). The extraction (A) was carried out in triplicates and two
spectra were acquired with different number of scans (NS = 8, NS = 128) for each individual sample.
Both spectra were analyzed and compared regarding the detection limit using the signal-to-noise ratio
(SNR) of the singlet. The signal region was set from 3.0394 ppm to 3.0575 ppm and the noise region from
-1.000 ppm to -2.000 ppm. We used a Python script to extract the calculated SNR automatically from a
series of spectra (supplementary material S9). The calculated mean values of the admixture of peanut
were plotted versus the mean SNR values obtained. The coefficients of determination (R2) in a range from
0.994 to 0.999 for both NS = 8 and NS = 128 show an adequate fit of the linear regressions. The
13
ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry
Page 14 of 38
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
coefficients of determination, the approximate limit of detection and the calibration models obained are
shown in table 3. Similar results of linear regression with comparable R2 were obtained for the series of
mixtures using the relative integral of the signal of the marker. For this purpose, the integrals in the spectra
were determined automatically with the same signal region set from 3.0394 ppm to 3.0575 ppm. For
hazelnut, both quantitation methods are compared graphically in figure 2. The graphical plots of almond
and walnut mixture series are shown in supplementary figures S7 – S8 as well as the lists with the SNR
values obtained with means and standard deviations for all series of mixtures in supplementary
tables S2 – S4.
The change in the signal intensity of the marker in spectra of hazelnut extract with increasing peanut
content is shown in comparison between NS = 8 and 128 as an example in figure 3. It is shown that an
increase of the numbers of scans shows a slight improvement in qualitative detection of impurities with
peanut. Using a lower number of scans resulted in very fast analysis times (~2 min), which allows a high
sample throughput. Due to the natural variability of the signal of the marker, already discussed in the
previous section, it is difficult to determine an exact value for the limit of detection (LOD). With regard
to the sample used for the hazelnut/peanut (peanut 2) and almond/peanut (peanut 1) calibration model,
the signal of the marker showed an SNR ≥ 3 in the spectra (NS = 8) with an admixture of 4% peanut.
However, this method cannot be applied e.g. in the walnut/peanut mixture series since a certain signal-to-
noise ratio is already present in the blank sample due to the doublet which is present in the walnut spectra
with a similar chemical shift of 3.04 ppm. Nevertheless, with the calibration models obtained semi-
quantitative predictions of peanut adulteration in powdered matrices can be made.
QUANTITATION OF PEANUT SAMPLES
For quantitation of unknown mixtures with the models obtained from linear regression, samples with
peanut content from 1.5% to 27% were prepared and extracted (method A) in triplicates. The samples
14
ACS Paragon Plus Environment

Page 15 of 38
Journal of Agricultural and Food Chemistry
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
used for these mixtures, both peanut and adulterated food, were from different countries and harvest years.
The focus of the evaluation was on the errors that occurred in the calculations due to the natural variance
of the marker. The true value of peanut content was calculated by using the mean value of the triplicate.
The linear calibration models (Table 3) of the three series of mixtures resulted from evaluation by SNR
were used for calculating the peanut content using mean of the obtained SNR values from triplicates. In
table 4 the results of quantitation with the linear regression model (almond) are shown. As an example,
the true peanut content of the last sample was 27% and the calculated impurity was 16% peanut by means
of NS = 8. The result shows the highest deviation of 40% and in this case, the peanut sample 1 (declared
origin: Egypt) was used for calibration and the sample (peanut 12) used in this mixture has a significantly
lower concentration of the marker metabolite (Figure 4). Peanut sample 12 was conspicuous in the set of
all 12 peanut samples because the concentration of the marker was the lowest, which is highlighted in red
in the box plot in Figure 4. This affects the calculation, as the calculated value is also significantly lower
than the true value. The calculated value depends on the peanut sample used as adulterant and the
similarity of the concentration of the chemical marker in relation to the sample used for calibration. This
is a difficult problem for accurate quantitation as the content of the marker metabolite in a peanut sample
used as adulterant is most likely unknown. If the concentration of the metabolite in the peanut sample
used in calibration is similar to the concentration of the added sample (e.g. peanut 10), then a small
deviation in calculation is obtained, resulting in an accurate calculation of the peanut content. This can be
seen in the example of using peanut 10 as adulterant, where the deviation is only 1%. On average, for nine
peanut adulterated almond samples a deviation of 15 ─ 16% was calculated for both number of scans
(NS = 8, 128). Similar results were obtained in quantitation of walnut and hazelnut samples
(supplementary table S6 and S7). When adulterating walnut with peanut, the calculated peanut content
deviates on average by 23 ─ 26% from the true value whereas in hazelnut samples the average deviation
from true peanut content was 20 ─ 29%. The deviations reflect the error resulting from the different
15
ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry
Page 16 of 38
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
content of the marker metabolite in the peanut samples. Nevertheless, the semi-quantitative calculation
provides a good method for estimating the peanut content. An important advantage is that variations of
adulterated food, e.g. geographical origin or varieties, do not affect on the calculation as long as no other
signals occur in the range of the peanut marker signal at 3.05 ppm. In addition, the signal of N-methyl-
trans-4-hydroxy-L-proline (1) provides a reliable qualitative indication of peanut in unknown food
samples.
If peanut mixed with foods that are visually and chemically similar, this poses a challenging problem for
quality control and therefore it is important to develop fast methods for detection of peanut adulteration.
The detection of peanut adulteration in various food products (almond, hazelnut, walnut) was performed
using NMR spectroscopy in combination with an individual marker signal (singlet at 3.05 ppm). The
selected signal of the marker is universally applicable for the detection of economic relevant adulteration
with peanuts in visually similar raw materials as presented in this study. By extraction from powdered
peanut in combination with reversed-phase chromatography and HPLC, the signal used as a marker was
assigned to the metabolite N-methyl-trans-4-hydroxy-L-proline (1). Identification was based on the
comparison with the synthesized product 1 and was confirmed by MS and NMR data. This metabolite has
high potential to be used in routine analysis with fast acquisition times for the identification of
adulterations with peanuts. A major advantage of using this marker is the belonging to the metabolic
profile of peanuts, as the metabolite was already discovered in samples from 1965 and is also present in
the samples used in this study that were acquired in 2019.26 With a spectral acquisition time of two minutes
applying 8 number of scans, an approximate limit of detection of 4% adulteration of peanuts was reached.
The LOD can easily be reduced by extending the acquisition time. However, it should be noted that the
method is not sensitive enough to detect peanut allergens, but economic relevant adulterations, such as
those shown in the operation OPSON VI (8%) can be detected.24 The semi-quantitative method is
influenced by the biological variance of the chemical marker, but is not dependent on the adulterated raw
16
ACS Paragon Plus Environment

Page 17 of 38
Journal of Agricultural and Food Chemistry
352
353
material. In the context of food fraud, NMR spectroscopy thus represents a viable screening method for
detection of peanut adulteration.
17
ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry
Page 18 of 38
TABLES
Table 1: Ratios of the three series of mixtures prepared from two peanut samples (peanut 1/peanut 2). The
geographical origins of the food products were labelled as: Egypt/Italy (peanut/almond), Egypt/France
(peanut/walnut) and USA/France (peanut/hazelnut).
food product
[%]
food product
[%]
peanut [%]
peanut [%]
100
90
80
70
60
50
40
30
20
10
0
8
6
92
94
10
20
30
40
50
60
70
80
90
4
96
2
98
1
99
0.8
0.6
0.4
0
99.2
99.4
99.6
100
18
ACS Paragon Plus Environment

Page 19 of 38
Journal of Agricultural and Food Chemistry
Table 2: Relative integral of the singlet (3.05 ppm) in the polar extract of twelve different peanut samples.
The biological variance of the chemical marker was observed calculating the standard deviation of 14%.
declared origin
peanut sample
relative integral
Egypt
USA
Israel
South Africa
USA
1
2
3
4
5
6
7
8
9
1.36
1.25
1.25
1.31
1.31
1.09
1.18
1.23
1.29
1.28
1.68
0.89
Egypt
Egypt
USA
USA
USA
10
11
12
Egypt
Egypt
1.26
14
mean
standard
deviation [%]
19
ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry
Page 20 of 38
Table 3: Approximate limit of detection of the series of mixtures and calibration model obtained. The
coefficients of determination of linear regression for both evaluations show similar results. With a higher
number of scans of 128 a lower LOD can be obtained due to higher signal-to-noise ratio.
peanut
mixture
R2
linear calibration model
(SNR)
NS
LOD
(SNR/integral)
8
128
8
0.994/0.998
0.997/0.999
0.996/0.998
0.995/0.998
0.997/0.998
0.997/0.998
>4%
>2%
>4%
>2%
>4%
>2%
y = 0.77721 + 1.16327 · x
y = 4.92591 + 4.69144 · x
y = 0.31802 + 1.16597 · x
y = 3.36174 + 4.66916 · x
y = 0.99714 + 0.98465 · x
y = 5.23451 + 3.93275 · x
walnut
almond
hazelnut
128
8
128
20
ACS Paragon Plus Environment

Page 21 of 38
Journal of Agricultural and Food Chemistry
Table 4: Results of the quantitation of almond adulteration with peanut samples from various countries.
Almond samples were of different geographical origin and harvest years. Due to the biological variance
of the chemical marker, the calculated values deviate on average by 15 ─ 16% from the true value.
almond
sample1
calculated
value [%]3
1.84
peanut
NS
SNR2
peanut [%]2
1.51
deviation [%]
sample1
8
128
8
128
8
128
8
128
8
128
8
128
8
128
8
128
2.46
9.92
3.51
15.94
4.83
21.41
8.56
33.01
11.59
45.50
15.35
63.10
20.38
81.76
38.02
154.73
19.12
79.28
21.57
7.05
8.83
10.32
24.50
24.56
0.23
9.91
7.14
0.02
0.94
1.70
1.32
1.14
40.48
40.85
40.38
39.89
1
9
2
7
8
3
4
5
6
4
5
1.40
2.74
2.69
3.87
3.87
7.07
6.35
9.67
3.00
5.12
6
7.05
8
9.02
7
9.02
12.89
12.79
17.21
16.79
32.34
32.42
16.13
16.26
13.02
17.98
23.02
27.05
9
10
11
12
8
128
1specifications of samples as detailed in the supplementary material table S1; 2mean of triplicates; 3calculation was performed
using the calibration model in table 3 (almond).
21
ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry
Page 22 of 38
FIGURES
Figure 1
22
ACS Paragon Plus Environment

Page 23 of 38
Journal of Agricultural and Food Chemistry
Figure 2
23
ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry
Page 24 of 38
Figure 3
24
ACS Paragon Plus Environment

Page 25 of 38
Journal of Agricultural and Food Chemistry
Figure 4
25
ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry
Page 26 of 38
Figure 5
26
ACS Paragon Plus Environment

Page 27 of 38
Journal of Agricultural and Food Chemistry
354
ASSOCIATED CONTENT
Supporting Information
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
The following supporting information is available free of charge at ACS website http://pubs.acs.org:
Table S1: Information for nut samples used in this study including supplier, harvest year, variety and
declared origin.
Table S2: List of peanut additives with mean and standard deviation used in the almond series of mixtures.
The signal-to-noise ratio (NS = 8, NS = 128) obtained with mean and standard deviation for the linear
regression is shown.
Table S3: List of peanut additives with mean and standard deviation used in the walnut series of mixtures.
The signal-to-noise ratio (NS = 8, NS = 128) obtained with mean and standard deviation for the linear
regression is shown.
Table S4: List of peanut additives with mean and standard deviation used in the hazelnut series of
mixtures. The signal-to-noise ratio (NS = 8, NS = 128) obtained with mean and standard deviation for the
linear regression is shown.
Table S5: Relative integrals of singlet at 3.05 ppm of individual peanut kernels with declared origin and
calculation of mean and standard deviation.
Table S6: Results of the quantitation of hazelnut adulteration with peanut samples from various countries.
Hazelnut samples were of different geographical origin and harvest years. Due to the biological variance
of the chemical marker, the calculated values deviate on average by 20 ─ 29% from the true value.
Table S7: Results of the quantitation of walnut adulteration with peanut samples from various countries.
Walnut samples were of different geographical origin and harvest years. Due to the biological variance
of the chemical marker, the calculated values deviate on average by 23 ─ 26% from the true value.
27
ACS Paragon Plus Environment

Journal of Agricultural and Food Chemistry
Page 28 of 38
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
Figure S1: 1H NMR spectra (400 MHz, NS = 128) of the methanol-d4 extract of food products. The area
of the potential marker signal around 3.05 ppm is shown expanded. The extraction method was not used
for mixture analysis, because of other signals appearing in the same region as the marker singlet.
Figure S2: 1H NMR spectra (400 MHz, NS = 128) of the acetonitrile-d3 extract of food products.
Figure S3: 1H NMR spectra (400 MHz, NS = 128) of the methanol-d4/acetonitrile-d3 (1:1) extract of food
products.
1
Figure S4: H NMR spectra (400 MHz, NS = 128) of the stability measurement of a peanut extract
measured on the day of extraction (day 0), on the following two days (day 1, day 2) and after one week
(day 7) are shown. The area of the marker signal around 3.05 ppm is shown expanded. The signal of the
marker is stable, because no changes in signal intensities were observed.
Figure S5: 1H NMR spectra (400 MHz, NS = 128) of ten kernels of purchased peanut sample 8 (roasted),
which were shock-frozen and mortared before polar extraction. The marker metabolite was still present
after roasting ten kernels of the same sample for six hours at 180 °C.
Figure S6: 1H NMR spectra (400 MHz, NS = 128) of twelve peanut extracts using extraction method A.
The presence of the chemical marker in peanut samples was confirmed as the singlet at 3.05 ppm was
detected in all measured spectra.
Figure S7: Graphical plots of the calculated peanut quantity in mixture samples with almond versus the
obtained signal-to-noise ratio of spectra measured with NS = 8 (A) and NS = 128 (B). The evaluation was
also performed via the relative integrated area of the marker singlet at 3.05 ppm. The linear regression of
the relative integrated area of the spectra obtained with NS = 8 (C) and NS = 128 (D) is shown.
Figure S8: Graphical plots of the calculated peanut quantity in mixture samples with walnut versus the
obtained signal-to-noise ratio of spectra measured with NS = 8 (A) and NS = 128 (B). The evaluation was
also performed via the relative integrated area of the marker singlet at 3.05 ppm. The linear regression of
the relative integrated area of the spectra obtained with NS = 8 (C) and NS = 128 (D) is shown.
28
ACS Paragon Plus Environment

Page 29 of 38
Journal of Agricultural and Food Chemistry
400
S9: Python script: The script is for extracting automatically signal-to-noise (sino) values obtained with
401
402
TopSpin 3.5 (Bruker cooperation) from the raw of NMR spectra into a text file.
403
404
AUTHOR INFORMATION
Corresponding Author
405
406
* Corresponding author: Dr. Thomas Hackl, Institute of Organic Chemistry, University of Hamburg, Martin-Luther-King-Platz 6, 20146
Hamburg, Germany Tel.: +49-40-428382804 E-Mail: Thomas.Hackl@chemie.uni-hamburg.de
407
408
Author Contributions
409
410
The manuscript was written through contributions of all authors. / All authors have given approval to the final version of the manuscript.
411
412
413
414
415
416
417
418
419
NOTE
The authors declare no competing financial interest.
FUNDING
The project is supported by funds of the Federal Ministry of Food and Agriculture (BMEL; “Food
Profiling”, funding code: 2816500914) by a decision of the German Bundestag (parliament). Project
support is provided by the Federal Institute for Agriculture and Food (BLE) within the scope of the
program for promoting innovation. The consortium is headed by the HAMBURG SCHOOL OF FOOD
SCIENCE. The focus lies on developments in the area of instrumental analysis for the authentication of
foodstuffs.
420
421
ACKNOWLEDGMENT
The authors are very grateful to René Bachmann and Maike Arndt providing us with authentic nut
422
423
samples. The authors thank Vera Priegnitz and Claudia Wontorra for their support in sample
measurement.
424
425
ABBREVIATIONS
AF, Afghanistan; AU, Australia; ELISA, enzyme-linked immunosorbent assay; EG, Egypt; ES, Spain;
426
ESI, electrospray ionization; FID, free induction decay; IT, Italy; LOD, limit of detection; NIR, near-
29
ACS Paragon Plus Environment