Subscriber access provided by Iowa State University | Library
New Analytical Methods
Determination of calystegines in several tomato varieties based on GC-Q-Orbitrap analysis and their classification by ANOVA Ana Romera-Torres, Francisco Javier Arrebola, José Luis Martínez Vidal, and Antonia Garrido Frenich J. Agric. Food Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jafc.8b06952 • Publication Date (Web): 13 Jan 2019 Downloaded from http://pubs.acs.org on January 17, 2019
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 31
Journal of Agricultural and Food Chemistry
1
Determination of Calystegines in Several Tomato Varieties Based on GC-Q-
2
Orbitrap Analysis and Their Classification by ANOVA
3 4 5
Ana Romera-Torres, Javier Arrebola-Liébanas, José Luis Martínez Vidal, Antonia
6
Garrido Frenich*
7 8
Department of Chemistry and Physics, Research Centre for Agricultural and Food
9
Biotechnology (BITAL), University of Almería, Agrifood Campus of International
10
Excellence, ceiA3, Carretera de Sacramento s/n, E-04120 Almería, Spain.
11 12 13
*Corresponding
14
[email protected])
author (Tel. +34 950015985; fax: +34 950015008. E-mail:
15
ACS Paragon Plus Environment
Journal of Agricultural and Food Chemistry
16
Abstract
17
In this study, several calystegines (A3, A5, B1, B2, B3, B4 and C1) have been determined
18
in tomato. A simple extraction followed by a derivatization step with silylating agents has
19
been performed prior to their analysis by gas chromatography coupled to high resolution
20
mass spectrometry (GC-HRMS-Q-Orbitrap), which allowed the monitoring of several
21
ions at accurate mass. The validation of the method has provided suitable values of
22
linearity, trueness (73.7-120.0%) and precision (≤ 20.0%, except for calystegines B3 and
23
B4 at 0.5 mg/kg). The limit of quantitation (LOQ) was set at 0.5 mg/kg for all analytes.
24
The validated method was successfully applied to the analysis of nine different tomato
25
varieties, and calystegines A3, A5, B2 and C1 were found at concentrations ranging
26
between 0.65 mg/kg (C1) and 12.47 mg/kg (B2). Tomato varieties were classified
27
according to their calystegines content applying an analysis of variance (ANOVA).
28 29
Keywords: Calystegines, Solanum lycopersicum, tomato varieties, GC-MS analysis, Q-
30
Orbitrap, ANOVA.
31
ACS Paragon Plus Environment
Page 2 of 31
Page 3 of 31
Journal of Agricultural and Food Chemistry
32
Introduction
33
Calystegines are a group of alkaloids discovered in 1988 during a determination of
34
opines by paper electrophoresis. They were initially discovered as positives spots in
35
transformed root cultures of Calystegia sepium and further screening of intact Atropa
36
belladonna roots also revealed their presence.1 Their chemical structure contains a
37
nortropane ring system with 3, 4 or 5 hydroxyl groups (calystegines A, B or C,
38
respectively) located at various positions and with differing stereochemistry, and also
39
having an aminoketal functionality.2 Other group of calystegines, named calystegines N,
40
are characterized by a bridgehead amino group and also contains a methylated nitrogen.3
41
As iminosugars, they usually act as competitive glycosidase inhibitors and are considered
42
for treatment of lysosomal storage diseases, e.g. Gaucher disease and Fabry disease.4
43
Although it has not been established that they are the main agents, calystegines are
44
presented in Solanum dimidiatum and Solanum kwebense, which causes crazy cow
45
syndrome and maldronksiekte in South Africa.5
46
Because of their closely related structure to tropane alkaloids, calystegines occurrence
47
were investigated in several genera, such as Atropa, Datura, Duboisia, Hyoscyamus, and
48
Scopolia belonging to Solanaceae family,6 and plant families such as Convolvulaceae,7
49
Erythroxylaceae8 and Brassicaceae,9 which are well-known for the presence of tropane
50
alkaloids. Solanaceae and Convolvulaceae families have many edible fruits and
51
vegetables rich in calystegines. Tomatoes (Solanum lycopersicum L.) are the most
52
popular vegetables in the world and around 150.000 million kg were produced in 2017
53
according to data from the United Nations Food and Agriculture Organization (FAO).10
54
Recently, analytical methods focused on the determination of calystegines have been
55
reviewed.11 These polyhydroxylated alkaloids have a high water solubility and are usually
56
extracted using different proportions of methanol/water, such as 50:50 (v/v),7,8,12–17 80:20,
ACS Paragon Plus Environment
Journal of Agricultural and Food Chemistry
57
v/v5,18 or 20:80, v/v.14 They have also been extracted with ethanol/water (50:50, v/v)19 or
58
with 0.2% of formic acid in acetonitrile/water (50:50, v/v).20 Then, extracts were passed
59
through cation and/or anion exchange resins columns in order to purify them and after
60
that, calystegines were usually silylated by a derivatization process. Finally, although
61
calystegines have been analyzed by thin layer chromatography21 or capillary
62
electrophoresis using an indirect UV detector,22 they are usually analyzed by gas
63
chromatography coupled to mass spectrometry with a single quadrupole analyzer (GC-
64
Q-MS).7,12,15,18,19 In contrast, neither purification nor derivatization are widely reported
65
for liquid chromatography-MS (LC-MS) analysis. In these cases, triple quadrupole
66
(QqQ),16,20 Orbitrap17 and a hybrid quadrupole coupled to Orbitrap (Q-Orbitrap)20
67
analyzer have been employed.
68
High resolution MS (HRMS) instruments, e.g. time of flight (TOF) or Orbitrap
69
instruments, measure ions at a high resolution power, and provide full-spectrum data with
70
a high mass accuracy. Moreover, the capacity to determine the molecular formula of
71
analytes from accurate-mass measurements has become the most important feature of
72
HRMS,23 allowing screening of targeted and untargeted compounds. While several
73
studies have employed HRMS to analyze similar compounds such as tropane
74
alkaloids,24,25 until now, only two LC-HRMS methods have been applied to the
75
determination of calystegines.17,20
76
In general, studies have been focused on the content of calystegines in several potato
77
varieties,5,13,14,26 but so far, little information about calystegines occurrence in tomato is
78
available17,27,28 and only four calystegines have been studied. Although one study
79
analyzed seven calystegines in tomato and tomato-based product using a 250 x 4.6 mm,
80
3 µm, ACE HILIC-A column (Advanced Chromatography Technologies LTD, Aberdeen,
81
Scotland),17 a thorough investigation of several varieties has not been carried out. The
ACS Paragon Plus Environment
Page 4 of 31
Page 5 of 31
Journal of Agricultural and Food Chemistry
82
aim of this work was to evaluate the variation of calystegines (A3, A5, B1, B2, B3, B4 and
83
C1) content within and between several tomato varieties obtained under the same
84
agronomic and environmental conditions. A GC coupled to a hybrid quadrupole-Orbitrap
85
analyzer (GC-Q-Orbitrap) was used for first time allowing a highly reliable identification
86
of the compounds.
87 88
Material and methods
89
Chemicals, reagents and equipment
90
Calystegine standards (A3, A5, B1, B2, B3, B4 and C1), structures from 1-7 shown in
91
Figure 1, were kindly provided by Professor Naoki Asano of Hokuriku University
92
(Kanazawa, Japan) and they were used as reference standards for GC analysis.
93
LC-MS grade methanol (purity ≥99.9%) was supplied by Merck-Sigma (St. Louis,
94
MO). LC-MS grade water was obtained from Scharlab (Barcelona, Spain), formic acid
95
(Optima LC-MS) was acquired from Fisher Scientific (Erembodegem, Belgium), and
96
HPLC grade n-hexane (purity 99.9%) was provided by VWR Chemicals (Radnor, PA).
97
Anhydrous pyridine (purity 99.8%), hexamethyldisilazane (HMDS) (reagent grade, ≥
98
99%) and chlorotrimethylsilane (TMCS) for GC derivatization (purity ≥ 99.0%) were
99
provided by Merck-Sigma (St. Louis, MO).
100
A Trace 1300 GC (Thermo Fisher Scientific, Bremen, Germany) was used for the
101
chromatographic analysis. It was equipped with a TriPlus RSH autosampler (Thermo
102
Fisher Scientific), and the column used was a 30 m x 0.25 mm i.d., 0.25 μm, VF-5ms
103
(Agilent, Santa Clara, CA). As MS detector, a Q-Exactive Orbitrap mass analyzer
104
(Thermo Fisher Scientific) was used.
105 106
Samples collection
ACS Paragon Plus Environment
Journal of Agricultural and Food Chemistry
107
Samples of nine tomato varieties were supplied by a greenhouse farmer from Almería
108
(southeast Spain). All they were grown under the same climatic and agronomic conditions
109
and into the same greenhouse. Three different samples of each var. were selected in order
110
to evaluate potential differences between varieties and within varieties. The tomato types
111
were: one oxheart, var. “Monterosa”; one black tomato, var. “Kumato”; two cherry, (one
112
common cherry, var. “Zoraida”, and another cherry oval, var. “Satyplum”); two saladette,
113
var. “Bernal” (A and B); and three cluster or tomato on the vine (TOV), varieties
114
“Ateneo”, “Guanche” and “Motto Fi”.
115 116
Sample extraction
117
Each sample (≥1 kg) was ground and homogenized. An aliquot of 1 g was weighed
118
and mixed with 10 mL of methanol/water (50:50, v/v). The mixture was shaken for 1 min
119
with a vortex. Afterwards, the sample was centrifuged for 10 min at 5000 rpm (4136 g).
120
The supernatant was filtered using a 0.22 μm Nylon filter vials (Agilent, Santa Clara,
121
CA), and 1 mL was collected for the derivatization step.
122 123
Derivatization
124
One mL of the extract was evaporated under a nitrogen stream, frozen and then,
125
lyophilized for 20 h. Next, 250 µL of pyridine, 40 µL of HMDS and 10 µL of TMCS
126
were added and the mixture was kept at 70 ºC for 30 min. After that, 700 µL of n-hexane
127
were added and homogenized increasing the total volume up to 1 mL, ready to be injected
128
into the GC-Q-Orbitrap system.
129 130
GC-Q-Orbitrap analysis
ACS Paragon Plus Environment
Page 6 of 31
Page 7 of 31
Journal of Agricultural and Food Chemistry
131
One µL of sample was injected at the injector at 250 ºC using splitless mode for 2 min.
132
Helium (with electronically controlled constant flow of 1 mL/min) was used as carrier
133
gas. For the separation of the calystegines, the GC oven was programmed as follows:
134
initially, the temperature was set at 100 ºC and held for 2 min, then, it was increased to
135
240 ºC at 5 ºC/min and finally, it was held at 240 ºC for 10 min, with a total running time
136
of 40 min.
137
The positive electron ionization (EI) source was operated at 70 eV at a temperature of
138
250 ºC, and the transfer line temperature was set at 250 ºC. The full scan-MS acquisition
139
mode was performed at a resolution power of 60000 full width at half maximum (FWHM)
140
in a scan range of m/z 60-600 with 1 µscan. Lock masses were used for column bleed at
141
m/z 73.04680 for C3H9Si+, m/z 133.01356 for C3H9O2Si2+, m/z
142
C5H15O3Si3+, m/z 281.05114 for C7H21O4Si4+ and m/z 355.06990 for C9H27O5Si5+.
207.03235 for
143
The chromatogram was processed using Xcalibur version 4.1, with Quanbrowser and
144
Qualbrowser, and the data was evaluated using TraceFinder 4.1 software (Thermo Fisher
145
Scientific, Les Ulis, France). Statistical analysis (analysis of variance, ANOVA) was
146
carried out with IBM SPSS Statistics v23 (Armonk, NY).
147 148
Method validation
149
A validation protocol was carried out in order to ensure an adequate identification and
150
quantitation of the target compounds. The performance characteristics of the method were
151
established evaluating linearity, trueness, precision and lower limits, following the
152
criteria established in SANTE/11813/2017.29 As many calystegines are natural
153
compounds which occur in several Solanaceae plants, there is not available a blank matrix
154
of tomato for all the target compounds. Therefore, and because the sample used for
155
method validation purposes was not a standard reference material, all the calystegines
ACS Paragon Plus Environment
Journal of Agricultural and Food Chemistry
156
that were not naturally present in this sample (B1, B3, B4 and C1) were evaluated for the
157
validation of the method.
158
Standard calibration solutions prepared in methanol at concentrations that ranged from
159
10 to 1000 µg/L were subjected to the sample pretreatment (evaporation, lyophilization
160
and derivatization) and analyzed. Linearity was studied from the least-squares regression
161
of peak area versus concentration and determination coefficients (R2) were calculated.
162
Intra-day precision, expressed as relative standard deviation (% RSD), was estimated
163
by analyzing three aliquots of tomato sample at three different spiked concentrations: 0.5,
164
1 and 5 mg/kg. Additionally, three aliquots of non-spiked samples were also extracted
165
and repeatability of the method for those calystegines (A3, A5 and B2) naturally present
166
in the sample was calculated. Trueness was determined through the recoveries of
167
calystegines B1, B3, B4 and C1 at three different spiking levels (0.5, 1 and 5 mg/kg).
168
Bearing in mind that no blank matrix was available, LOD and LOQ were established
169
from the solvent calibration curve. LOD was considered as the lowest concentration that
170
provided a mass spectrum with a characteristic ion measured with a mass error lower than
171
5 ppm. LOQ was established as the lowest concentration providing a mass spectrum with
172
a fragment ion having a mass error lower than 5 ppm, at the same retention time as the
173
characteristic ion and with the same chromatographic profile. Additionally, the LOQ
174
should present good linearity, precision and trueness.
175 176
Results and discussion
177
Optimization of GC-Q-Orbitrap conditions
178
A derivatized standard solution of each calystegine was injected into the GC-Q-
179
Orbitrap system. A full-MS was acquired for each calystegine in EI mode in order to
180
select the characteristic ions.
ACS Paragon Plus Environment
Page 8 of 31
Page 9 of 31
Journal of Agricultural and Food Chemistry
181
A predicted GC-MS spectrum was used as reference for each calystegine.30 Some
182
possible characteristic ions were identified and their masses were extracted from the total
183
ion chromatogram (TIC). Calystegines that belong to the same group (A, B or C) are
184
positional isomers, so they present the same molecular formula. It was observed that
185
calystegines within the same group provided the same ions, although slight differences
186
between relative abundances of these ions were observed.
187
Extracting the masses of the possible characteristic ions from the full-MS spectra, the
188
retention time, the theoretical mass and the chemical formula were determined from the
189
experimental mass. Several ions were confirmed for each calystegine as well as some
190
identified as common ions. The selected GC-Q-Orbitrap parameters are presented in
191
Table 1.
192
Figure 2 shows an example of the full-MS spectra including the proposed mechanism
193
for calystegines A3, B4 and C1. As observed, the loss of a methyl group (m/z 15.02293)
194
and trimethylsilanol (OH-TMS m/z 90.04954) are the most common fragmentations. it
195
should be mentioned that none of the molecular ions were detected, [M-CH3]·+ ion being
196
the highest intensity ion detected.
197
The described chromatographic method, based on a previously published work,15 was
198
used and proved that reliable identification and quantitation of the compounds was
199
possible due to goof chromatographic separation of the calystegine isomers studied.
200
Figure 3 shows the obtained TIC using the optimized experimental conditions.
201 202
Optimization of sample pretreatment
203
Sample pretreatment was divided into two phases: sample extraction and
204
derivatization. On the one hand, for the optimization of the sample extraction and
205
considering the highly polarity of the studied compounds, two different extractive options
ACS Paragon Plus Environment
Journal of Agricultural and Food Chemistry
206
were tested. Three aliquots of 10 g of tomato were spiked at 1 mg/kg and extracted using
207
methanol with 1% of formic acid (QuPPe-Method)31 and methanol/water (50:50, v/v),
208
which was frequently employed for their extraction.8,13,21 The QuPPe-Method had been
209
developed for the extraction of highly polar pesticides and because of the sample
210
pretreatment comprises an evaporation step, a non-aqueous solvent was thought to be
211
appropriate and tested by analyzing three aliquots of tomato spiked as described above.
212
However, recoveries ranging from 26-106% were obtained when 1% formic acid in
213
methanol was used (Figure 4A), while methanol/water (50:50, v/v) provided better
214
recoveries (64-118%) and higher sensitivity (Figure 4B). Bearing in mind that a marked
215
matrix effect was observed, the sample amount was reduced from 10 to 1 g. Three aliquots
216
of 1 g of sample spiked at 1 mg/kg were extracted. A diminution of the matrix effect was
217
observed and a clearer TIC were obtained with recoveries similar to those seen when 10
218
g of sample were tested (Figure 4C). Thus, for further experiments, this last extraction
219
procedure based on the use of only 1 g of sample was used.
220
The derivatization step was also submitted to optimization (Figure 4). Different
221
volumes of pyridine (30, 130 or 250 µL), HMDS (40 or 80 µL) and TMCS (10 or 20 µL)
222
were tested and the best results were obtained employing 250 µL pyridine, 40 µL HMDS
223
and 10 µL TMCS (Figure 4D). Although pyridine does not interfere with the reaction, the
224
entire solid residue obtained from the lyophilization step should be covered in order to be
225
completely dissolved and derivatized. Additionally, the use of a final clean-up step was
226
also evaluated, mixing the final extract with water (50:50, v/v), but improvements in the
227
results were not observed and this additional clean-up step was finally discarded.
228 229
Method validation
ACS Paragon Plus Environment
Page 10 of 31
Page 11 of 31
Journal of Agricultural and Food Chemistry
230
The performance characteristics of the optimized method are shown in Table 2.
231
Linearity was evaluated through the obtained determination coefficients (R2) which were
232
always higher than 0.99 for all the compounds, except calystegines A3 and B4 (Table 2).
233
The standard addition methodology was used for quantitation purposes due to a high
234
matrix effect observed (not reported). LOD was established at 0.25 (calystegine A5, B1,
235
B3 and C1) and 0.5 mg/kg (calystegine A3, B2 and B4), whereas LOQ was set at 0.5 mg/kg
236
for all calystegines.
237
As shown in Table 2, adequate intra-day precision (≤ 20.0%) was obtained at all the
238
concentrations studied and for all the compounds, but slightly higher values were
239
obtained for calystegine B3 at 0.5 mg/kg (22.4%) and calystegine B4 at 0.5 mg/kg (21.4%).
240
Recoveries ranged from 73.7 (calystegine B4) to 120.0% (calystegine B3) at 0.5 mg/kg;
241
from 89.5 (calystegine B4) to 118.8% (calystegine B1) at 1 mg/kg and from 95.8
242
(calystegine B3) to 100.7% (calystegine B4) at 5 mg/kg. For calculating recovery rates to
243
those calystegines natively present in the spiked samples, the concentrations found for
244
those compounds in the blank analysis were subtracted.
245
Very few reports determine calystegine in tomato, and only Romera-Torres et al.17
246
include validation parameters; the sensitivity is similar to that presented herein, although
247
slightly higher for some calystegines (LOQ 0.10 mg/kg for calystegine B4, 0.25 mg/kg
248
for calystegine B3 and 0.50 mg/kg for calystegines A3, A5, B1, B2 and C1); similar results
249
were observed for trueness and precision. However, comparing to other studies of
250
calystegines in other matrices, the proposed GC-Q-Orbitrap method is more sensitive and
251
also lower than alternative GC methodologies that provided LOQ from 3-10 mg/L or 2-
252
20 mg/kg,5,7–9,13–15,18,19and than LC methods, with LOQ from 0.4 to 2.5 mg/kg.16,20
253 254
Analysis of samples
ACS Paragon Plus Environment
Journal of Agricultural and Food Chemistry
255
It can be observed (Table 3) that ‘Zoraida’ and ‘Satyplum’ varieties presented the
256
highest concentrations of calystegines A3, A5, B2 and C1, while the concentration of the
257
analytes in ‘Monterosa’, ‘Bernal B’, ‘Ateneo’ and ‘Motto Fi’ varieties were below the
258
LOQ of the proposed analytical method. The concentrations calculated for the
259
calystegines found in the studied samples ranged from 0.65 mg/kg of C1 (‘Bernal A’) to
260
12.47 mg/kg of B2 (‘Zoraida’). Figure 5 shows the chromatograms and full-MS spectra
261
of calystegine A3 and calystegine B2 in two of the studied tomato samples.
262
Previous studies found concentrations in tomato samples of calystegine A3 (0.2-21
263
mg/kg), calystegine A5 (0.9-7 mg/kg) and calystegine B2 (0.4-21 mg/kg).27 These results
264
are in accordance to the results presented herein, although the wide variability in the
265
concentrations of natural compounds must be taken into account, which is highlighted
266
when observing the extensive range of concentrations found for instance in potato,
267
ranging from 0.2 mg/kg of calystegine A318 to 10145 mg/kg of calystegine B2.5
268
Finally, with the aim of checking whether the difference observed between the content
269
of calystegines depends on the var., an ANOVA study was performed. The results
270
obtained for calystegine A5 show that no significance difference was observed between
271
varieties (‘Kumato’, ‘Satyplum’, ‘Zoraida’, ‘Bernal A’ and ‘Guanche’) (p value>10)
272
(Table 3). However, significant differences were observed between the concentrations
273
found for calystegine C1 between varieties that in ‘Zoraida’ was significantly higher than
274
in ‘Kumato’, ‘Satyplum’, ‘Bernal A’ and ‘Guanche’ (p value