Evaluation of a quick one-step sample preparation method for the determination of the isotopic fingerprint of rapeseeds (Brassica napus) - Investigation of the influence of the use of 2,2-Dimethoxypropane (DMP) on C- and H-CSIA by GC/C/Py-IRMS.

RATIONALE
Gas-chromatographic analyses for vegetable oils require transesterification, which generally involves multiple steps, mainly to generate fatty acid methyl esters (FAMEs). A quick method based on acid-catalyzed transesterification using 2,2-dimethoxypropane (DMP), enables the conversion in one-step, in a single reactor. For compound-specific stable carbon and hydrogen isotope analyses (C- and H-CSIA) of individual fatty acids (FAs) in oil, the verification of this one-step method has not yet been reported.


METHODS
In this study, we have evaluated the feasibility of the one-step method for C- and H-CSIA of individual FAMEs in rapeseed samples. The focus was on the investigation of the influence of methanol, which was produced from the reactions of DMP with glycerol and water during transesterification, on the accuracy of isotope composition of FAMEs, consequently of the FAs. The reproducibility of the one-step method was assessed by the measurement of the FAMEs from the rapeseed and rapeseed oil. For the C- and H-CSIA of individual FAMEs, a Gas Chromatography-Combustion/Pyrolysis-Isotope Ratio Mass Spectrometry (GC/C/Py-IRMS) system was used.


RESULTS
Our results showed that no significant differences arise in the carbon and hydrogen isotope compositions of the selected main FAMEs produced with and without DMP except for the H-CSIA value of C18:3. The reproducibility of the one-step method for the rapeseeds was in the range of ± 0.1 mUr to ± 0.3 mUr for C-CSIA and ± 1 mUr to ± 3 mUr for H-CSIA of the main FAMEs.


CONCLUSIONS
DMP improves the transesterification efficiency without influencing the accuracy of the C- and H-CSIA of FAMEs. The performance of the one-step method for rapeseed samples for the determination of C- and H-CSIA values of FAMEs is satisfactory. Thus, the applicability of the one-step method for isotopic fingerprint analyses of FAs in oilseeds is reported for the first time.


| INTRODUCTION
The combination of analytical techniques and chemometrics has been successfully applied to investigate agricultural commodities for the protection of high-quality origin varieties against adulteration and fraud. [1][2][3] For edible vegetable oils (e.g. olive oil, sesame oil, palm oil, pumpkin oil), chromatographic techniques with stable isotope analysis are often used to determine isotope compositions, i.e. "fingerprints," of individual oil components. [4][5][6][7][8][9] For instance, Spangenberg et al. showed that the ratio of the stable carbon isotope composition (δ 13 C values) of two fatty acids (FAs), namely oleic acid (C18:1) and palmitic acid (C16:0), could indicate the adulteration of olive oil with other vegetable oils. 8 6 Ehtesham et al. showed that the stable hydrogen isotope composition (δ 2 H values) and δ 13 C values of four FAs, i.e. butyric acid (C4:0), myristic acid (C14:0), C16:0 and C18:1, can be applied as biomarkers to trace the regional origin of milk powders. 10 To determine the isotope composition of FAs in vegetable oil, triacylglycerols (TAGs), the main components of oils, and/or free fatty acids (FFAs) are chemically converted to fatty acid methyl esters (FAMEs) in the presence of methanol and certain catalysts. [4][5][6][7][8][9] The stable isotope compositions of FAMEs are then determined using gas chromatography combustion/pyrolysis isotope ratio mass spectrometry (GC-C/Py-IRMS). 7,11,12 Due to the addition of the CH 3 group of methanol to FAs when producing FAMEs, the δ 13 C and δ 2 H values of FAMEs differ from the corresponding values of FAs in oil.
Thus, the isotope composition of a certain FA (e.g. C16:1) is calculated based on the isotope composition of its corresponding FAME (e.g. C16:1 FAME) and the methanol employed, using mass balance correction equations. [4][5][6][7][8][9]12 Potentially, kinetic isotope effects could cause additional differences in isotope compositions between reactant FAs and product FAMEs. 11 Whereas for carbon no kinetic isotope effects that change the isotope composition are expected in the reaction using excess methanol with catalyst, 13 hydrogen isotopes can undergo fractionation due to a secondary kinetic isotope effect. However, this effect was determined to be small when boron trifluoride in methanol was used to carry out the methylation reaction. 14 In addition, isotope fractionation of hydrogen may be caused by organic 1

H/ 2 H exchange in C-H bonds in FAs if oil is transesterified under acidic and hot
conditions, although isotope fractionation due to hydrogen exchange was found to be trivial during that reaction. 15 GC-C/Py-IRMS is a powerful technique for precisely determining slight differences in the abundances of isotopes (i.e. C, H, O and N) in various compounds. 16 For compound-specific stable carbon analysis (C-CSIA), individual target compounds are first separated in a GC column then transferred to an online-connected combustion chamber.
In the combustion chamber, each target compound is converted to CO 2 and the CO 2 gas is analyzed in the IRMS system where the abundances of the masses 44 ( 12 C 16 12 For compound-specific stable hydrogen isotope analysis (H-CSIA), target compounds separated in the GC column are transferred to a high-temperature conversion system and pyrolytically converted to H 2 gas. 16 The abundances of the masses 2 ( 1 H 2 ) and 3 ( 1 H 2 H) of the H 2 gas are determined in the IRMS system. The δ 13 C and δ 2 H values are determined from the carbon stable isotope ratio 13 C/ 12 C and hydrogen stable isotope ratio 2 H/ 1 H, respectively, of each compound in a sample and of corresponding reference standards.
To implement C-CSIA and H-CSIA for a vegetable oil, the oil should be first extracted from the oilseed. Depending on the form of samples, extraction of oil may require multiple steps. For instance, to extract oil from maize and 16 nonmaize seeds, nuts or kernels, Woodbury et al. first milled the samples and placed them in petroleum ether for a total of 6 h. The remaining solvent in the extracted oil was then removed under vacuum at 60 C. 9 Jeon et al.
collected sesame oil by centrifuging the roasted and pressed seeds for 10 min. The extracted sesame oil was then completely dried under nitrogen flushing. 17 The extracted oil is converted to FAMEs for GC analysis to determine the FA profile of the oil or for stable isotopic analysis of individual FAs. For authentication studies, FAMEs are normally produced using a boron trifluoride (BF 3 ) catalyst as described in DIN EN ISO 12966-2. 7,9,18 This DIN method also consists of multiple steps, i.e. saponification, transmethylation/methylation, purification and extraction of FAMEs prior to the instrumental analysis. 19 Briefly, TAGs are first transmethylated in the presence of methanolic sodium hydroxide to form FAMEs. All the FFAs present are converted to sodium soaps in an alkaline-catalyzed step. The soaps are then converted into methyl esters by reaction with a BF 3 -methanol complex. Isooctane and saturated sodium chloride solution are subsequently added to separate the FAMEs into the upper isooctane phase. After the removal of trace amount of water by the addition of anhydrous sodium sulfate, the FAMEs are ready for GC analysis.
The involvement of multiple steps and the use of BF 3 are known disadvantages of the DIN method when analyzing a large number of samples for chemometrics evaluation. Labor-intensive multistep sample preparation is not only unsuitable for processing a high number of samples but also may lead to certain sample loss and increase the chances of contamination. BF 3 is a toxic substance and the methanolic BF 3 reagent is highly reactive, and thus not suitable for long-term storage. 19,20 To overcome these disadvantages of the DIN method, Garcés and Mancha developed a unique and quick one-step sample preparation method. 21 This one-step method enables lipid extraction from seeds, transesterification, methylation and extraction of FAMEs simultaneously in a single vial within a couple of hours with the aid of 2,2-dimethoxypropane (DMP), toluene and n-heptane as components of a transmethylation mixture. DMP enables the transesterification and methylation reactions to be performed simultaneously. A transesterification product, glycerol, reacts with DMP to produce isopropylidine glycerol and methanol. 22 Removal of glycerol accelerates the transesterification reaction and the methanol produced via the reaction will be fed as a reactant for the transmethylation/methylation processes. In addition, DMP reacts with water generated from the methylation process to produce acetone and methanol. Water hinders the transmethylation reaction; thus, the removal of water from the system helps the methylation process to proceed. 23 The FAMEs produced in these two simultaneous reactions can then be directly extracted in the n-heptane available in the same reaction tube.
The applicability of the quick one-step method of Garcés and Mancha to determining the lipid content and the FA profile of various oilseeds was verified. However, its verification for the determination of the isotope composition of individual FAs has not yet been reported.
In the study reported here, we evaluated the feasibility of the one-step method of Garcés and Mancha for the C-and H-CSIA of rapeseed (Brassica napus) with the following two objectives:

| Chemicals and reagents
All the solvents and reagents were of analytical grade or higher purity.
For the production of transmethylation mixtures, methanol, toluene, DMP and sulfuric acid (H 2 SO 4 ) were used. Heptane was also added to each transmethylation mixture for the purpose of extracting the FAMEs produced in the reaction.
Supelco ® nonadecanoic acid methyl ester (C19:0 FAME; Merck, Darmstadt, Germany) was used as an internal standard to determine the recovery rate of FAMEs in the n-heptane phase and the FAME yield. Supelco ® 37 component FAME Mix and ROTICHROM ® FO 7 FAMEs mixture (Carl Roth, Karlsruhe, Germany), which consists of 11 FAMEs, were applied for the quantification of individual FAMEs.
ROTICHROM ® oleic acid methyl ester (C18:1 FAME) was applied as in-house reference material for quality control of the isotope measurements. The stable isotopic reference materials USGS 70 and USGS 71 were purchased from the USGS Reston Stable Isotope Laboratory (Reston, VA, USA). The icosanoic acid methyl esters (C20:0) in these reference materials were used for the normalization of stable carbon (δ 13 C) and hydrogen (δ 2 H) measurement of FAMEs.

| Samples
A bottle of rapeseed oil was purchased from a shop in Dottenfelderhof (Germany), and a rapeseed sample cultivated in Hesse (Germany) was provided by the Landesbetrieb Landwirtschaft Hesse. For each experiment, 20 mg of rapeseed oil or 50 mg of rapeseed was used for the one-step extraction. Each experiment was done in triplicate. For the rapeseed samples, 10 g of rapeseed was ground with a coffee grinder for 1 min to obtain a fine and homogeneous powder.

| Determination of FA profile
For the FA profile analysis, an Agilent 7890B gas chromatograph with FID (Agilent, Santa Clara, CA, USA) was employed. The GC instrument was equipped with a TG-WAX column (30 m × 0.32 mm, 0.5 μm; Thermo Fisher Scientific, Dreieich, Germany). Helium was used as the carrier gas at a flow rate of 1.5 mL/min. A PAL autosampler (CTC Analytics, Zwingen, Switzerland) was used to inject 1 μL of the sample into the GC inlet heated at 250 C with a split ratio of 1:10. The GC oven temperature program was started at 160 C for 1 min, heated to 190 C at 20 C/min and further to 220 C at 7 C/min where it was held for 29 min. Each sample was analyzed twice. The FAs in rapeseed were identified using GC-FID. For the determination of the FA profile, the samples and the ROTICHROM ® FO 7 FAME mixture were analyzed ing GC-FID. In accordance with DIN EN ISO 12966-4, the weight percentage of the individual FAME (wi) was calculated using the following equation: where Ai is the chromatographic area of the FAME. The correction factor Fi was determined by the ROTICHROM ® FO 7 FAME mixture using the equation: where mi/Σm is the known mass percentage of the FAME.

| δ 13 C and δ 2 H normalization
The stable isotope compositions of carbon and hydrogen are reported in delta notation (δ) as the per mille deviation of the isotope composition relative to reference materials: where R sample and R RM are the ratio of the heavy to light isotopes of sample and reference materials, respectively. The unit of δ is mUr (milliurey) instead of per mille based on the IUPAC guidelines for SI units. 25 The measured δ 13 where δ T spl , δ T RM1 and δ T RM2 denote the true isotope compositions of samples, USGS 70 and USGS 71, relative to international reference materials; and δ M spl , δ M std1 and δ M std2 represent the measured isotope compositions of the sample, USGS 70 and USGS 71, relative to the working gas.

| Quality assurance and quality control
To control the day-to-day performance of the GC-C/Py-IRMS system and the data normalization, the in-house reference material C18:1 FAME was used. The carbon and hydrogen isotope compositions of C18:1 FAME were measured after every 10-12 measurements and normalized using the USGS 70 and USGS 71 reference materials. The  Table S1 (supporting information).
The main FA components of rapeseed oil are C18:1 (61.8 ± 0.6%), C18:2 (18.1 ± 0.4%), C18:3 (9.8 ± 0.5%) and C16:0 (5.5 ± 0.1%). The FA composition remains unchanged regardless of the transmethylation mixtures, with or without DMP. Furthermore, the absolute mass of each of the four main FAs was determined by the addition of a known amount of an internal standard C19:0 FAME in each reactor. The results are presented in Figure 2 and Table S2 (supporting information). For all major FAMEs, the absolute mass per 100 g of rapeseed determined using TM+D was greater than that using TM-D and TX-D.
Based on the results presented in Table S2 ( catalyst, to achieve the maximum FAME yield from frying oil. 27 The yield of FAME was reduced by 27% when the molar ratio of methanol to oil was 106:1. 27 The TAGs are composed of FAs esterified on a glycerol backbone. 28 As presented in the FA profile in Table 2, the average molar mass of the TAGs in rapeseed oil is estimated to be 881 g/mol (see Table S3, supporting information). Thus, 50 mg of rapeseed, which contains 20 mg of rapeseed oil, is equivalent to 0.02265 mmol.
Based on this value, the molar ratios of methanol to oil in the transmethylation mixtures were calculated and are presented in Table 3. The methanol-to-oil molar ratios in TM+D, TM-D and TX-D are 2128:1, 2401:1 and 4256:1, respectively, indicating that the amount of methanol for the reaction is sufficient with or without DMP. Thus, the increase of the yield of TM+D is probably not due to the increase of methanol produced by the use of DMP.
We assume that the main effects of DMP in the transmethylation reaction are (1) consumption of a reaction product (glycerol) and (2) removal of water from the system. The addition of DMP to the system ensures complete conversion of lipid by consuming glycerol, a reaction product of the transmethylation reaction (see Figure 1). From T A B L E 2 FA profile of a rapeseed sample determined using three transmethylation mixtures  Furthermore, the ability of DMP to react with water promotes the transmethylation reaction (see Figure 1). The presence of water in the system may result in the oil hydrolysis reaction, which suppresses the transmethylation reaction. Commonly, rapeseed contains 7-10% water and 0.5-1.8% FFAs under optimal storage conditions. 28 The methylation reaction of FFAs also produces water (see Figure 1). In

| Precision of GC-C/Py-IRMS
Two in-house FAME reference materials, namely oleic acid methyl ester (C18:1) and nonadecanoic acid methyl ester (C19:0), were analyzed nine times (n = 9) using GC-C/Py-IRMS at the Institute of Applied Geosciences (IAG), Technical University Darmstadt, Germany and the mean and standard deviation were determined. The standard deviation (1σ) of the δ 13 C and δ 2 H values of the reference materials determined from the repeat measurements was considered as the precision of the C-and H-CSIA achieved by this GC-C/Py-IRMS system. The results are presented in Table 4. For both FAMEs, a precision of ±0.1 and ±1.7 mUr was achieved for C-CSIA and H-CSIA, respectively. These values showed the good stability of the instrument for the C-and H-CSIA of FAMEs.

| Influence of aggregate states of samples on reproducibility of H-and C-CSIA of FAMEs
One advantage of this one-step method is to save the step of extracting fluid oil from seeds, which may result in a certain degree of isotopic fractionation. 14 On the other hand, compared with fluid oil, the solid seeds are coarse and inhomogeneous, and this may influence the reproducibility in transmethylation and methylation reactions. An inconsistent and incomplete conversion of TAGs to FAs to FAMEs leads to a poor reproducibility, i.e. precision, of the isotope analysis.
The FAMEs were prepared from rapeseed oil and ground rapeseed under the same conditions using the TM+D transmethylation mixture.
Three reactors were prepared from three aliquots of each sample.
The ranges of the standard deviation of the carbon and hydrogen compositions of the four major FAMEs of oil and rapeseed samples are summarized in Table 5. Since the origin of the fluid oil and the rapeseed samples are different, the mean values differ (see Figure S1,  Regarding the C-CSIA measurements, the standard deviations of the We selected the TM+D and TX-D mixtures to produce FAMEs from the same batch of rapeseed sample. One of the reasons of this selection was that the reactions in both TM+D and TX-D reactors must have reached equilibrium, as proved by Garcés and Mancha under comparable conditions. 21 Another reason was that the FAME yields of TM+D (100%; see Table 3) and of TX-D (91%; see Table 3) are comparable. Panetta and Jahren showed that FAME yields influence the δ 13 C values, if the reactions had not reached equilibrium. 29 From their results, however, the difference in δ 13 C values between the FAME yield of 90% and 100% was less than 0.1 mUr. 29 Figure 3 shows the isotope compositions of the four major FAMEs produced using TM+D (with DMP) and TX-D (without DMP).
A T-test was carried out to compare the carbon and hydrogen isotope composition of individual FAMEs between these two reactors. As presented in Tables S4 and S5 (supporting information), statistically, there is no significant difference in isotope compositions between TX-D and TM+D, except for the δ 2 H value of C18:3. The difference in hydrogen isotope composition of C18:3 is 2.6 mUr, somewhat greater than the reproducibility of the seed sample preparation (see Table 5), but similar to that of the oil sample preparation.
The transmethylation reaction is a complicated process with regard to bond formation between the carboxyl carbon and the incoming methoxide oxygen. 14 In simpler terms, every mole of FAME generated is composed of 1 mol of -CH 3

PEER REVIEW
The peer review history for this article is available at https://publons. C/ 12 C isotope ratios of (E)-methyl cinnamate from different sources using isotope ratio mass spectrometry. J Agric Food Chem. 2004;52 (10):3065-3068.

SUPPORTING INFORMATION
Additional supporting information may be found online in the Supporting Information section at the end of this article.