Ultra-performance liquid chromatography-mass spectrometry for precise fatty acid profiling of oilseed crops

Plant material

Seeds from 50 sunflower (Helianthus annuus) and 50 rapeseed (Brassica napus) lines (Table S1) from the Pustovoit All-Russia Research Institute of Oil Crops Collection (Russia, Krasnodar) were used in the study.

Reagents

The following reagents were used: Methanol LC-MS (Scharlau, Barcelona, Spain), Methyl tert-butyl ether HPLC grade (Scharlau, Barcelona, Spain), Chloroform HPLC grade (Fisher Scientific, Waltham, MA, USA), Heptane LC/MS grade (Honeywell Fluka, Morris Plains, NJ, USA), Water UHPLC-MS grade (Scharlau, Barcelona, Spain), Potassium hydroxide solution 45% in water (Sigma Aldrich, USA), NaCl USP grade (Helicon, Moscow, Russia), HCl 37% (PanReac AppliChem, Omaha, NE, USA), Acetonitrile LC/MS grade (Fisher Scientific, Waltham, MA, USA), Isopropanol LC-MS grade (Honeywell Fluka, USA), Ammonium acetate (Honeywell Fluka, Morris Plains, NJ, USA), Formic acid 98%–100% LC-MS grade (LiChropur Merck Millipore, Burlington, MA, USA), Acetic acid Optima, LC-MS grade (Fisher Scientific, Waltham, MA, USA), and FA standards (Oleic acid-¹³C₁₈ (Sigma Aldrich, St. Louis, MO, USA), Palmitic acid-¹³C₁₆ (Sigma Aldrich, St. Louis, MO, USA), and Stearic acid-¹³C₁₈ (Sigma Aldrich, St. Louis, MO, USA)).

GC-FID analysis

A total of 4–5 g of seeds were mixed together and homogenized and 0.5 g was taken for fatty acid extraction with 4 ml of hexane.

To obtain the methyl esters of fatty acids, 2–3 ml of homogenized seed-hexane mixture was transferred into the new tube and 0.1 ml of sodium methylate was added, and mixed intensively for two minutes. The tube content was further transferred onto the paper filter with Na₂SO₄ on the bottom. Obtained filtrate was then placed into the DAG-2M automatic dispenser tube.

GC-FID analysis was carried out using the < <Chromateck-Crystall 5000 > > GC chromatograph with the DAG-2M automatic dispenser. GC separation was performed in a SolGelWax column with the dimensions of 30 m × 0.25 mm × 0.5 µm; gas mobile phase - helium; speed - 25 cm/sec; temperature range - 185–230 °C.

UPLC-MS analysis

For lipid extraction, 10 mg (for each line) of sunflower (1 sample - 1 seed) and rapeseed (1 sample - several seeds) seeds were homogenized using six 2.8 mm zirconium oxide beads (Bertin Instruments, Montigny-le-Bretonneux, France) in the Precellys Evolution homogenizer (Bertin Instruments, Montigny-le-Bretonneux, France) coupled with Cryolis filled with dry ice at the temperature not exceeding 10 °C. Homogenization parameters were as follows: 6,800 rpm, 3* 20 s, pause 30 s. 400 µL of methanol/methyl tert-butyl ether mixture (1: 3 v: v) was added prior to homogenization. After homogenization, another 400 µl of methanol/methyl tert-butyl ether mixture was added, and the sample was mixed by vortexing. After sonication for 10 min in the ice-cooled ultrasonic bath and incubation at 4 °C for 30 min with shaking, the sample was transferred into the new 1.5 ml Eppendorf tube, and 560 ml of water/methanol mixture (3: 1 v: v) was added. After the addition of water/methanol the sample was mixed by vortexing for 10 min and centrifuged for 10 min at 4 °C at 12,700 rpm. This leads to the separation of two phases: the lipophilic phase and the polar phase. The upper, lipophilic, phase was collected and vacuum dried in Concentrator plus (Eppendorf) for 1.5 h at 30 °C and the pellet was stored at −80 °C before FA measurement. This protocol is based on the protocol described by Giavalisco et al. (2011) (Giavalisco et al., 2011). For FA quantification, three isotopically labeled internal FA standards were added to the extraction mixture (3 µg of each FA per sample).

For the analysis of FAs, the extracts obtained at the previous step were hydrolyzed using the protocol adopted from Bromke (Bromke et al., 2015). Lipid extracts were resuspended in 200 µl of the mixture of methanol and 6% KOH (4: 1 v: v). The tubes were incubated for 2 h at 60 °C with continuous shaking (1,800 rpm). After cooling to room temperature, 100 µl of saturated NaCl solution was added. The reaction mixture was acidified by the addition of 50 µl of 29% HCl. Tubes were thoroughly vortexed and spun for 30 s at full speed using a table centrifuge. FAs were extracted with 200 µl of chloroform/heptane mixture (1: 4 v: v). After mixing by vortexing and 15 s centrifugation, the organic phase was collected. Extraction with chloroform /heptane mixture was repeated the second time and the collected FA-containing organic phases were combined. The extract was washed by the addition of 200 µl of water followed by short vortexing and 15 s centrifugation at 12,700 rpm, which resulted in the separation of two phases. Finally, the organic upper phase was collected, dried under vacuum conditions in Concentrator plus (Eppendorf) (30 min, 30 °C) and stored at −80 °C before FA measurement.

To prepare samples for the injection into the UPLC-MS system, the dried extracts were first resuspended in acetonitrile/isopropanol mixture (70: 30), vortexed for 10 s, and incubated for 10 min at 4 °C with shaking, followed by sonication for 10 min on ice and centrifugation for 7 min at 4 °C at the speed 12,700 rpm. After the completion of these procedures, final dilutions with acetonitrile/isopropanol (70: 30) were prepared in MS vials.

Samples were processed using mass spectrometry (UPLC-MS) coupled with reversed phase ultra-performance liquid chromatography (ACQUITY UPLC System; Waters, USA) on Q-TOF (Quadrupole-Time-of-Flight) Maxis Impact II mass spectrometer (Bruker Daltonics, Wien, Austria). Parameters for analysis were set according to the negative ion mode with the spectra acquired over a mass range from m/z 50 to 1,200. The optimum values of the ESI-MS parameters were as follows: capillary voltage, + 4.0 kV; drying gas temperature, 200 °C; drying gas flow, 6.0 l/min; nebulizing gas pressure, 2 bar.

UPLC separation was carried out using the C8 Acquity Beh column (2.1 mm ×  100 mm, 1.7- µm particle size; Waters) and the Acquity BEH C8 1.7 µm Vanguard precolumn (Waters) at 60 °C. For UPLC gradient the mobile phases consisted of two solvents. Two different solvent systems and gradients were tested.

1. Solvent A: 1% of 1 M NH₄Ac solution and 0.1% formic acid in water; and solvent B: acetonitrile/isopropanol (7: 3, 1% of 1M NH₄Ac solution and 0.1% formic acid), with an injection volume of 3 µl. The following gradient profile was applied: 55% B, 1 min; linear gradient from 55% B to 80% B, 3 min; linear gradient from 80% B to 85% B, 8 min; linear gradient from 85% B to 100% B, 3 min. After washing the column for 4 min 50 s with 100% B, the proportion of buffer B in the mixture was set back to 55%, and the column was re-equilibrated for 4 min 10 s (24.5 min total run time), with the mobile phase flow rate of 400 µl/min.

2. Solvent A: 1% of 1M NH₄Ac solution and 0.1% acetic acid in water; and solvent B, acetonitrile/isopropanol (7: 3, 1% of 1M NH4Ac solution and 0.1% acetic acid), with an injection volume of 3 µl. The following gradient profile was applied: 55% B, 50 s; linear gradient from 55% B to 75% B, 1 min; linear gradient from 75% B to 89% B, 5 min; linear gradient from 89% B to 100% B, 1 min 10 s. After washing the column for 2 min with 100% B, the proportion of buffer B in the mixture was set back to 55%, and the column was re-equilibrated for 1 min 50 s (11.5 min total run time), with the mobile phase flow rate of 400 µl/min.

The sample final dilution was 5 (UPLC-5) and 400 times (UPLC-400) for the buffer systems 1 and 2, respectively.

Method validation procedure

Validation procedure included the evaluation of linearity, repeatability, reproducibility, and calculation of the limit of detection (LOD) and the limit of quantification (LOQ). Linearity was evaluated by building seven-point calibration curves (six replicates) in pure solvents. Solvent (acetonitrile/isopropanol, 70/30) was spiked with a standard mixture of isotopically labeled fatty acids (Oleic acid-13C18, Palmitic acid-13C16, and Stearic acid-13C18) in order to obtain all standards in the concentrations of 5, 10, 50, 100, 500, 1,000, and 2,500 ng/ml. The mean correlation coefficients (n = 6) of the calibration curves were 0.997 or higher for the target analytes, indicating good linearity in the selected concentration range (Tables 1A and 1B).

The limit of detection (lod) and the limit of quantification (LOQ) were determined from spiked samples, as the minimum detectable concentration of analyte with a signal-to-noise ratio of 3 and 10, respectively.

For repeatability evaluation, MS technical repeats were performed. The results are presented in the Table S2.

Data processing analysis

GC-FID: FAs detection was performed based on retention times using FA methyl ester standards (Fluke). The percentage of each FA was calculated based on the peak area using GC software.

Xcms package (Smith et al., 2006) was used for UPLC-MS data processing. Optimal parameters were chosen with the aid of the IPO package (Libiseller et al., 2015). Peak intensities were obtained using the peakTable function implemented in the xcms package with method=‘maxint’. As a result, set of peaks characterized by the retention time, mass divided by charge, and intensity for each sample was obtained. Isotopically labeled internal FA standards were used for retention time correction and alignment.

To identify peaks that correspond to FAs, all possible chemical formulas that correspond to FAs with the chain length varying between 10 and 28 carbon atoms (n) and with the number of double bonds varying from 0 to 6 (k) defined as C_nH_2n−2kO₂ were generated. Since the method does not distinguish between isomeric FAs, in what follows, the term ‘FA’ will be used to denote the groups of isomers (example of isomeric pattern presented on Fig. S1) defined by certain n and k, for example, FA18:1 denote all fatty acids with 18 carbons and single double bond. The masses of all these FAs in deprotonated state were calculated and all peaks matched masses within 20 ppm (part per million, for two masses m1 and m2 ppm = abs(m1 − m2)∕max(m1, m2)∗10⁶) were searched for.

Then peaks were manually selected based on the net-like patterns as described in [15]. Briefly, it was observed that the extension of the FA length increases retention time, while the addition of the double bond reduces retention time, thus FA peaks form net-like pattern.

For the sunflower LC-5 FA18:3 two peaks with the ten-second difference between their retention times were obtained. Since the intensities of both these peaks exhibited high Pearson correlation with the FA18:3 intensity in LC-400 data the sum of these peaks was used in the analysis. Intensities of all detected FAs were at least three orders of magnitude higher in plant samples than in blank samples. To obtain FA mass fractions, the intensities of all FAs were divided by the sum of intensities of FAs identified using all three methods (GC-FID, UPLC-5, and UPLC-400) and multiplied by 100.

Statistical analysis and data visualization was carried out using R (R Core Team, 2013).

GC-FID data

Fifty lines of either of the two oilseed crops were analyzed using GC-FID. As a result we have detected 11 and 13 FAs in sunflower and rapeseed, respectively (Figs. 1C and 2C, Table S3). The lists of FAs which were detected in the two crops differed by minor FAs, while C16:0, C16:1, C18:0, C18:1, C18:2, C18:3, C20:0, C20:1, C22:0, and C24:0, were common for both of them. FA C20:2, C22:1, and C24:1 were detected in rapeseed only, while C14:1 FAs we found in sunflower but were absent in rapeseed (Figs. 1C and 2C).

In rapeseed oil, the most abundant FAs were C16:0, C18:0, C18:1, C18:2, C18:3, and C20:1 which together made for 98.6% of all FAs. In sunflower seed oil, the most abundant FAs were C16:0, C18:0, C18:1, C18:2, C22:0, which constituted 98.9% of all FAs. In both species, two FAs (18:2 and 18:1) constitute more than 80% of the total FA content. However, in rapeseed, 18:1 is three times more abundant than 18:2, while in sunflower it is 18:2 which dominates (Figs. 1C and 2C).

The relative abundance of FAs was demonstrated to vary between samples (Table S3).

UPLC-MS data

For UPLC-MS fatty acids profiling we used two different buffer systems: the UPLC buffer with formic acid and the buffer with acetic acid. The former buffer is the most common buffer used in UPLC-MS experiments. In the latter buffer, ionization is known to be more effective since acetic acid is a weaker acid, so lower sample amount, and, therefore, higher sample dilution is required. For this reason for the system with acetic acid we used higher dilution (1:400), and for the system with formic acid we used lower sample dilution (1:500).

Hydrolyzed extracts of all lines of both sunflower and rapeseed were analyzed using both UPLC-MS-based methods.

At low dilution (UPLC-5), we are able to detect 29 FAs for sunflower and 35 FAs for rapeseed. At high dilution (UPLC-400), 25 and 31 FAs were detected for sunflower and rapeseed, respectively (Figs. 1C and 2C, Table S2). Hence, the use of lower dilution rates proved to be a more efficient approach which allowed to reveal the highest number of FAs. However, in this case it becomes impossible to perform the relative quantification of major FAs: 18:1 and 18:2, because they cause detector saturation (Figs. 3A, 3B). Apart from even-chain FAs, UPLC-MS also revealed odd-chain FAs (for example 17:0, 17:1, and 17:2). FAs with 18-carbon chain (stearic, oleic, and linoleic) were the most abundant. A net-like pattern on the M/z-RT plots can be observed (Figs. 4A, 4B and 5A, 5B).

FA content show a big variability between the lines (Table S2 , Fig. S2).

Results of both UPLC-MS methods show strong correlation: Spearman rho = 0.927 and 0,854 for sunflower and rapeseed, respectively (Figs. 4C, 5C).

Comparison of UPLC-MS with GC-FID

Results of UPLC-MS were compared with the results obtained using GC-FID. All 11 FAs detected in sunflower and 13 FAs detected in rapeseed were also identified using UPLC-MS. In the case of both plants, UPLC-MS proved to be more sensitive and detected about 2.5 times higher number of FAs than GC-FID (Figs. 1C, 2C). The majority of the minor FAs (mass fraction below 0.5%) are missed by GC-FID. The longest FA detected by GC-FID was the FA with 24-carbon chain, whereas UPLC-MS provided insight into the changes in longer FAs with the chains containing up to 28 carbon atoms.

As it can be seen in Figs. 1A, 1B and 2A, 2B, although the relative amounts of FAs measured by UPLC-MS and GC-FID are different, there exist significant correlation between them. Spearman rho = 0.908 and 0,918 for sunflower GC-FID and UPLC-5 and GC-FID and UPLC-400, respectively. Spearman rho = 0.979 and 0,947 for rapeseed GC-FID and UPLC-5 and GC-FID and UPLC-400, respectively.