Development, Optimization, and Comparison of Different Sample Pre-Treatments for Simultaneous Determination of Vitamin E and Vitamin K in Vegetables

The absence of vitamin E from the diet can lead to cardiovascular disease, cancer, cataracts, and premature aging. Vitamin K deficiency can lead to bleeding disorders. These fat-soluble vitamins are important nutritional factors that can be determined in different methods in vegetables. In this work, the simultaneous determination of α-tocopherol, α-tocopheryl acetate, phylloquinone, and menaquinone-4 by gas chromatography–mass spectrometry (GC–MS) has been optimized using both direct injection and solid phase microextraction (SPME). Three different sample pre-treatment approaches based on: (A) solid–liquid–liquid–liquid extraction (SLE–LLE), (B) SLE, and (C) SPME were then applied to extract the target analytes from vegetables samples using menaquinone as internal standard. All the procedures allowed the determination of the target analytes in onion, carrot, celery, and curly kale samples. Similar results were obtained with the three different approaches, even if the one based on SPME offers the best performance, together with a reduced use of solvent, time consumption, and experimental complexity, which makes it the preferable option for industrial applications.


Introduction
The determination of natural products of interest to human health (bioactive compounds) is carried out through a complex workflow that generally includes many steps such as extraction, fractionation, separation, and isolation of the compounds. However, the complex nature of natural extracts and low levels are often an obstacle to their correct determination.
The traditional strategies to isolate bioactive compounds from vegetables are based on liquid-liquid extraction (LLE), solid-liquid extraction (SLE), or solid phase extraction (SPE) using solvents of different coupled to GC-MS for the determination of α-tocopherol in olive oil as other important bioactive compound in different food matrix [16][17][18][19].
A fast and simple method for the determination of vitamins, including vitamin E acetate, in infant formula by DLLME combined with HPLC-UV has been also recently reported [20]. Very recently, an overview of methods for assessment of vitamin K has been provided by Górska [21]. Existing methods are usually laborious and there is still a great need for the development of robust, reliable methods for the quantitative determination of vitamin K. For instance, SPE-LC approaches were proposed for the determination of Vitamin K1 in fish feeds and shrimp post larvae [22] and edible microgreens [23] while DLLME coupled to liquid chromatography with diode-array and ESI (Electrospray ionization)-mass spectrometry detection (proved to be useful for routine control analysis of vitamins D and K in different food commodities [24]. Lately, two approaches based on liquid and solid phase extraction followed by ultra-high-performance liquid chromatography-atmospheric pressure chemical ionization tandem mass spectrometric (UHPLC-APCI-MS/MS) and liquid chromatography-with diode array-fluorescence detector (HPLC-DAD-FLD) were successfully developed for the determination of eight vitamin K compounds in 17 fermented food products at micrograms level [25] and of tocols, γ-oryzanols, phytosterols, squalene, cholecalciferol, and phylloquinone in rice bran and vegetable oil samples [26], respectively. A faster method for the determination of vitamin K in leaves was based on SPME-GC-FID using a polydimethylsiloxane (PDMS) fiber [27].
In the light of these considerations in the present work, the task of the simultaneous determination of α-tocopherol, α-tocopheryl acetate, phylloquinone and menaquinone by gas chromatography-mass spectrometry has been addressed. Three different sample pre-treatment approaches, of which the first two based on conventional extraction techniques and the third on SPME, were applied to extract the target analytes from vegetables samples (onion, carrot, celery, and curly kale) using menaquinone as the internal standard. All of the procedures permitted the determination of the target analytes in the selected food at satisfactory LODs, limits of detection, even if SPME proved to be the simpler and faster approach.

Vegetable Samples
All samples (onion, carrot, celery, and curly kale) were of local origin and collected from supermarkets. Each vegetable was finely grated and immediately subject to extraction procedures or dried for preservation. In this regard, the grated sample was portioned into aliquots (0.5 g) that were dried in a CHRIST-vacuum concentrator (Centrifuge Concentrator Centrifuge) (RVC 2-18, Osterode am Harz, Germany) at 30 • C until reduced to 43% ± 1% their weight. Dry residues were stored at 8 • C in the dark until analyzed. Spiked samples were prepared in triplicate by adding suitable amounts of standard solutions to fresh or dry aliquots.

Programmed Temperature Vaporization (PTV) and GC Parameter Optimization Using Direct Injection
The method was initially developed working in direct injection (1 µL) mode. Initially, the parameters of the PTV injector were optimized. The low starting temperature (70 • C) avoided the overflow of the liner with solvent vapors, the heating rate (14.5 • C/s) and the splitless time of 1 min optimized the transfer of the analytes into the column, the final temperature (300 • C) assured the cleaning of the inlet. The chromatographic conditions were then optimized to achieve good separation and chromatographic efficiency. The GC-MS chromatogram was obtained by directly injecting a standard of α-tocopherol, α-tocopheryl acetate, phylloquinone, and menaquinone-4 at a concentration level of 50 µg mL −1 using the optimized temperature gradient program and shows its resolution capability as reported in Figure 1. Calibration curves were linear in the explored concentration range with correlation coefficients greater than 0.999 and intercepts significantly close to zero at the 95% confidence level. The within-day (n = 5) and between-days (n = 5 over 5 days) coefficients of variation, estimated by an ANOVA test, ranged from 2.8 (α-tocopherol) to 4.0 (menaquinone) % and from 3.4 (α-tocopherol) to 5.9 (menaquinone) %, respectively, and remained practically unchanged at 5 and 50 µg mL −1 . The limit of detection (LOD) and limit of quantification (LOQ) were calculated as three and ten-fold the standard deviation of the intercept of the calibration curves. The estimated LODs were in the ranges 0.1 (α-tocopherol and α-tocopherol acetate)-0.6 µg mL −1 (menaquinone-4), respectively, while LOQs were in the range 0.4 (α-tocopherol)-1.9 µg mL −1 (menaquinone-4), respectively. All the validation parameters are resumed in Table 1. The within day and between days RSD% were always better than 4.0 and 5.9, respectively.
Molecules 2020, 25, x 4 of 12 concentration range with correlation coefficients greater than 0.999 and intercepts significantly close to zero at the 95% confidence level. The within-day (n = 5) and between-days (n = 5 over 5 days) coefficients of variation, estimated by an ANOVA test, ranged from 2.8 (α-tocopherol) to 4.0 (menaquinone) % and from 3.4 (α-tocopherol) to 5.9 (menaquinone) %, respectively, and remained practically unchanged at 5 and 50 µg mL −1 . The limit of detection (LOD) and limit of quantification (LOQ) were calculated as three and ten-fold the standard deviation of the intercept of the calibration curves. The estimated LODs were in the ranges 0.1 (α-tocopherol and α-tocopherol acetate)-0.6 µg mL −1 (menaquinone-4), respectively, while LOQs were in the range 0.4 (α-tocopherol)-1.9 µg mL −1 (menaquinone-4), respectively. All the validation parameters are resumed in Table 1. The within day and between days RSD% were always better than 4.0 and 5.9, respectively.

SPME Procedure Optimization
Experiments were then devoted to the optimization of the SPME conditions and to the validation of the method using SPME. A 7-µm thick PDMS-coated fiber was chosen for the simultaneous extraction of the selected vitamins from aqueous solutions since it has been successfully employed [17,26,27] for the extraction of α-tocopherol, vitamin K1 and KIt was also reported that the moderate addition of ethanol in water increases the solubility of the analytes [28] due to lower hydrophobic repulsion between ethanol and vitamin molecules, while the addition of HCl (10 mM) to the water-ethanol mixture [17,29] has a positive effect on the extraction of α-tocopherol with the PDMS fiber. Thus, the effect of the addition of ethanol and HCl, respectively, on the extraction efficiency were evaluated. Figure 2 reports how the extraction efficiency was affected by the addition of varying concentrations of ethanol and HCl (10 mM). While the addition of 10% ethanol has a positive effect on the extraction of all the analytes, a further increase in its concentration (20%) produced opposite results, i.e., response increase for the more hydrophobic

SPME Procedure Optimization
Experiments were then devoted to the optimization of the SPME conditions and to the validation of the method using SPME. A 7-µm thick PDMS-coated fiber was chosen for the simultaneous extraction of the selected vitamins from aqueous solutions since it has been successfully employed [17,26,27] for the extraction of α-tocopherol, vitamin K1 and KIt was also reported that the moderate addition of ethanol in water increases the solubility of the analytes [28] due to lower hydrophobic repulsion between ethanol and vitamin molecules, while the addition of HCl (10 mM) to the water-ethanol mixture [17,29] has a positive effect on the extraction of α-tocopherol with the PDMS fiber. Thus, the effect of the addition of ethanol and HCl, respectively, on the extraction efficiency were evaluated. increases the ionic strength of the solution and the organic compounds become less soluble increasing the partition coefficients between the phases. Four different concentrations of NaCl (0, 10%, 20%, and 30%) in water with 10% ethanol and 10 mM HCl were tested and the relevant results in terms of extraction efficiency are shown in Figure 2. It was apparent, in agreement with Aresta et al. [17,29] but in contrast with [27] that the extraction efficiency for all the analytes was negatively affected by salt addition and was not further investigated.
Molecules 2020, 25, x 5 of 12 vitamins E and response decrease for vitamins K. The addition of HCl shows rather positive results for all the investigated compounds. Consequently, a concentration of 10% ethanol with the presence of HCl (10 mM) was used for further experiments. Usually, salt addition increases the ionic strength of the solution and the organic compounds become less soluble increasing the partition coefficients between the phases. Four different concentrations of NaCl (0, 10%, 20%, and 30%) in water with 10% ethanol and 10 mM HCl were tested and the relevant results in terms of extraction efficiency are shown in Figure 2. It was apparent, in agreement with Aresta et al. [17] and [29] but in contrast with [27] that the extraction efficiency for all the analytes was negatively affected by salt addition and was not further investigated. As far as the extraction volume is concerned, 15 mL was selected as optimal value, since it was found that lower volumes produced a response decrease for all the analytes, while no effects were observed using higher volumes. Figure 3 shows the extraction time profiles obtained on Vitamins E and K at the temperature of 20 and 50 °C, respectively. The equilibrium conditions were almost always reached after 30 min (50 °C) and 40 min (20 °C) of extraction and the higher extraction efficiencies were obtained at 50°C for all the analytes except for α-tocopherol that showed a significant response decrease and, according to the literature [17,29] never reached equilibrium conditions. The extraction time of 30 min and the temperature of 50 °C were selected for the following experiments. The desorption temperature should be high enough to ensure the complete desorption of the extracted compounds from the fiber coating. In the present work, desorption was carried out for 5 min at 300 °C, since a significant "carry-over" was observed when desorbing at lower temperatures or shorter desorption times. The method was validated using the optimized SPME conditions and the relevant results are reported in Table 2. The dynamic range of the developed SPME-GC-MS procedure resulted in linear from the LOQ values for over two orders of magnitude, with correlation coefficients better than 0.997 and intercepts not significantly different from zero at 95% confidence level. The within day and between days RSD% were always better than 4.8 and 8.8, respectively. The estimated LOD and LOQ, calculated as three and ten-fold the standard deviation of the intercept of the calibration curves, were always better than 10 and 40 ng mL −1 , respectively. As far as the extraction volume is concerned, 15 mL was selected as optimal value, since it was found that lower volumes produced a response decrease for all the analytes, while no effects were observed using higher volumes. Figure 3 shows the extraction time profiles obtained on Vitamins E and K at the temperature of 20 and 50 • C, respectively. The equilibrium conditions were almost always reached after 30 min (50 • C) and 40 min (20 • C) of extraction and the higher extraction efficiencies were obtained at 50 • C for all the analytes except for α-tocopherol that showed a significant response decrease and, according to the literature [17,29] never reached equilibrium conditions. The extraction time of 30 min and the temperature of 50 • C were selected for the following experiments. The desorption temperature should be high enough to ensure the complete desorption of the extracted compounds from the fiber coating. In the present work, desorption was carried out for 5 min at 300 • C, since a significant "carry-over" was observed when desorbing at lower temperatures or shorter desorption times. The method was validated using the optimized SPME conditions and the relevant results are reported in Table 2. The dynamic range of the developed SPME-GC-MS procedure resulted in linear from the LOQ values for over two orders of magnitude, with correlation coefficients better than 0.997 and intercepts not significantly different from zero at 95% confidence level. The within day and between days RSD% were always better than 4.8 and 8.8, respectively. The estimated LOD and LOQ, calculated as three and ten-fold the standard deviation of the intercept of the calibration curves, were always better than 10 and 40 ng mL −1 , respectively. Table 2. Linear range, detection limits and precision of α-tocopherol, α-tocopherol acetate, phylloquinone, and menaquinone-4 obtained by SPME.

Discussion
The three sample pre-treatment approaches were applied on carrot, celery, curly kale, and onion for the simultaneous determination of α-tocopherol, α-tocopheryl acetate, and phylloquinone, using menaquinone-4 as the internal standard, since it is known to be absent in these vegetables. Figure 4 reports, as an example, the GC-MS extracted-ion chromatograms of a carrot dry sample pre-treated with the three procedures (A, B, and C) optimized in this work. The insets show the spectra acquired in the samples compared to those of the NIST (National Institute of Standards and Technology) library. As apparent, all the described extraction procedures permitted the easy determination of all the analytes and the internal standard. Quantitation was performed with the standard addition method. Recovery studies were performed at the concentration levels of 5, 50, and 100 µg mL −1 and the relevant results are shown in Table 3. SPME (C) was not considered in these experiments since it is notoriously a non-exhaustive technique. Recoveries were found to be concentration independent (p < 0.05) and procedure B provides the highest recoveries (>80%). The matrix effect for the SPME procedure [30] was calculated as described in the experimental section, and the values were 69 ± 15, 80 ± 12 and 96 ± 5 for each dilution ratio (1:1 or 0.1:1 or 0.005:1, w/v, respectively). As apparent, the matrix effect was minimized working at the dilution ratio of 0.005:1. The optimized conditions are reported in the experimental

Discussion
The three sample pre-treatment approaches were applied on carrot, celery, curly kale, and onion for the simultaneous determination of α-tocopherol, α-tocopheryl acetate, and phylloquinone, using menaquinone-4 as the internal standard, since it is known to be absent in these vegetables. Figure 4 reports, as an example, the GC-MS extracted-ion chromatograms of a carrot dry sample pre-treated with the three procedures (A, B, and C) optimized in this work. The insets show the spectra acquired in the samples compared to those of the NIST (National Institute of Standards and Technology) library. As apparent, all the described extraction procedures permitted the easy determination of all the analytes and the internal standard. Quantitation was performed with the standard addition method. Recovery studies were performed at the concentration levels of 5, 50, and 100 µg mL −1 and the relevant results are shown in Table 3. Table 3. Percentage recoveries obtained with the sample pretreatments A and B.

Chemicals
All chemicals used were supplied by Sigma-Aldrich (Milan, Italy) and were analytical grade. The standard of vitamins E (dl-α-tocopherol), E acetate (dl-α-tocopheryl acetate), K1 (phylloquinone) and K2 (menaquinone-4) have a purity level greater than 96%. Vitamin stock solutions (0.4 mg mL −1 ) were made in 50% (v/v) acetonitrile in ethanol and stored in aliquots (0.5 mL) at −20 • C in the dark. More dilute working solutions were obtained just before use in the range 0.05-150 µg mL-SPME working solutions were prepared in water with 10% ethanol, HCl (10 mM) just before use.

Sample Pre-Treatment
Vitamins were extracted from vegetables using three different approaches. All samples were always added with the internal standard before the sample pretreatment. In the first approach (A), each sample (1 g of fresh or 0.1 g of dry sample) was homogenized (Ultra Turbax IKA T18 Basic) (Merck Life Science S.r.l. 20149 Milano Italy) with 10 mL of water and 0.4 mL of 200 µg mL −1 menaquinone (internal standard) at the speed of 15.000 rpm for 1 min, after which, 15 mL of 6% (v/v) 2-propanol in hexane was added. The mixture was vortexed for 1 min and centrifuged (ALC Multispeed Refrigerated Centrifuge PK 121R) (Universal Resource Trading LTD, United Kingdom) at 3500 rpm for 5 min at 21 • C. The organic phase was then recovered and dried under a gentle stream of the residue was dissolved in 0.4 mL of pure hexane and 1 µL was injected in the GC-MS system.
For the second extraction procedure (B), each sample (1 g of fresh or 0.1 g of dry sample) was subjected to a solid-liquid extraction (30 min at room temperature in the dark) with 10 mL of 50% (v/v) acetonitrile in ethanol with 0.1% (w/v) butylated hydroxy anisole and 0.4 mL of internal standard under magnetic stirring. Then, the mixture was centrifuged for 5 min at 3500 rpm; the supernatant was transferred into an amber tube containing 0.25 g of Na 2 SO 4 anhydrous, briefly shaken and 1 µL injected.
The third procedure was based on SPME (C). Samples (0.075 g, fresh or dry) were placed in 15 mL amber vials, weighed and 15 mL of water with 10% (v/v) ethanol, HCl (10 mM) and 30 µL of internal standard were added. The solution was stirred for 15 min at 50 • C, and finally subjected to SPME for 30 min at 50 • C under magnetic stirring. Desorption of the analytes was performed into the GC injector kept at the constant temperature of 300 • C in splitless mode with 5 min of sampling time. The fiber was removed from the injector after 8 min, clean and ready for the next extraction. Quantifications were based on internal standard method. The matrix effect on the SPME was calculated [28] as follows. Different amounts (1.5 g, 0.15 g or 0.075 g) of vegetable samples were placed in 15 mL amber vials, spiked with the analytes at the concentration level of 100 µg g −1 and added with 15 mL of the extraction solution. Each sample was mixed for 15 min at 50 • C before being subjected to SPME-GC-MS analysis. Matrix effects were calculated by comparing the area responses obtained for each dilution ratio (1:1 or 0.1:1 or 0.005:1, w/v), with standard solutions at the corresponding concentrations as a reference

Apparatus
The gas chromatography-mass spectrometry system consisted of a Finnigan Trace GC Ultra gas chromatograph equipped with a split/splitless programmed temperature vaporization (PTV) injector coupled to an ion trap mass spectrometer (FinniganPolarisQ A TRACE™ TR-5 GC Column (Thermo Fisher Scientific Milan, Italy) (30 m, 0.25 µm i.d., 0.25 µm film thickness, Thermo Fisher Scientific, Milan, Italy) was used with helium as carrier gas (flow rate 1 mL min −1 ). The injector i.d. was 1 mm. The oven temperature program was 3 min at 180 • C, from 180 to 300 • C at 25 • C min −1 , and 13 min at 300 • C. When working with direct injection, the injector worked in PTV splitless mode, to avoid thermal degradation and/or adsorption. The relevant parameters are reported in Table 6. For SPME, the injector temperature and the GC transfer line were both set at 300 • C. The mass spectrometer was operated in the electron impact positive ion mode with a source temperature of 250 • C. The electron energy was 70 eV and the filament current 250 µA. Mass spectra were acquired in the m/z range 50-Detection of the vitamins were made from extracted ion chromatograms (m/z 165, for vitamin E and E acetate, and 186 and 225 for K1 and K2, respectively) obtained in total ion current (TIC) mode.

Conclusions
In conclusion, although the benefits of SPME (high productivity and low solvent consumption) are well known in academic research, the technology is not yet used as widely as it should by regulatory agencies, perhaps due to the perception that quantitative analysis is not solid. as much as traditional techniques. In this study, however, the equivalence of the non-exhaustive SPME with the exhaustive extractions of the conventional SLE and LLE techniques is demonstrated. Sample preparation is much simpler (0.075 g of sample in 15 mL of water with 10% ethanol) and faster (extraction time with the fiber of only 30 min, fully automatic) compared to the other two conventional procedures. Gas chromatography-mass spectrometry (GC-MS) was employed for first time for the analysis of α-tocopherol, α-tocopheryl acetate and phylloquinone in vegetables samples (onion, carrot, celery, and curly kale) using menaquinone-4 as the internal standard. Samples were extracted using different pre-treatment methods, i.e., (A) solid-liquid-liquid-liquid extraction (SLE-LLE), (B) SLE, and (C) SPME. Even if all the approaches allowed the determination of the analytes in the selected samples, SPME procedure provided the best performances, together with a reduced use of solvent, time consumption, and experimental complexity. In summary, in this manuscript a good proof-of-concept report of the combination of different methods with a high-throughput and "green" sample preparation approach is reported. This approach could be extended to more target compounds and more Vitamins (Vitamin A for instance) in future work.