Comparative analysis of fatty acids, tocopherols and phytosterols content in sunflower cultivars ( from a three-year multi-local study

. Sunflower oil is naturally rich in both unsaturated fatty acids and numerous minor components, with high levels of α -tocopherol and phytosterols that could play an important role in plant protection. These micronutriments, tocopherols and phytosterols, also exhibit health benefits for humans including reducing both cardiovascular diseases and aging disorders. α −Tocopherol represents more than 95% of total sunflower tocopherol and has the greatest vitaminic activity. Furthermore, this oil contains high amounts of β -sitosterol which is a promising health beneficial phytosterol. The objective of this study was to compare the content and composition of fatty acids, tocopherols and phytosterols in sunflower oil between different cultivars and locations in a three-year experiment, and to estimate the relationships between these minor components, fatty acids and climatic conditions. Results revealed that content and composition of these molecules are mainly determined by plant genotype. Nevertheless, the environmental conditions seemed to strongly influence the tocopherol content. The results suggest possible improvement in sunflower breeding for minor components.


INTRODUCTION
Sunflower world production represents up to 20% of the total world production of oil seeds (FAO, 2013). France was the 9 th world sunflower producer in 2003 and 2004 with 2 million hectares; this surface has increased up to 2.5 millions in 2011 (France was 14 th in the sunflower producers ranking) (FAO, 2013). Sunflower oil is rich in natural micronutrients, and various studies have shown the health benefit properties of polyunsaturated fatty acids (PUFA), tocopherols and phytosterols. The ingestion of PUFAs can have beneficial effects on human health such as the reduction of fatal coronary heart disease by decreasing the plasma low density lipoproteins (LDL)-cholesterol (Ruiz-Gutiérrez et al., 1997;Zock & Katan, 1998). Vitamin E isomers and specially α-tocopherol are able to reduce the risk of cardiovascular disease and exhibits anti-carcinogenic activities (Bramley et al., 2000;Beardsell et al., 2002). This can be achieved through prevention of polyunsaturated fatty acids oxidation and free radicals scavenging (Niki, 2004). Since it is retained by the liver by means of a specific transfer protein, α-tocopherol is considered to be the most biologically active form (Packer et al., 2001). Sterols inhibit cholesterol absorption in humans by limiting micellar solubility of cholesterol, leading to a reduction of cholesterol absorption (Kritchevsky, 1997;Thompson & Grundy, 2005;von Bergmann et al., 2005).
Since Seiler (1983), where the factors affecting the variability of fatty acids in sunflower were studied, other studies like Lagravère et al. (2004) have confirmed that genotypes and environmental growing conditions affect fatty acid composition. Various works on the variation of tocopherols and phytosterols have been implemented for different vegetal species studying the genotype and environmental effects; however, most of them have focused only in one minor component: Almonor et al. (1998) reported that γ-tocopherol content diminished in soybeans with reduced temperatures; Goffman and Becker (2002) found significant variations between genotypes and locations for the tocopherol content and composition in rapeseed; Määttä et al. (1999) described a significant difference for the phytosterol content in 7 oat cultivars or for 6 of rye (Zangenberg, 2004), Vlahakis and Hazebroek (2000) found differences in total phytosterol contents among canola, sunflower and soybean crops and also within each crop. Regarding sunflower, Demurin et al. (1996) and Velasco et al. (2002), have observed that genetic and different environmental conditions lead to variation of tocopherols content. However, in these studies only one year of cultivation was considered. Results concerning simultaneous variations in the content and composition of several families of oil components have been rarely addressed in the field. Fatty acids and tocopherols have also been studied in sunflower by Baydar and Erbas (2005), who found a positive correlation between total tocopherol and oleic acid contents. In soybean oil, Dolde et al. (1999) found a positive correlation between linoleic acid and tocopherol levels. A study including the three families (fatty acids, phytosterols and tocopherols) was also conducted by El-Shami et al. (2001) in rapeseed oil, but no correlations between these components were treated.
Regarding the composition of sunflower oil in health promoting molecules (Bramley et al., 2000;Vlahakis & Hazebroek, 2000;Holliday & Phillips, 2001;Velasco et al., 2002;Lagravère et al., 2004) and the importance of sunflower production in France, the objectives of our study were to (1) investigate the variability and (2) understand the relationship of these three families of components (PUFA, tocopherols and phytosterols) in sunflower oil. Our comparative study was carried out in four different commercial varieties grown in different locations in France during three consecutive years. At the same time, climatic parameters were determined in order to take into account their effects. To the best of our knowledge, this is the first time that a study comparing different varieties and over three years has been performed to study the variability in sunflower of tocopherol and phytosterol contents, in a multilocal study over a three-year period. In each location, four blocks of around 30 m 2 for each variety were sown between April the 1 st and 20 th . As recommended (Metropolitan Oleaginous Interprofessional Technical Center -CETIOM, France), the seedling rate was 65000 seeds/ha with 0.6 m between rows and 0.25 m between plants. Main plot treatments were: two tillages in Autumn and March-April; weeding using trifluralin, and at pre-seedling and flurochloridone at post-seedling, metaldehyde (7 kg/ha) to control slugs; fungicidal treatments were made when necessary with a mix of flusilazole and carbendazime (0.8 L/ ha); and two fertilizations applied: one before spring tillage (60 kg P/ha and 60 kg K/ha) and the other one during the seedling morphological developmental stage (50 kg N/ha with 1.2 kg B/ha). Thus, 97 samples (each one a bulk of the four blocks) were obtained for further analysis (Table 1) . These stations were selected due to their proximity to a greater number of experimental sites (fields), as well as for representing the 6 major climatic zones in France. This allowed us to consider a larger variability of weather conditions in France, all over the locations of the study during the three years. Four parameters were considered: rainfall (mm per day), maximun temperature (°C per day, T MAX ), minimun temperature (°C per day, T MIN ) and radiation (kW/m 2 ) for the period from April 1 st to September 30 th . Figure 2 sums-up the meteorological dataset for six locations that represented the six main regions where sunflower is cultivated in France. That allows us to estimate the climate differences in these areas, which could influence minor components content and composition.

MATERIALS AND METHODS
Sample Preparation. For each sample, 40 g of akenes were freeze-dried (HETO CT 110, Saint Herblain, France) and stored at −18 °C.
Oil extraction. Sunflower dry-frozen grains were ground with a sample mill (KnifeTec 1095; Foss Tecator AB, Sweden) during three periods of 10 s. Around 18 g were carefully weighed in a cellulose cartridge (Whatman ø = 30 mm, h = 100 mm) covered with backcombed cotton (Lohmann & Rauscher). Full cartridges were placed in a six soxhlet ramp for extraction (four hours) with hexane, following the reference method (AFNOR, 1998). Oil was recovered after a low pressure evaporation of the extraction solvent with a rotavapor (HS 40 Huber, Bioblock Scientific, Heildolph) and frozen (−20 °C) to minimize oxydative reactions before analysis.
Fatty acid analysis. Fifty mg of oil were weighed in a glass tube with a refrigerated system. As internal standard, 1 mL of arachidic acid was added and the extracts were saponified    Ayerdi Gotor A et al., FYTON 84 (2015) by adding 1 mL of 0.5 M methanolic NaOH, and heated to 65 °C for 30 min. Then, 2.5 mL of methanolic BF 3 (12% v/v) were added, and the mixture was heated for exactly 3 min at 65 °C, followed by a rapid refrigeration and the addition of 5 mL of ultrapure water and 10 mL of hexane (HPLC grade). After vigorous shaking and settle, the organic phase was recovered and analyzed immediately by gas chromatography.
Analyses were carried out with a Fisons 8000 series gas chromatograph (MMFC 800 Multi-function controller, EL 980 electrometer Unit, Italy) with an autosampler. Splitless injections were used with 0.5 mL sample volumes at an inlet temperature of 200 °C. Nitrogen carrier gas pressure was 200 kPa.
The fatty acids were separated using a Phenomenex fused silica capillary ZB-FFAP columnn (30 m x 0.25 mm x 0.25 μm, Paris, France). The oven temperature was programmed to (1) begin at 185 °C, (2) hold for 15 min; (3) increase to 200 °C at 5 °C/min and (4) remain constant for 7 min at 200 °C. The flame ionization detector (FID) was operated at 200 °C with hydrogen pressure at 100 kPa and air flow at 60 kPa. The major sunflower oil fatty acids were determined by comparison of retention times of the fatty acids methylated standards. The calibration results were based on the amount of arachidic acid detected in the sample. Table 2. Calculation parameters used to determine the real concentration of the solution with an UV-Spectrophotometer.

Tocopherol
Wave length (

Solution concentration (µg/ml) = Read absorbance value (mm) Division factor
Tocopherols analysis. Following the simplified international norm (ISO, 1997), a methanolic solution of each standard (0.1 mg/mL) was prepared. Solutions were measured with a UV lamp spectrophotometer (Hitachi U-1100 photometer) as it is shown in Table 2. The external calibration was carried out with dilutions of the stock solution with hexane and propan-2-ol (norm modification) to have a range of concentrations between 5 and 40 μg/mL of each form. For each sample, 1g of oil was carefully weighed (Sartorius Analytical balance Precisa 205 A, Italy), and adjusted up to 25 mL with hexane. The mixture was directly injected in a high performance liquid chromatography (HPLC) system (SpectraPhysics, Thermo Separation Products, USA). The complete separation of all native tocopherols was achieved with a normal phase LiChrorosorb Si60 column (250 cm x 4 mm x 5 μm, CIL Cluzeau, France). The mobile phase was a mixture of hexane/ isopropanol (99.7:0.3 v/v) at 1 mL/min flow rate. Detection was performed with a fluorescence detector (excitation wavelength = 298 nm and emission wavelength = 344 nm; Waters 2475 multi λ). The concentrations of the tocopherol homologs were calculated from the peak area.
Sterols extraction and CPG analysis. Sterols analysis procedure is an adaptation of the norm (AFNOR, 1999). The oil (250 mg) was dissolved in ethanol containing 1 mL of a freshly prepared solution of betulin (1mg/mL acetone) as internal standard. Saponification was achieved by adding 5 mL of 5 % (w/v) ethanolic KOH and heating at the boiling point, and kept up for 15 minutes in a refrigeration system. Saponification was stopped with 5 mL of ethanol. The mixture (5 mL) was placed in a glass column fill up with 10 g of aluminum oxide (Neuter, Akt. 1; 50-200 μm; pH 7.5; ICN Prolabo/Subra France) dampened with ethanol. After elution a new balloon was placed to recover the unsaponifiable matter washed with 5 mL of ethanol and 30 mL of diethyl ether. Unsaponifiable matter was concentrated in the balloon by evaporation under pressure of solvents and recovered with 2 mL of diethyl oxide. Then the mixture was evaporated to dryness with nitrogen (N 2 , Air liquide, France) . Before injection, TMS derivatives were prepared by dissolving the samples of the unsaponifiable fraction into 100-μL of 1-methyl imidazol and N-methyl-N (trimethylsilyl)-heptafluorobutyramide (5:95 v/v), and heating the mixture for 15 minutes to 105 °C in an oil bath. The saponification procedure was duplicated for each oil extract. Analyses were performed by a Fisons gas chromatograph (GC 8000 series MMFC 800 Multi-function controller, Italy) with an autosampler. Splitless injections were made with 1 μL sample volumes. Separations were achieved with a Phenomenex fused silica capillary ZB-5 (30 m x 0.25 mm x 0.25 μm, Paris, France). Nitrogen carrier gas pressure was 130 kPa. The injector was at 320 °C and the FID was 320 °C. The initial temperature of the column was 240 °C and increased to 320 °C at 4 °C/min, and held for 10 minutes until it decreased to initial conditions. Identification of sterols was based on the retention time of standards referred to above and with the relative times to betulin given in the norm. The calibration results were based on the integrated areas and adjusted according to the amount of betulin detected in the samples.
Method validation and quality control. Samples have been analysed by the ITERG reference laboratory (Paris, France). At the same time, one control sample was analyzed daily with each analytical series and for each kind of analysis (fatty acids, tocopherols, sterols) to determinate ageing changes in columns.
Statistical analysis. The data were statistically analyzed using Statistica V7 (StatSoft, Tulsa, USA). A three-way-ANOVA with three crossed factors (genotype, year) and one imbricated factor (location) with second order interaction was performed to determine the relationships between fatty acids, total tocopherol, total phytosterol contents, hybrids, planting location and year of study. It was also used to calculate bilateral correlation between variables (Pearson correlation coefficients).

RESULTS
Meteorological dataset. The six meteorological locations showed significant differences for the rainfalls; mean maximal and minimal temperatures (p<0.01) between the 3 years (Table 3). Significant differences in temperature were also observed between locations. In 2003, the plants were exposed to extreme growth and maturation conditions (highest temperatures and lowest rainfall) compared with 2002 (coldest and high precipitations) and 2004 (middle conditions).
Oil and fatty acid content. The sunflower seeds yielded around 50% oil by soxhlet extraction (Table 4). However in Table 3. Mean values for each year for the 6 locations except for solar radiation that is analyzed only in Carcassonne.

Year
Precipitation ( 2003, we observed significant differences between oil yields: Melody in the 16 th location: 0.38 g/g of dry matter (DM) vs. Prodisol in the 26 th location: 0.61 g/g DM. Nevertheless, in the ANOVA analysis, neither the effect of year and genotypes, nor their interactions were significant as shown in Table 7.
Concerning fatty acid (FA) composition (Table 4), three cultivars (Allstar RM, Melody and Prodisol) had high linoleic acid content, moderate oleic acid content and traces of linolenic acid content, whereas Aurasol presented a high level in oleic acid and moderate linoleic acid content. Palmitic acid was the only major saturated acid (4-6%) either in high oleic or in high linoleic cultivars. There was a significant difference in fatty acid composition (percentage) both between varieties within a year (p<0.001) and between the years for the four cultivars (p<0.001). The genotype x year (G x Y) interaction was not significant for fatty acid composition. In 2002, with a colder and a more rainy summer, 3 varieties (Allstar RM, Aurasol and Prodisol) showed higher content in oleic acid compared to 2003 and 2004.  Tocopherol composition and content. α−tocopherol represented the greatest proportion of total tocopherols, ranging from 88.5 to 97.9%, whereas β-tocopherol was less expressed (1.6 − 9.1%). We observed only traces of γ-tocopherol and δ-tocopherol. The tocopherol content in the seed oil varied between 303.8 mg/kg oil (Allstar in 2003, 85 th location) and 1187.9 mg/kg oil (Prodisol in 2004, 54 th location) (Table 5).

Compound
Genotype ( mainly by climatic conditions during grains filling) significantly affect each tocopherol and total tocopherol content in the oil (p<0.001). No statistical significance for α-tocopherol or total tocopherol content between locations was observed. The correlations between tocopherol isomers were generally positive and highly significant ( Table 8). The α-tocopherol content was closely related to the total tocopherol content (r² = 0.99; p<0.05). Moreover, β-tocopherol and total tocopherol contents were also positively correlated (r² = 0.83; p<0.05).
Phytosterol composition and content. The most abundant sterol in all cultivars tested was β-sitosterol (around 60%), followed by campesterol stigmasterol, Δ7-campesterol and Δ5-avenasterol (Table 6). Δ7-stigmasterol and Δ7avenasterol were also detected and quantified but at very low levels. Total phytosterol content in oils tested ranged between 228.9 mg/100 g oil (Aurasol in 2004 in the 26 th location) and 501.6 mg/100 g oil (Prodisol in 2002 in the 11 th location). The total phytosterol content was significantly higher in 2002 than in 2003 or 2004, but these variations were less important than those observed for tocopherol content. Differences in phytosterol content between years 2002 and 2003 ranged from +4% to -28% among (or: depending on varieties) varieties.
ANOVA (Table 7) showed that the year affect total phytosterol content in oil (p<0.001). Genotype affects also campesterol and stigmasterol contents (p<0.01) but not β-sitosterol content. Location (Year) effect was only significant for campesterol content, but not for the total phytosterol content.
Relationships between fatty acid, tocopherol and phytosterol composition. The total tocopherol content was slightly but significantly correlated with oleic acid (r² = + 0.40; p<0.01), and negatively correlated with linoleic acid (r² = -0.36; p<0.01) ( Table 8). Total tocopherol content (as well as α-tocopherol and β-tocopherol contents) was negatively correlated with palmitic acid (r² = -0.70; p<0.01). Generally, the individual phytosterols and total phytosterol content were not significantly correlated with fatty acid content. This was the case for β-sitosterol, the main sterol in sunflower oil, which did not correlate with any fatty acid. Finally, total tocopherols and total phytosterols were positively correlated (r² = 0.47; p<0.01). Relationships between components composition and meteorological conditions. Correlation between minor components and the meteorological parameters were made to have a first sight of the influence of temperature and rainfall during seed growth. The cumulative temperatures in August showed a negative correlation with the total tocopherol content (r² = −0.69; p<0.05). This negative correlation was also observed for each tocopherol isomer (r² = −0.60; p<0.05) and for the total phytosterol content (r² = −0.38; p<0.05). The same tendency was also found for the cumulative temperatures in June, July and September. On the contrary, there was a positive correlation between the sum of precipitations in August and the total tocopherol and total phytosterol contents (Fig. 3). Campesterol and stigmasterol contents were positively correlated to the sum of precipitation in June (r² = +0.70; p<0.05).

DISCUSSION
Sunflower oil yield is generally influenced by temperature during seed development. Thus, high temperatures during seed maturation may have a negative effect in oil synthesis ( Baydar & Erbas, 2005;Ayerdi Gotor et al., 2006).
Alpha-tocopherol was the dominant isomer detected in this study, similar to the results obtained by Dolde et al. (1999) and Velasco et al. (2002). They only detected traces of δ-tocopherol in the sunflower materials analyzed. Our tocopherol total content was similar to the values reported by Nolasco et al. (2004) (500 to 1200 mg/kg oil). However, the maximum measured value was lower than the one found by Velasco et al. (2002) (562-1872 mg/kg oil). Despite this, Velasco et al. (2002) also found a similar correlation between tocopherols homologues.  This study clearly showed that the variation of total tocopherol content is dependent on genotype and cultivation year. The annual effect was mainly influenced by the meteorological conditions such as temperature. Results obtained show that the tocopherol content in sunflower oil is highly correlated to temperature. This is in agreement with results of Almonor et al. (1998) on soybean oil and Ayerdi Gotor et al. (2006) on sunflower oil. Both studies showed a reduction in total tocopherol content when plants were exposed to high temperatures during grain filling. However, Britz and Kremer (2002) reported that total tocopherol content in soybean oil remains constant despite differences in growth temperatures. The reduction in tocopherol content during a hot year could be explained by the role of these molecules in the plant. Tocopherols, particularly α-tocopherol protects the plants against heat stress by scavenging singlet oxygen and lipid peroxil radicals leading to decreased (detected) amounts of tocopherols (Munné-Bosch, 2005). Previous studies concerning tocopherol variation in sunflower and in other plant species (Peterson & Qureshi, 1993;Goffman & Becker, 2002;Marwede et al., 2004;Fernández-Cuesta et al., 2012) show that genotype and environment either individually or their interaction affect the content and composition in seeds.
The range found on phytosterol content is close to the one reported by Vlahakis and Hazebroek (210 to 454 g/100 g oil) and others authors (Vlahakis & Hazebroek, 2000;Ayerdi Gotor et al., 2008;Fernández-Cuesta et al., 2012). As mentioned previously, the climate differed between years; 2002 was cold and wet, while 2003 was excessively warm and dry. The data presented indicate that low temperatures together with precipitation correlate with an increase in the total phytosterol content. These results are in agreement with those found by Zangenberg et al. (2004) in rye (Secale cereale L.) grains. On the contrary, Vlahakis and Hazebroek (2000) showed that the total phytosterol levels in greenhouse grown soybean seed increase under higher temperatures under controlled conditions. However, Hamama et al. (2003) did not show significant differences concerning phytosterol content between different genotypes of canola (Brassica napus L.), indicating that environmental conditions may affect the various species in a different manner.
The positive correlation observed between campesterol, stigmasterol and β-sitosterol indicate that the amount of different phytosterols could be simultaneously increased. In the literature, little is known concerning correlations between phytosterols contents. Fernandez-Cuesta et al. (2012) found a negative correlation between campesterol and stigmasterol, and between stigmaterol and β-sitosterol , but a positive correlation between campesterol and β-sitosterol. In canola oil, the same order of slight positive correlations between phytosterol components were observed (Hamama et al., 2003). In contrast, Hamama et al. (2003) found negative correlations between both campesterol and β-sitosterol content and campesterol and Δ5-avenasterol contents. In this study, the only significant positive correlation was between brassicasterol and total phytosterol contents (r² = +0.40; p<1%).
Similar to the results of our study, Baydar and Erbas (2005) reported significant correlations between tocopherols and oleic acid content (r² = +0.79) as well as tocopherols and linoleic acid content (r² = -0.82).
The negative correlation between β-sitosterol and campesterol or stigmasterol could be explained due to the different pathways of biosynthesis of β-sitosterol and the others phytosterols (Schaller, 2003). Indeed, Schaller (2004) identified in (1) Arabidopsis the Δ24-isomerasereductase which produces β-sitosterol and campesterol, and (2) higher plants a Δ22-desaturase implicated in the transformation of β-sitosterol to stigmasterol and campesterol to brassicaterol (Vlahakis & Hazebroek, 2000). Such enzymes were species, genotype and growthtemperature dependent (Garcés et al., 1992;Almonor et al., 1998). These previous results could contribute to explain the absence of correlation between β-sitosterol and stigmasterol, and the high correlation between campesterol and stigmasterol observed herein.
The correlation between total tocopherols and total phytosterols for sunflower oil in this study are in agreement with earlier results obtained on 12 cultivars from the same planting location (r² = +0.69, p<0.05) (Vlahakis & Hazebroek, 2000).
The relationships between fatty acids, tocopherols and phytosterols and the climatic conditions during plant growth found in this study cannot be compared with data obtained by El-Shami et al. (2001) in rapeseed oil. In their study, correlations were obtained with values from only one genotype and one cultivation year, study which did not reflect any environmental influence.
However, this study indicated that both genotype and the growing conditions had a stronger influence on tocopherol composition and content than on phytosterol ones. Moreover, the contents of tocopherol, oleic acid percentage and phytosterol contents were positively correlated. Thus, it would be possible to increase tocopherol and phytosterol contents including high oleic acid content simultaneously through a breeding program.