Milk thistle seeds were collected from plants growing wild in Tunisia (region of Sousse), during April and June, 2012. Seeds were picked according to external color; green seeds were chosen as immature stage, mahogany brown seeds as the intermediate stage and dark brown seeds as the last stage of maturity (mature stage). 100 g of seeds were dried at 50 °C and then ground to fine powder in a grinder.

The oils were extracted using petroleum ether in a Soxhlet extractor for 4 h. The solvent was initially removed using a rotary evaporator at 40 °C. Oil samples were placed at ambient temperature (25–35 °C).

The unsaponifiable fraction was determined by saponifying 5 g of oil extracts with 50 mL ethanolic KOH 12% (w/v) and heating at 60 °C for 1.30 h. After cooling, 50 mL of H₂O were added. The unsaponifiable matter was extracted four times with 50 mL of petroleum ether. The combined petroleum ether extract was washed with 50 mL of ethanol–water (1:1). The extracted ether was dried over anhydrous Na₂SO₄ and evaporated to dryness using a rotary evaporator. The dry residue was dissolved in chloroform for TLC analysis.

The unsaponifiable matter was separated into sub-fractions on preparative silica gel thin-layer plates (silica gel 60G F254) using one-dimensional TLC with hexane–diethyl ether (6:4, v/v) as the mobile solvent. The unsaponifiable fraction diluted in chloroform was applied on the silica gel plates. After development, the plate was sprayed with 2,7-dichlorofluorescein and viewed under UV light. The band corresponding to sterols was scraped, extracted three times with chloroform–diethyl ether (1:1, v/v), filtered to remove the residual silica, dried in a rotary evaporator and stored at −10 °C.

GC-MS analyses were performed using a capillary HP-5MS column (30 m×0.25 mm I.D., 0.25 µm film thickness; Agilent Technologies) with gas chromatography (Agilent Technologies 7820A) coupled directly to the mass detector (Agilent Technologies 5975 series MSD). Helium was used as carrier gas, with a constant flow rate of 1 ml/ min. The injector and detector temperatures were 230 °C. The oven temperature was programmed from 150 to 320 °C at 10 °C·min⁻¹ from 150 to 250 °C and at 5 °C·min⁻¹ from 250 to 320 °C. Electronimpact mass spectra were measured at acceleration energy of 70 eV. Manual injection of 1 µL of the sterol solution was performed in the split mode at a 10:1 split ratio. The phytosterol compounds were identified by comparing their relative retention times and mass spectra with those of the authentic standard. The peaks were also confirmed by comparison with the Wiley 275.L Mass Spectral Library.

A statistical analysis was performed by using the Proc ANOVA in SAS (Software version 8). Duncan’s Multiple Range Test was used. For each oil sample, three determinations have been made.

During seed maturation the oil content increased to a maximum value of 35.8±1.3% and then declined to reach a value of 30.5±0.9% at full maturity. More oil was synthesized during the early stage of seed development. Malekzadeh et al. (2011) reported that the total oil content of milk thistle seeds decreased under drought stress. In mature seeds, oil was stored in the form of oil bodies (Voelker and Kinney, 2001). The extracted oil from milk thistle seed has been suggested as suitable as an edible oil (EL-Mallah et al., 2003).

The results obtained showed that the amount of total unsaponifiable matter decreased during seed maturation (Table 1). The greatest change occurred during the early stage of seed development. Thus, the highest level of unsaponifiable matter (3.8±1.2%) was detected in immature seeds. At full maturity, the unsaponifiable lipid content of the studied seeds was 1.9±0.2% of the total oil. The total amount of unsaponifiable matter in olive oils ranged from 1 to 2% of the total lipids (Alonso-Salces et al., 2009). Overall, the level of unsaponifiable matter ranged from 0.5 to 2.5% of the total lipids (Małecka, 2002). These minor lipids greatly influence the organoleptic quality and stability of the oil (Alonso-Salces et al., 2009). The effectiveness of unsaponifiable matter in retarding oil deterioration has been demonstrated by many researches (Mohamed and Awatif, 1998; Gopala Krishna et al., 2003).

The unsaponifiable fraction is made up of minor constituents (sterols, triterpene alcohols, aliphatic alcohols, hydrocarbons, etc.), which may vary both qualitatively and quantitatively depending on genetic factor, climatic conditions, extraction and refining procedures, as well as storage conditions (Canabate-Díaz et al., 2007). 4-Desmethylsterols were isolated from the unsaponifiable matter and represented 55.2% of this fraction and about 1% of the total oil. This high proportion shows that milk thistle seed oil is one of the richest natural products in phytosterols versus other vegetable oils frequently used in the diet, such as olive oil (0.17%) (Weihrauch and Gardner, 1978). These results are useful for the consumer and also for the oil producer. Indeed, the unsaponifiable fraction of vegetable oils has applications in cosmetics and pharmacology due to its biological properties.

Phytosterols are important due to their impact on health. Therefore, readily available food products have been engineered to be enriched in phytosterols and marketed to help lower serum cholesterol and reduce the risk cardiovascular disease. 4-Desmethylsterols are the major components of phytosterol matter in most vegetable oils (Azadmard-Damirchi and Dutta, 2006). Eight 4-desmethylsterols were detected in the oil analyzed (Figure 1). The changes in sterol composition during the development of seeds were summarized in Table 2. The sterol content of milk thistle seeds was affected by the ripening degree of the seeds. At the early stages of seed maturation, Δ⁷-stigmastenol was the most abundant sterol followed by β-sitosterol. As the seed developed, the level of Δ⁷-stigmastenol reduced from 52.84±1.5 to 27.81±0.5% of the total sterol content. This result could be explained by the high activities of Δ⁷-sterol Δ⁵-desaturase and Δ⁷-sterol reductase during the early stages of seed development. In fact these two enzymes are involved in the conversion of Δ⁷-sterols to Δ⁵-sterols (Benveniste, 2002). The Δ⁵-sterols are mainly accumulated in the plasma membrane, where they are believed to regulate the membrane fluidity (Grandmougin et al., 1989). The amounts of campesterol and stigmasterol increased gradually during seed maturation. However, the level of Δ⁷-avenasterol decreased as the seed developed. The results obtained show that the levels of Δ⁷-campesterol and Δ⁵-avenasterol were relatively constant. The amount of cholesterol was very low in immature seeds (<2%). The rate of cholesterol accumulation was found to be greatest at the late stage of seed maturation.

β-Sitosterol (46.50±0.8%) was the most abundant compound followed by Δ⁷-stigmastenol (27.81±0.5%). The β-sitosterol content determined was much higher than that in the milk thistle cultivars grown in Iran (33–37%) (Fathi-Achachlouei and Azadmard-Damirchi, 2009), but was lower than that of the Egyptian cultivars (57.4%) (El-Mallah et al., 2003). The β-siosterol amount in milk thistle seeds is affected by environmental conditions and genotypes. The β-sitosterol range in olive oil is 34–66% (Mezghache et al., 2010). The health aspects of β-sitosterol, the most common phytosterol, have recently been reported in several studies (Awad et al., 1998).

The Δ⁷-sterol compounds were mainly represented by Δ⁷-stigmastenol (27.81%), Δ⁷-avenasterol (3.80±0.2%), and Δ⁷-campesterol (2.88±0.2%). The level of total Δ⁷-sterols (34.49%) detected in this studied oil sample was much higher than that in the Iranian samples (19–22%) (Fathi-Achachlouei and Azadmard-Damirchi, 2009). This contrasts with the composition of corn oil where Δ⁷-stigmastenol amounted to 2% of the total desmehyl sterol content (Harrabi et al., 2007). Important differences may occur in some plant families. For instance, many plants belonging to the order of caryophillales contain large amounts of Δ⁷- sterols; spinach and chenopodiumrubrum contain almost only Δ⁷-sterols such as spinasterol or stigmast-7-ene-3β-ol (Benveniste, 2002).

The campesterol (4.2±0.5%) and stigmasterol (5.47±0.4%) contents of the seed oil of this study were comparable to those of the Iranian milk thistle cultivars (Fathi-Achachlouei and Azadmard-Damirchi, 2009), but were higher than those of olive oil where the sum of these two sterols was less than 5% (Mezghache et al., 2010). The amount of Δ⁵-avenasterol was 3.20±0.1%. In the literature, this component has been associated with antioxidant effects (Williamson, 1998; Blekas and Boskon, 1999). Yoshida and Niki (2003) reported that campesterol, stigmasterol and clerosterol exerted antioxidant effects on the oxidation of a methyl linoleate oil solution. Clerosterol was not detected in the studied seed oils and in the Egyptian samples (El-Mallah et al., 2003), but was detected in the Iranian samples (Fathi-Achachlouei and Azadmard-Damirchi, 2009). Moreover, sitostanol and campestanol were absent in this studied oil sample. This result agrees with these reported by Fathi-Achachlouei and Azadmard-Damirchi (2009) and El-Mallah et al. (2003). However, Dabbour et al. (2014), detected these two stanols in cold-pressed milk thistle seed oil (campestanol 0.21%, sitosanol 1.67%).

The results obtained show that cholesterol represents 3.91±0.3% of the total 4-desmethyl sterol fraction. It was lower than that in the milk thistle cultivars grown in Iran (9.5%) (Fathi-Achachlouei and Azadmard-Damirchi, 2009) and in the cold-pressed seed oil of the cultivar grown in Jordan (15.14%) (Dabbour et al., 2014). Consequently, it could be suggested that the unrefined oils obtained by extraction with an organic solvent had a lower level of cholesterol than the cold-pressed oils. Cholesterol is the predominant sterol in animal fats and fish oils, but is very rare in vegetable oils. It is known that cholesterol occurs in the sterol fraction of many vegetable oils as a minor component and usually amounts to 1% of the total sterol content (Phillips et al., 2002), significantly lower than that detected in the milk thistle seed oil. It has been recognized that plant sterols could reduce plasma cholesterol levels in humans (Ostlund et al., 2002). The mechanism of cholesterol reduction in the presence of phytosterols is based on the blocked absorption of it in the digestive tract. Since the level of cholesterol is low and in the presence of excessive amounts of phytosterols, it can be expected that its absorption will be minimal and the positive effect of phytosterols will overcome it (Ostlund et al., 2002). Furthermore, cholesterol biosynthesis in higher plants has not been studied extensively and thus, uncertainties exist in the sequence of intermediates. The enzymatic approach to understanding and controlling the formation of the sterol structure was hampered by the low existence of sterol enzymes in higher plants. Therefore, milk thistle seeds which had a higher level of cholesterol as compared with the other seeds could be a good example for the study of the cholesterol biosynthetic pathway.