CG fruit (P. peruviana L.) of Colombian origin was purchased from a local supermarket in January 2019. Undamaged fruit of uniform ripeness were selected, then two separate sub-samples were formed, seeds and berry peels. Those were obtained by careful manual isolation of peels and seeds from individual berries in order to make distinguishable samples of berry structural elements. Fruit samples then were kept in the refrigerator at a temperature of -18 °C until analysis. Another sample constituted air-dried seed and peel mixture (seed/peel waste, seed/peel pomace) remaining after a high-speed vacuum separation of fruit juice. Seed cakes, in turn, were taken after the Soxhlet extraction of seed oil with n-hexane and air dried. The moisture content of the samples was determined by oven-drying at 103 ± 2 °C to constant weight.

The isolation of the oil (%, v/w) from fruit peels and seeds was carried out by extraction with n-hexane in a Soxhlet apparatus for 8 h. The solvent was then completely evaporated in a rotary vacuum evaporator at a temperature of 40 °C (ISO 659:2014). The FA composition of the oils was determined by GC analysis after transmethylation of the sample with 2% H₂SO₄ in CH₃OH at 50 °C (ISO 12966-1:2014, ISO 12966-2:2011). Determination was performed on a Hewlett Packard 5890 А gas chromatograph equipped with a capillary Supelco 2560 column, 75 m × 0.25 mm × 18 μm (i.d.), and a flame ionization detector (FID). The column temperature was programmed from 130 °C (4 min) to increase at 15 °C/min to 240 °C (5 min); injector and detector temperatures were 250 °C; hydrogen was the carrier gas at a 0.8 mL/min flow rate; the split ratio was 50:1. The identification of FAs was performed by comparison of the retention times with those of a standard mixture of fatty acid methyl esters (FAME) (37 component FAME mix, Supelco, USA), according to ISO 12228-1:2014.

Tocopherols were determined directly by HPLC analysis using a Merck-Hitachi unit equipped with a 250 mm × 4 mm Nucleosil Si 50-5 column and a fluorescent Merck-Hitachi F 1000 detector. The operating conditions were as follows: mobile phase n-hexane:dioxan (96:4, v/v); flow rate of 1.0 mL/min; detector excitation at 295 nm, emission at 330 nm; injected sample volume of 20 μL (1 g/100 mL solution of crude oil in n-hexane). Tocopherols were identified by comparison with the retention times of reference tocopherol standards (DL-α-, DL-β-, DL-γ- and DL-δ-tocopherols with 98% purity, purchased from Merck, Darmstadt, Germany), according to ISO 9936:2016.

For the determination of sterols, the unsaponifiable fraction was isolated after saponification of the oil and extraction with n-hexane, according to ISO 18609:2000. Sterols were identified by comparison of the retention times with those of a standard sterol mixture containing cholesterol (stabilized, purity 95%, Across Organics, New Jersey, USA), stigmasterol (purity 95%, Sigma-Aldrich, St. Louis, MO, USA) and β-sitosterol (with ca 10% campesterol, ca 75% β-sitosterol, Across Organics, New Jersey, USA), according to ISO 12228-1:2014.

The phospholipid content was determined by spectrophotometry, measuring the content in phosphorus at 700 nm after mineralization of the oil with a solution of HClO₄ and H₂SO₄ (1:1, v/v) (ISO 10540-1:2014).

The seed/peel fraction and seed cakes were mineralized at 450 °C; the residue was first dissolved in concentrated HCl, evaporated to dryness and then dissolved again in 0.1 mol/L HNO₃. The concentrations of mineral elements were determined by using an atomic absorption spectrophotometer (AAS) Perkin Elmer/HGA 500 (Norwalk, USA). The respective wavelengths for the AAS detection were: Na - 589.6 nm, K - 766.5 nm, Mg - 285.2 nm, Ca - 317.0 nm, Zn - 213.9 nm, Cu - 324.7 nm, Fe - 238.3 nm, and Mn - 257.6 nm. Elemental identification was carried out by comparing with a standard solution of metal salts, and the calculation of the respective metal concentrations were based on a standard solution of 1 μg/mL, using a calibration curve.

The total protein content in CG seed cakes was determined by the Kjeldahl method, as described in AOAC Method 976.06 (2016), on a UDK 152 System (Velp Scientifica, Italy).

The hydrolysis of protein to free aminoacids was completed by reacting 30 mg dried seed cake with 3 mL 6N HCl in a sealed glass ampule at 105 °C for 24 h. The solvent was evaporated in a vacuum chamber at 40 °C and the residue was fully diluted in 20 mM HCl (2 mL). After filtration, 20 μL of the solution were derivatized with AccQ-Fluor kit (WATO52880, Waters Corporation, USA). Finally, the solution was heated to 55 °C and 20 μL were injected into an ELITE LaChrome HPLC chromatograph (Hitachi) equipped with a diode array detector (DAD) and a reverse phase С 18 AccQ-Tag (3.9 mm × 150 mm) column. The mobile phases in the gradient elution were WATO52890 buffer (Waters Corporation, USA) and 60% acetonitrile. The detection wavelength was 254 nm and column temperature was 37 °C.

The content of cellulose (crude fiber) in the CG seed cakes was determined by a modification of the method of Brendel et al., (2000). The hydrolysis of cellulose and hemicellulose was carried out by boiling 1 g of seed cakes in 16.5 mL of 80% CH₃COOH and 1.5 mL concentrated HNO₃ for 1.5 h. After filtration of the suspension, the solid residue was dried at 105 °C for 24 h and weighed.

The measurements were made in triplicate (n = 3), and the results were presented as mean values of the individual measurements with the corresponding standard deviation (mean ± SD). Statistical tools, such as ANOVA and Tukey’s multiple comparison test, were used to determine significant differences (p < 0.05).

As stated above, CG seeds and seed/peel waste have been reported as promising sources for obtaining edible oil (Ramadan and Mörsel, 2003; Ramadan, 2012). In turn, the shiny surface of the CG fruit is also associated with the presence of lipid fractions in the peel, so in this study the two oil-containing fruit elements - seeds and peels, were evaluated separately in order to trace their individual contributions. The pulp (juice) was not analyzed alone, as previous studies showed that it contained considerably smaller amounts of oil (0.2% FW) (Ramadan, 2011). Results on the total oil content and the general composition of the oil isolated from the respective fruit elements are presented in Table 1.

As anticipated, the data revealed considerable numerical differences in the lipid fractions of CG fruit elements. The oil content in the peels was about seven times lower than that in the seeds and seed/peel waste; the latter two showing insignificant variation. The content of the oil extracted from CG seeds and seed/peel waste (22.93% and 21.03%, respectively) was higher than that of some common oilseed crops, e.g. corn (3-5%), cotton (16%) and soybean (18%) (Popov and Ilinov, 1986), or some prospective fruit seeds, such as grape seeds (8-20%) (Heuzé and Tran, 2017). The phospholipid concentration in the peel oil was about twice as high as that in the seed/peel oil, and about four times higher than that in seed oil. The contents in phospholipids in the three CG oils were higher than the respective data for sunflower, corn germ and safflower oil (0.4-0.9%) or for soybean oil (1.0-3.0%) (Popov and Ilinov, 1986). The contents in sterols, a known group of biologically active dietary cholesterol-reducing agents, were similar in the three fruit fractions, and higher than many seed oils, such as sunfower, soybean, cotton, safflower, and others (0.24-0.64%) (Popov and Ilinov, 1986; FAO/WHO, 1999). The total amount of biologically active tocopherols in the two seed-containing CG fruit samples (5096 and 5634 mg/kg, respectively) was twice as high as that in the peel oil (2648 mg/kg). These results were comparable to the tocopherol values for common oils, such as soybean (600-3370 mg/kg), maize (330-3720 mg/kg) and rapeseed (430-2680 mg/kg) (FAO/WHO, 1999).

The FA composition of the analyzed GC oil is presented in Table 2. The results revealed significant differences in the FA composition of CG peel oil compared to the seed and seed/peel oils. The most abundant FAs in the CG peel oil, out of 14 identified, were capric, palmitic and oleic acids. The ratio between saturated (SFA) and unsaturated (UFA) FAs was 67.7:32.3, and the ratio between monounsaturated (MUFA) and polyunsaturated (PUFA) FAs was 23.7:8.6. Linoleic, oleic and palmitic acids were the dominating FAs in the seed and seed/peel waste oils. These results suggested that both CG wastes from juice production could be used as alternative sources of vegetable oils rich in linoleic (63.19-67.89%) and oleic (14.69-16.56%) acids, which are important for the prevention of cardiovascular diseases - as are the seed oils of grapes, melon, tobacco, poppy, and others (Popov and Ilinov, 1986). The ratios of SFA to UFA were similar in the two oils, 17.7:82.3 and 16.2:83.8, respectively, as well as the PUFA-to-MUFA ratios (63.6:18.7 and 68.3:15.5, respectively).

The comparative analysis of our results and literature data about CG fruit oil revealed that there were some differences in oil yield and FA composition. It should be noted that most of the available studies on CG oil and its FAs were carried out either on whole (intact) fruit, seeds or pomace (the combined seed and peel waste from juice separation), so it was difficult to compare our results for CG peel alone. In the study by Yıldız et al., (2015) the n-hexane extracted oil from whole fruits harvested at Bursa, Turkey, was 0.18% (on a fresh weight basis, FW), while the yield reported by Ramadan and Mörsel, (2003) was 2.0% oil of berry weight, in which seed oil comprised 1.8% and pulp/peel oil, 0.2%. In another study, the n-hexane extracted oil from seed/peel pomace was 19.3% (Ramadan, 2012), which was very close to our results.

In terms of FA profiles, our results differed only numerically from the data provided by previous studies. For example, Rodrigues et al., (2009) identified linoleic acid (72.42%), oleic acid (10.03%) and palmitic acid (9.38%) as the main FAs in the lipid fraction of fruit from Brazil; the SFAs were 12.87% of total FAs. Similar results were reported by Mokhtar et al., (2018) for seed/peel waste powder, in which four FAs were quantified, linoleic (77.78%), oleic (11.32%), palmitic (7.39%), and stearic (3.51%), at a SFA:UFA ratio of 10.9:89.1. In the study by Ramadan and Mörsel, (2003) the contents of the main FAs of CG seed oil were: linoleic (76.1%), oleic (11.7%), palmitic (7.29%), and stearic (2.51%), with a SFA:UFA ratio of 12.8:87.2. The FA profile of the seed/peel extracted oil was also dominated by linoleic (77.1%), oleic (10.3%), palmitic (7.95%), and stearic (2.61%), at a SFA:UFA ratio of 12.8:88.2 (Ramadan, 2012). As it can be seen, the ratio of linoleic to oleic acid in these studies was about 7:1. While our results suggested a lower value, about 4:1; the relative share of UFA was slightly lower in this study, too (SFA:UFA ratios of 17.7:82.3 and 16.2:83.8, for seed/peel and seed oils). Nevertheless, the results from this study supported the evaluation of CG fruit oil as especially suitable for nutritional application, due to its high proportions of PUFAs.

The individual sterol composition of the oils is presented in Table 3. The results revealed that the total sterol content in the unsaponifiable fraction was comparable in seed and seed/peel oils (33.33-44.59%), but that it was considerably lower in the peel oil (2.14%). The total amount of unsaponifiables was in a reversed order, i.e. higher in the peel oil (61.33%) compared to seed and seed/peel oils (3.02-4.21%). With a share of 80.23% in the total sterol content, β-sitosterol was the dominating sterol in the oil of CG peels, while the other two oils contained campesterol, Δ⁵-аvenasterol and β-sitosterol as their main sterols (constituting 67.65% and 77.80% of the sterol fraction in seed and seed/peel oil, respectively). These results were very close to the findings by Ramadan, (2012), who also reported high levels of unsaponifiables in CG pomace oil (22.0 g/kg), and campesterol (4.70 g/kg), Δ⁵-аvenasterol (2.63 g/kg) and lanosterol (1.60 g/kg) as responsible for about 75% of total sterols.

The results from the analysis of the individual tocopherol composition of CG oil are presented in Table 4. Four tocopherols were identified in each of the oils, and the data revealed that γ-tocopherol was noticeably the dominating structure in the peel oil, while β-tocopherol, δ-tocopherol and γ-tocopherol had similar individual shares (31.40-34.15%) in both seed and seed/peel waste oils. As tocopherols are known to be potent antioxidants, it can be suggested that the high amount of tocopherols in the studied CG oils, and especially the high γ-tocopherol levels, would be a preventive factor against lipid oxidation processes during the storage of oil and oil-containing products.

These results were in compliance with the findings by Ramadan and Mörsel, (2003), who also identified γ- and α-tocopherols as the main constituents in the pulp/peel oil of CG fruit (45.5 mg/kg and 28.3 mg/kg, respectively), and β-tocopherol and γ-tocopherol in the seed oil. In another study (Ramadan, 2012), β-tocopherol comprised 47% of the tocopherol fraction in the CG seed/pulp oil (2.10 g/kg), γ-tocopherol was 26% (1.08 g/kg), δ-tocopherol 18.5% (0.85 g/kg), and α-tocopherol about 6% (0.34 g/kg). There were no sufficient data in the literature about the tocopherol content in CG fruit peel alone, so it was difficult to make a more detailed comparison of our results with previous findings.

As already stated, the two by-products obtained from CG fruit, seed cakes and seed/peel waste, represent raw materials which are underutilized, but valuable in terms of both quantity and functionality. In this study, the peels were found to constitute 11.4% of fresh fruit weight, and the seeds 13.2%. For example, the combined seed/peel fraction made up nearly a quarter of the fruit weight. Our results were very close to the findings by Ramadan, (2011) for CG seed/peel pomace at 27.4% of fruit weight. On the other hand, CG waste fractions were previously indicated as potential sources of nutrients, e.g. vitamins, minerals, carbohydrates, protein, etc. (Ramadan et al., 2008; Ramadan, 2011, 2012). On these grounds, the two waste products from fruit processing in this study were considered worthy of further analysis in order to determine the contents in some components with nutritional value, e.g. minerals, fiber, protein, and amino acids.