Comparative phytochemical, antioxidant, and hemostatic studies of fractions from raw and roasted sea buckthorn seeds in vitro

Various seeds, including sea buckthorn (Hippophae rhamnoides L.) seeds, are sources of different bioactive compounds. They can show anti-inflammatory, hypoglycemic, anti-hyperlipidemic, antibacterial, antioxidant, or other biological properties in in vitro and in vivo models. Our preliminary in vitro results have demonstrated that the extracts from raw (no thermal processing) and roasted (thermally processed) sea buckthorn seeds have antioxidant potential and anticoagulant activity. However, it was unclear which compounds were responsible for these properties. Therefore, in continuation of our previous study, the extracts were fractionated by C18 chromatography. Phytochemical analysis of three fractions (a, b, and c) from raw sea buckthorn seeds and four fractions (d, e, f, and g) from roasted sea buckthorn seeds were performed. Several in vitro assays were also conducted to determine the antioxidant and procoagulant/anticoagulant potential of the fractions and two of their major constituents—isorhamnetin 3-O-β-glucoside7-O-α-rhamnoside and serotonin. LC–MS analyses showed that serotonin is the dominant constituent of fractions c and f, which was tentatively identified on the basis of its HRMS and UV spectra. Moreover, fractions c and f, as well as b and e, contained different B-type proanthocyanidins. Fractions b and e consisted mainly of numerous glycosides of kaempferol, quercetin, and isorhamnetin. The results of oxidative stress assays (measurements of protein carbonylation, lipid peroxidation, and thiol groups oxidation) showed that out of all the tested fractions, fraction g (isolated from roasted seeds and containing mainly dihexoses, and serotonin) demonstrated the strongest antioxidant properties.


Plant material
Sea buckthorn was obtained from a horticultural farm in Sokółka, Podlaskie Voivodship, Poland (53° 24′ N, 23° 30′ E), the largest plantation (33 ha) of this fruit in Poland.The plant material was identified by Mr. Stanisław Trzonkowski, the owner of the plantation.The ripe fruit was harvested in the second half of August, 2017.Whole branches were provided.A voucher specimen (IUNG/HRH/2017/1) was deposited at the Department of Biochemistry and Crop Quality, Institute of Soil Science and Plant Cultivation, State Research Institute, Puławy, Poland.The use of sea buckthorn in the present study complies with international, national and/or institutional guidelines.
The fruit was picked manually and frozen.To isolate the seeds, the thawed fruit was homogenized in water, using a blender.The fruit pulp and skins were washed out with tap water, and the sedimented seeds were collected, rinsed with methanol (to remove excess of water and facilitate drying), drained, and air-dried.The dry seeds were stored at room temperature.
A portion of sea buckthorn seeds was subjected to thermal treatment in a baking and pastry company, as described before 8 .The applied roasting conditions were chosen to mirror the process of rye bread baking.The temperature in the baking chamber at the beginning (approx.10 min.)was 280 °C, the relative humidity was 35%.In the second phase, the temperature was decreased to 175 °C.The total roasting time was 36 min.

Preparation of fractions a-g
The method of preparation of the extract from roasted seeds of sea buckthorn was described previously 8 .Briefly, the seeds were powdered in a coffee grinder, defatted by maceration in hexane at room temperature (30 min.),drained, and air-dried.The ultrasound-assisted extraction of the plant material (91 g) with 80% methanol (v/v) and pure methanol comprised 4 steps: 500 mL 80% methanol (30 min.),400 mL 80% methanol (30 min.),300 mL 80% methanol (30 min.),and 200 mL 100% methanol (20 min.).The extracts were filtered, pooled, concentrated in a rotary evaporator, and defatted by liquid-liquid extraction with hexane.The defatted extract was rotaryevaporated to remove all organic solvents and subjected to liquid-liquid extraction with butanol.Pooled butanol extracts were dried in a rotary-evaporator, dissolved in 20% tert-butanol, and lyophilized (Gamma 2-16 LSC, Christ, Osterode am Harz, Germany) to yield 1.93 g of the roasted seed extract.
To obtain fractions with distinct groups of compounds, the roasted seed extract was separated by reversedphase liquid chromatography.A 1.04 g portion of the extract was shaken with 45 mL of 1.5% methanol with 0.1% formic acid (v/v), sonicated (3 min.),and centrifuged.The supernatant was loaded onto a short C18 open column (Cosmosil 140C18-PREP; self-packed, 66 mm × 44 mm), equilibrated with the same solvent.The column was washed with 300 ml of the solvent to remove highly polar compounds.The bound compounds were eluted sequentially with portions of 20%, 66%, and 85% methanol (300 mL each).The solubility of the extract in 1.5% methanol was poor.To separate all available extract, the pellet that remained after centrifugation was dissolved in 4.5 mL of methanol, and the solution was subsequently filled up with water to the volume of 45 mL, and acidified with 45 µL of formic acid.Then, it was centrifuged, and the obtained supernatant was loaded onto the same column, equilibrated with 10% methanol + 0.1% formic acid.The column was washed with the same solvent (200 mL), and the elution with 20%, 66% and 85% methanol was repeated, as described above.The pellet obtained after the second centrifugation was dissolved in 9 mL of methanol, and the solution was filled up to the volume of 45 mL and acidified with 45 µL of formic acid.The mixture was centrifuged, the supernatant was loaded onto the column equilibrated with 10% + 0.1% formic acid; the chromatographic separation was performed again.Corresponding eluates, obtained during the above-described chromatographic separations, were pooled, rotary evaporated to remove methanol, and freeze-dried, to yield 0.324 g of faction g (the 1.5% methanol washing), 0.106 g of fraction f (20% methanol eluates), 0.308 mg of fraction e (66% methanol eluates), and 0.147 g of fraction d (85% methanol eluates).
The extract from 52 g of raw seeds was prepared in a similar way (1.02 g).The fractionation of the extract was performed as described above, to yield 0.064 g of fraction c (20% methanol eluates), 0.223 g of fraction b (66% methanol eluates), and 0.121 g of fraction a (85% methanol eluates).

Phytochemical analysis of the fractions a-g
The composition of the investigated fractions was analyzed by UHPLC-ESI-HRMS/MS, using a Thermo Ultimate 3000RS (Thermo Fischer Scientific, MS, USA) chromatographic system, equipped with a charged aerosol detector and coupled with a Bruker Impact II Q-TOF mass spectrometer (Bruker Daltonics GmbH, Bremen, Germany).Fraction g was dissolved in 1% methanol; fractions c and f-in 20% methanol; fractions b and e-in 50% methanol; fractions a and d-in 70% methanol.All samples had the same concentration (2.4 mg mL −1 ), except for fraction g, which was dissolved four times (to avoid unnecessarily contaminating the CAD detector and the ESI ions source with large amounts of unretained substances).The dissolved samples were centrifuged and analyzed.Chromatographic separations were performed at 40 °C, on an ACQUITY UPLC HSS C18 column (2.1 × 100 mm, 1.8 µm; Waters, MA, USA), equipped with a respective pre-column.The mobile phase A was 0.1% formic acid in MilliQ water, the mobile phase B was acetonitrile containing 0.1% of formic acid.The flow rate was 0.400 mL min −1 .The injection volume was 3 µL.Fractions b, c, e, f, and g were chromatographed using the following elution program: from 0.0 to 1.0 min.-5%B, 1.0 to 26 min.-aconcave gradient to 80% B; the column was subsequently washed with 99% B (26.5-28.0min.),then the mobile phase returned to initial conditions (28.0-29.0min.),and the column was equilibrated for 1 min.In the case of fractions a and d, the described elution method was slightly modified, the concave gradient was replaced by a linear one.HRMS analyses were performed using negative and positive ion modes.The following settings were applied in negative mode: the scanning range was m/z 80-2000; capillary voltage was 3 kV; dry gas flow was 10 L min −1 ; dry gas temperature was 220 °C; nebulizer pressure was 2.0 Bar; collision RF was 750 Vpp; transfer time was 100 µs; prepulse storage time was 10 µs.Collision energy was set automatically in the range from 2.5 to 80 eV, depending on the m/z values of fragmented ions.Settings for the positive ion mode: the scanning range was m/z 100-2000; capillary voltage was 4 kV; the remaining parameters were almost the same as those described above, but the automatically set collision energies ranged from 2.5 to 50 eV.The obtained data were calibrated internally with a solution of sodium formate, which was introduced to the ion source via a 20 μL loop at the beginning of each separation.However, when the above-described MS method was applied, triterpenoid saponins were ionized very weakly, and no fragmentation data was obtained for many of them.For this reason, the fraction d was subjected to another LC-MS analyses.The flow rate and the column temperature were as described above.The solvent A was 0.1% formic acid and 10 mM ammonium formate in MilliQ water, the solvent B was acetonitrile containing 0.1% formic acid.The injection volume was 2 µL.The elution method was as follows: from 0.0 to 1.0 min.-5%B; from 1.0 to 17.0 min.-alinear gradient to 90% B; from 17.0 to 20.5 min.-90%B; from 21.0 to 25.0 min.-5%B. HRMS analyses were performed in negative and positive ion mode.The following settings were applied in negative ion mode: the ion source settings were not modified; collision RF was 750 Vpp; transfer time was 120 µs; prepulse storage time was 10 µs.Collision energy was set automatically in the range from 20 to 60 eV, depending on the m/z values of fragmented ions.The settings for positive mode: the ion source settings were not modified; collision RF was 750 Vpp; transfer time was 120 µs; prepulse storage time was 10 µs.Collision energy was set automatically in the range from 15 to 40 eV.The use of the mobile phase with ammonia formate caused that saponins were effectively ionized in the positive ion mode, frequently forming adducts with two ammonia ions, and were fragmented by collision-induced dissociation.Samples e, and f were additionally analyzed using an ACQUITY UPLC chromatographic system (Waters MA, USA), equipped with a diode array detector (DAD) and a triple quadrupole mass detector (TQD), to obtain UV spectra of the separated substances, and facilitate their identification; samples were chromatographed on an ACQUITY BEH C18 column (2.1 × 100 mm; Waters, as described by Sławińska et al. 8 . Constituents of the samples were tentatively identified on the basis of their MS spectra (with the help of Compound Crawler 3.3, a database search software (Bruker)), UV spectra (when possible), and literature data.

Isolation of pure compounds
Isolation and structure determination of isorhamnetin 3-O-β-glucoside7-O-α-rhamnoside based on the UHPLC-MS analysis was described in Skalski et al. 14 .

Preparation of stock solutions (fractions) and pure compounds for bioassays
The fractions were dissolved in 50% DMSO (a universal solvent for many different plant substances) and the pure compounds in 75% DMSO.The final concentration of DMSO in the tested human plasma was below 0.5%

Lipid peroxidation measurement
Lipid peroxidation was measured based on the concentration of thiobarbituric acid-reactive substances (TBARS) as described earlier by Sławińska et al. 8 .Plasma was incubated with fractions or compounds and an oxidative stress inducer (4.7 mM H 2 O 2 /3.8 mM Fe 2+ /2.5 mM EDTA) for 30 min at 37 °C (final concentrations: 1, 10, and 50 μg/mL).A negative control with 0.9% NaCl was set up.After the incubation, equal amounts of 15% TCA in 0.25 M HCl and 0.37% TBA in 0.25 M HCl were added to the samples; the samples were mixed and placed in a 100 °C thermoblock for 15 min.Next, the samples were cooled and centrifuged (10 000 × g, 15 min, 18 °C).The absorbance of the supernatant was measured in triplicate with a SPECTROstar Nano Microplate Reader (BMG LABTECH, Germany) at 535 nm.Equal amounts of 0.9% NaCl, 15% TCA, and 0.37% TBA were used as a blank sample.TBARS concentration was calculated with a molar extinction coefficient (ε = 156,000 M −1 cm −1 ).

Protein carbonylation measurement
Protein carbonylation was measured with a method employing 2,4-dinitrophenylhydrazine (DNPH) as described earlier by Sławińska et al. 8 .Plasma was incubated with the fractions or compounds and an oxidative stress inducer (4.7 mM H 2 O 2 /3.8 mM Fe 2+ /2.5 mM EDTA) for 30 min at 37 °C (final concentrations: 1, 10, and 50 μg/mL).After the incubation, the samples were diluted 10 times with 0.9% NaCl.Then, 40% TCA was added on ice.The samples were incubated for 5 min and centrifuged (5 min, 2500 rpm, 4 °C).Supernatant was discarded and 10 mM DNPH in 2 M HCl was added to the pellets.The samples were incubated for 1 h at room temperature, in the dark.40% TCA was added on ice, the samples were incubated for 5 min and centrifuged again (5 min, 2500 rpm, 4 °C).The supernatant was discarded.1.5 mL of 1:1 ethanol/ethyl acetate was added to the pellets on ice and the samples were vortexed for 5 min and centrifuged (5 min, 2500 rpm, 4 °C); this step was repeated two more times.Finally, supernatant was discarded, and the pellets were dissolved in 1 mL 6 M guanidine hydrochloride in 2 M HCl.The absorbance was measured in triplicate with a SPECTROstar Nano Microplate Reader (BMG LABTECH, Germany) at 280 and 375 nm.The levels of carbonyl groups were calculated with a molar extinction coefficient (ε = 22,000 M −1 cm −1 ) and expressed as nmol carbonyl groups/mg of plasma protein.

Thiol group oxidation measurement
The levels of thiol groups were measured with a method based on Ellman's reagent (5,5′-dithio-bis-(2-nitrobenzoic acid), DTNB) as described earlier by Sławińska et al. 8 .Plasma was incubated with fractions or compounds and an oxidative stress inducer (4.7 mM H 2 O 2 /3.8 mM Fe 2+ /2.5 mM EDTA) for 30 min at 37 °C (final concentrations: 1, 10, and 50 μg/mL).A negative control with 0.9% NaCl was set up.After the incubation 20 μL of the samples were added to a 96-well plate in triplicate.20 μL of 10% SDS in 10 mM phosphate buffer (pH 8) and 160 μL of 10 mM phosphate buffer (pH 8) were added to all samples.Absorbance was measured with a SPECTROstar Nano Microplate Reader (BMG LABTECH, Germany) at 412 nm and 280 nm.Afterward, 16.6 μL of 10 mM DTNB in 10 mM phosphate buffer (pH 8) was added to the samples (16.6 μL of 10 mM phosphate buffer (pH 8) was added to the blank sample instead of DTNB).The plate was covered with parafilm and incubated at 37 °C for 1 h.Afterward, the absorbance was read again.The levels of thiol groups were calculated with a molar extinction coefficient (ε = 13,600 M −1 cm −1 ) and expressed as nmol carbonyl groups/mg of plasma protein.

Measurement of prothrombin time
Plasma was incubated with the fractions for 30 min at 37 °C; the final concentrations were 10 and 50 μg/mL.A control sample with 0.9% NaCl was set up. 50 μL of the samples were added to coagulometer cuvettes in duplicate.The samples were incubated again for 2 min at 37 °C and 100 μL of Dia-PT reagent was added.The time of coagulation was recorded with a K-3002 Optic Coagulometer (Kselmed) as described earlier by Sławińska et al. 8 .

Measurement of thrombin time
Plasma was incubated with the fractions for 30 min at 37 °C; the final concentrations were 10 and 50 μg/mL.A control sample with 0.9% NaCl was set up. 50 μL of the samples were added to coagulometer cuvettes in duplicate.The samples were incubated for 1 min at 37 °C and 100 μL of thrombin (final concentration -5 U/mL) was added.The time of coagulation was recorded with a K-3002 Optic Coagulometer (Kselmed) as described earlier by Sławińska et al. 8 .

Measurement of activated partial thromboplastin time
Plasma was incubated with the fractions for 30 min at 37 °C; the final concentrations were 10 and 50 μg/mL.A control sample with 0.9% NaCl was set up. 50 μL of the samples were added to coagulometer cuvettes in duplicate.

Statistical analysis
Statistical analysis was carried out with Statistica 10 (StatSoft 13.3, TIBCO Software Inc. Palo Alto, CA, USA).The distribution of data was checked with Shapiro-Wilk test.The homogeneity of variance was checked with Levene's test.If the data had normal distribution and homogenous variance, one-way ANOVA with Tukey's post-hoc test was used to assess the differences between groups; otherwise, Kruskal-Wallis test was used.The results are presented as means ± SD or medians and interquartile ranges.The results were considered significant at p < 0.05.Dixon's Q-test was used to eliminate uncertain data.

Chemical characteristic of fraction a-g
Fraction g consisted of the most polar compounds, mainly dihexoses.It also contained small amounts of trihexoses, as well as putative dipeptide and serotonin.The composition of fractions c and f (eluted with 20% methanol) was similar, but they differed markedly in the content of some compounds.Serotonin seemed to be the dominant compound of both preparations.Serotonin hexoside was also detected.A sulfur-containing putative dipeptide and a putative indolylacetic acid hexoside were the other major constituents of fraction f.They were also present in fraction c but in significantly lower amounts (Fig. 1, Table 1).Moreover, both fractions contained putative nucleosides (uridine, guanosine, xanthosine), aromatic amino acids (phenylalanine, tryptophan), methylgallic acid, and diverse B-type proanthocyanidins (dimers, and trimers of (epi)catechin, and/or (epi)gallocatechin).Procyanidin peaks of fraction c were about two times higher than those of fraction f (Fig. 1).

Coagulation times (PT, TT, and APTT)
None of the fractions from raw and roasted seeds significantly impacted the coagulation times measured in plasma at both used concentrations (10 and 50 µg/mL) (data not presented).

A comparison of the antioxidative activity of various fractions
Effects of various tested fractions (at the highest used concentration-50 µg/mL) on oxidative stress in human plasma treated with H 2 O 2 /Fe 2+ were compared in Table 4. Out of all the tested fractions, fraction g (isolated from roasted seeds) demonstrated the strongest antioxidant properties.It reduced plasma protein carbonylation and oxidation of protein thiol groups.Moreover, two fractions (a and c) from raw seeds and two fractions (d and f) from roasted seeds had stronger antioxidant activity than other fractions (b and e).In addition, thermal processing increased the antioxidant activity of raw seeds-only fractions f and g from raw seeds inhibited thiol groups oxidation induced by H 2 O 2 /Fe 2+ .On the other hand, only fraction c from raw seeds significantly decreased lipid peroxidation.

Discussion
While there are many publications describing the composition of sea buckthorn seed oil, its carotenoids, tocochromanols, and sterols, the knowledge about other groups of natural products present in the seeds is lacking.Seed flavonoids have been characterized quite well, but the data concerning other groups of metabolites are scarce.
Our LC-MS analyses showed that serotonin was the dominant constituent of fractions c and f (Fig. 1, Table 1), it was tentatively identified on the basis of its HRMS and UV spectra.The presence of serotonin in the bark and other parts of sea buckthorn, including seeds, is quite well documented in the literature [15][16][17][18][19] .A putative sulfurcontaining dipeptide and a putative indolylacetic acid hexoside seemed to be other major constituents of fraction f, and were also present in fraction c.It cannot be excluded that the dipeptide is γ-glutamyl-S-methylcysteine.
The detected compound has the MS/MS spectrum similar to those reported by Joshi et al. 20 , and Chen et al. 21.This would be probably the first report about its presence in sea buckthorn.γ-glutamyl-S-methylcysteine was previously found in the seeds of certain legumes, which may increase the reliability of identification 20,[22][23][24][25] .It also seems to be probable that the detected putative derivative of indolylacetic acid is 7-hydroxy-2-oxindole-3-acetic acid 7'-O-3-β-glucoside, a product of auxin metabolism, which was isolated from maize seedlings; its UV spectrum is quite similar to that of the compound from fractions f and c (which has maxima at 255 and 296 nm) 26 .2).The presence of such flavonoid glycosides in sea buckthorn seeds is confirmed by publications of other research groups 12,28,29 .The fractions also contained many flavonol tri-and diglycosides acylated with (2E)-2,6-dimethyl-6-hydroxy-2,7-octadienolic acid ((E)-linalool-1-oic acid), phenolic acids (sinapic, ferulic), or both of them (Fig. 2, Table 2).It seems that such compounds are characteristic for sea buckthorn seeds, as many similar kaempferol glycosides were previously isolated from seeds of H. rhamnoides ssp.sinensis 28,30,31 .Apart from flavonol glycosides, the fractions also contained catechin, epicatechin, phenolic acid derivatives, sesquiterpenoid glycosides, vomifoliol hexoside-pentoside, alkaloids, triterpenoid saponins, as well as some unidentified constituents.Most of these compounds occurred in small amounts.The detected vomifoliol glycoside is probably vomifoliol 3'-O-β-D-apiofuranosyl-(1 → 6)-β-D-glucopyranoside, purified from seeds of H. rhamnoides ssp.sinensis by Gao et al. 15 .Regarding alkaloids, four compounds were detected.Two of them were preliminarily identified as β-carboline alkaloids, harmalol and harmol.Our data partly corroborate the results of Michel 33 , who used LC-HRMS to detect two putative β-carboline alkaloids in sea buckthorn fruit; one of them was tentatively identified as harmalol.Another alkaloid (Table 2, compound 18), a coumaric acid derivative found in fractions b and e, is most likely 4-[p-coumaroylamino]butan-1-ol, a compound earlier purified from the seeds of H. rhamnoides ssp.sinensis 34 .
Fractions a and d contained mainly triterpenoid saponins.Most of these compounds, including major constituents of the fractions, were acylated with (2E)-2,6-dimethyl-6-hydroxy-2,7-octadienolic acid, some of them were also acetylated (Fig. 3, Table 3).Small amounts of unacylated saponins were also present in fractions b and e.The same or similar compounds were found earlier in sea buckthorn leaves 35,36 .Six triterpenoid saponins were also previously isolated from H. rhamnoides ssp.sinensis seeds 32 , but they were different from those described in the current work.The fractions also contained free triterpenoids (Table 3), but in a smaller number than other parts of sea buckthorn 35,37 .Small amounts of diverse lysophospholipids were also detected, such as lysophosphatidylethanolamines, lysophosphatidylinositols, lysophosphatidylserines, lysophophatidylglycerols, lysophosphatidic acids.Moreover, both fractions contained glycerophospho N-acylethanolamines (GP-NAE, NA-GPE), similar to compounds reported from seeds of hemp or hazel 38,39 .
Oxidative stress is known as one of the major contributors in various diseases, including cardiovascular diseases and cancer.In general, it is not recommended to evaluate the antioxidant activity of plant preparations by only a single method due to the complex nature of phytocompounds.Therefore, in the present in vitro study, three assays (including thiol and carbonyl groups' determination, and lipid peroxidation) were used to study oxidative stress in human plasma.Thermal processing has various effects on the antioxidant properties of fruits, vegetables, and seeds.Heat treatment might decrease antioxidant activity, but in some cases, the effect can be opposite-antioxidant activity can improve.The outcome is largely dependent on the method of processing, temperature/power, and time [40][41][42][43][44] .In addition, the stability of phytocompounds during thermal processing depends on the type of compound and plant material in which the compound is encased during heating [45][46][47][48] .

Fig. 3. UHPLC-CAD chromatograms of fractions a (A) and d (B). Numbers above the chromatographic peaks correspond to numbers of compounds in
We reported for the first time that three fractions from raw sea buckthorn seeds and four fractions from roasted sea buckthorn seeds differed in terms of antioxidant properties, in an experimental model based on human plasma.The easiest way to explain these differences relates to the different chemical profiles of each of the used fractions.In addition, we suggest that thermal processing can increase the antioxidant activity of raw sea buckthorn seeds.
Fractions a (from raw seeds) and d (from roasted seeds) were very similar in composition and consisted mainly of various triterpenoid saponins.Both showed comparable antioxidant activity by decreasing protein carbonylation.Other researchers also reported that triterpenoid saponins possess antioxidant potential, which is in line with our results [49][50][51] .For instance, Liu et al. 50noted that triterpenoid saponins from the roots of Glycyrrhiza uralensis decreased Fe 2+ /cysteine-induced liver microsomal lipid peroxidation, while Yang et al. 52 reported the antioxidant potential of triterpenoids isolated from sea buckthorn branch bark.Our earlier results also demonstrated that a triterpenoid fraction from sea buckthorn twigs reduces lipid peroxidation and protein carbonylation in plasma treated with H 2 O 2 /Fe 2+53 .The small differences in the activity of fractions a and d might be caused by small changes in composition induced by thermal processing.
Only fractions f and g from roasted seeds significantly inhibited thiol groups oxidation.These observations are similar to other studies, where thermal processing improved the antioxidant activity of seeds.For example, roasting increased the antioxidant capacity of coffee beans 54 and peanuts (Arachis hypogaea L.) 55 .
Serotonin, a dominant constituent of fractions c and f showed the strongest antioxidant potential out of the three tested compounds by inhibiting both lipid peroxidation and protein carbonylation.Serotonin is an ancient indoleamine with important functions in both plants and animals.In the past, it was known mainly for its role as a neurotransmitter, but recently, other functions of serotonin, including its antioxidant activity, are being explored in depth 56 .Azouzi et al. 57 proposed that serotonin prevents lipid peroxidation by binding to lipid bilayers, preferentially to lipids with unsaturation on both alkyl chains.This would allow it to trap hydroxyl radicals before they have the chance to interact with lipids.Moreover, serotonin's hydroxyl group might contribute to its reactive oxygen species (ROS)-scavenging capacity.
Various glycosides of isorhamnetin, kaempferol, and quercetin (especially isorhamnetin 3-O-glucoside-7-Orhamnoside) were the main constituents of fractions b and e. Isorhamnetin 3-O-glucoside-7-O-rhamnoside showed some degree of antioxidant activity, but comparatively stronger activity of fractions b and f suggest that the combined action of other flavonol glycosides also contribute to their antioxidant properties.Skalski et al. 14 also noted that these compounds isolated from sea buckthorn berries have antioxidant potential in vitro.Different tested fractions, including b, c, e, and f contained B-type proanthocyanidins as well.B-type proanthocyanidins have a proven antioxidant potential which might enhance the antioxidant properties of the tested fractions 58,59 .
In addition, our present results may suggest that tested fractions are often more effective antioxidant factors than pure chemical compounds.These results may also indicate that various compounds have synergistic actions.d, e, f, and g) from roasted (B) sea buckthorn seeds on the level of thiobarbituric acid reactive substances (TBARS) in human plasma treated with an oxidative stress inducer (Fe 2+ /H 2 O 2 ) (n = 9).Control (+) and test samples (0.5-50 μg/mL) were incubated (30 min, 37 °C) with Fe 2+ /H 2 O 2 .The results are presented as % of control (+).The data are expressed as means ± SD.One way-ANOVA with Tukey's test was used to assess the differences between the control and the test samples.The results were considered significant at p < 0.05 (***p < 0.001).The numbers above significant results are % of lipid peroxidation inhibition.The difference between control (−) and control (+) was statistically significant (p < 0.001).
Oxidative stress can influence hemostasis, including the coagulation process.Moreover, it is known that thermal processing can change not only the antioxidant activity of seeds but also other biological properties; for example, the anti-inflammatory potential of coffee beans was decreased by high-intensity roasting 60 .However, in our previous study on the extracts from raw and roasted sea buckthorn seeds, both used extracts (0.5-50 µg/ mL) had no effect on plasma coagulation times 8 .In the present study, the seven used fractions (a-g) from the seeds also did not change the coagulation activity of plasma in vitro.and c) from raw (A) and four fractions (d, e, f, and g) from roasted (B) sea buckthorn seeds on the level of carbonyl (CO) groups in human plasma treated with an oxidative stress inducer (Fe 2+ /H 2 O 2 ) (n = 8).Control (+) and test samples (0.5-50 μg/mL) were incubated (30 min, 37 °C) with Fe 2+ /H 2 O 2 .The results are presented as % of control (+).The data are expressed as medians and interquartile ranges.Kruskal-Wallis test was used to assess the differences between the control and the test samples.The results were considered significant at p < 0.05 (*p < 0.05; **p < 0.01; *** p < 0.001).The numbers above significant results are % of protein carbonylation inhibition.The difference between control (−) and control (+) was statistically significant (p < 0.001).
Despite the promising in vitro antioxidant activity of fractions from sea buckthorn seeds, poor bioavailability of most phenolic compounds might be an obstacle in achieving the same results in vivo.Suomela et al. 61 reported that administration of 0.4 g of sea buckthorn flavonol extract (containing 78 mg of flavonol aglycones) daily for 4 weeks failed to reduce CVDs risk markers and oxidative stress in healthy subjects, which suggests that a higher dosage might be needed to induce significant effects 61 .The bioavailability of most triterpenoid saponins is relatively low, however, it can sometimes be improved by microbial biotransformation 62 .Thermal processing can be a way of improving the bioaccessibility and bioavailability of various compounds by detaching them from protein complexes or releasing them from the food matrix 63,64 .For example, phenolic compounds from highland barley had higher bioaccessibility after thermal processing 64 .Fig. 6.The effect of three fractions (a, b, and c) from raw (A) and four fractions (d, e, f, and g) from roasted (B) sea buckthorn seeds on the level of thiol groups in human plasma treated with an oxidative stress inducer (Fe 2+ /H 2 O 2 ) (n = 9).Control (+) and test samples (0.5-50 μg/mL) were incubated (30 min, 37 °C) with Fe 2+ / H 2 O 2 .The results are presented as % of control (+).The data are expressed as medians and interquartile ranges.Kruskal-Wallis test was used to assess the differences between the control and the test samples.The results were considered significant at p < 0.05 (*p < 0.05, **p < 0.01).The numbers above significant results are % of thiol groups oxidation inhibition.The difference between control (−) and control (+) was statistically significant (p < 0.001).The phenolic content of sea buckthorn berries was changed after enzymatic digestion; their antioxidant properties were significantly enhanced by this process 65 .Results of Baeza et al. 66 also found that various metabolites of phenolic compounds have stronger biological activity than their precursors.
Our previous studies 8 together with the present experiments indicate that sea buckthorn seeds are a rich source of antioxidants; the observed properties can be linked not only with the presence of phenolic compounds but also with triterpenoids.In addition, we suggest that not only raw but also roasted sea buckthorn seeds can be a new source of natural antioxidants (as functional food or supplements), beneficial in the prevention and treatment of various diseases associated with oxidative stress.However, more research is needed to provide a better understanding of the antioxidative mechanism of bioactive compounds of sea buckthorn seeds after thermal processing and determine their effectiveness in vivo.

Fractions c and
f, as well as b and e, also contained different B-type proanthocyanidins.The detected compounds were similar to those found in seeds of H. rhamnoides ssp.sinensis by Fan et al.27 .Fractions b and e consisted mainly of numerous glycosides of kaempferol, quercetin, and isorhamnetin.3-O-dihexoside-7-O-deohyhexosides of these aglycones (most probably 3-O-sophoroside-7-O-rhamnosides), as well as isorhamnetin 3-O-β-glucoside-7-O-α-rhamnoside were dominant constituents of both preparations (Fig.2, Table

Fig. 4 .
Fig. 4. The effect of three fractions (a, b,and c) from raw (A) and four fractions (d, e, f, and g) from roasted (B) sea buckthorn seeds on the level of thiobarbituric acid reactive substances (TBARS) in human plasma treated with an oxidative stress inducer (Fe 2+ /H 2 O 2 ) (n = 9).Control (+) and test samples (0.5-50 μg/mL) were incubated (30 min, 37 °C) with Fe 2+ /H 2 O 2 .The results are presented as % of control (+).The data are expressed as means ± SD.One way-ANOVA with Tukey's test was used to assess the differences between the control and the test samples.The results were considered significant at p < 0.05 (***p < 0.001).The numbers above significant results are % of lipid peroxidation inhibition.The difference between control (−) and control (+) was statistically significant (p < 0.001).

Fig. 5 .
Fig. 5.The effect of three fractions (a, b,and c) from raw (A) and four fractions (d, e, f, and g) from roasted (B) sea buckthorn seeds on the level of carbonyl (CO) groups in human plasma treated with an oxidative stress inducer (Fe 2+ /H 2 O 2 ) (n = 8).Control (+) and test samples (0.5-50 μg/mL) were incubated (30 min, 37 °C) with Fe 2+ /H 2 O 2 .The results are presented as % of control (+).The data are expressed as medians and interquartile ranges.Kruskal-Wallis test was used to assess the differences between the control and the test samples.The results were considered significant at p < 0.05 (*p < 0.05; **p < 0.01; *** p < 0.001).The numbers above significant results are % of protein carbonylation inhibition.The difference between control (−) and control (+) was statistically significant (p < 0.001).

Table 3 .
Tentatively identified constituents of fractions a and d.FA formic acid; Hex hexose; dHex deoxyhexose; Pen pentose; AcA acetic acid; CouA coumaric acid; LinA (E)-linalool-1-oic acid; PE phosphoethanolamine; PG phosphoglycerol; PI phosphoinositol; PS phosphoserine; P phosphate; italic -data from the LC-MS analysis, in which the mobile phase contained ammonium formate buffer, as described in the Materials and methods.