Flavonoid Preparations from Taraxacum officinale L. Fruits—A Phytochemical, Antioxidant and Hemostasis Studies

Dandelion (Taraxacum officinale L.) roots, leaves, and flowers have a long history of use in traditional medicine. Compared to the above organs, dandelion fruits are the least known and used. Hence, the present paper was aimed at the phytochemical analysis of T. officinale fruit extract and estimating its antiradical, antiplatelet, and antioxidant properties related to hemostasis. Methanolic extract of fruits (E1), enriched with polyphenols (188 mg gallic acid equivalents (GAE)/g), was successfully separated into cinnamic acids (E2; 448 mg GAE/g) and flavonoids (E3; 377 mg GAE/g) extracts. Flavonoid extract was further divided into four fractions characterized by individual content: A (luteolin fraction; 880 mg GAE/g), B (philonotisflavone fraction; 516 mg GAE/g), C (flavonolignans fraction; 384 mg GAE/g), and D (flavone aglycones fraction; 632 mg GAE/g). High DPPH radical scavenging activity was evaluated for fractions A and B (A > B > Trolox), medium for extracts (Trolox > E3 > E2 > E1), and low for fractions C and D. No simple correlation between polyphenol content and antiradical activity was observed, indicating a significant influence of qualitative factor, including higher anti-oxidative effect of flavonoids with B-ring catechol system compared to hydroxycinnamic acids. No cytotoxic effect on platelets was observed for any dandelion preparation tested. In experiments on plasma and platelets, using several different parameters (lipid peroxidation, protein carbonylation, oxidation of thiols, and platelet adhesion), the highest antioxidant and antiplatelet potential was demonstrated by three fruit preparations–hydroxycinnamic acids extract (E2), flavonoid extract (E3), and luteolin fraction (A). The results of this paper provide new information on dandelion metabolites, as well as their biological potential and possible use concerning cardiovascular diseases.


Introduction
The occurrence of some civilization diseases, such as atherosclerosis, diabetes, heart attack, and some cancers, has been associated with elevated levels of oxidative stress and dysfunctional hemostasis. Hemostasis is defined as a sequence of reactions that maintain the fluidity of circulating

DPPH Free Radical Scavenging Activity
The DPPH radical scavenging activity of dandelion fruits phenolic preparations (E1-E3, Fr A-D) was expressed as Trolox Equivalents (TE) and IC 50 values ( Table 3). The estimated TE values varied from 0.05 to 2.01 of Trolox activity, with 1.00 indicating equivalence to Trolox. Fraction A (luteolin) and B (philonotisflavone) demonstrated the highest scavenging activity (2.01 and 1.09 TE, respectively), followed by the three extracts E1-E3 (0.26-0.55 TE), and the fractions D and C (0.05 and 0.06 TE, respectively).
tri-hexose (raffinose)  most widely-studied dandelion aerial organs. Besides, several metabolites not yet reported in T. officinale, such as biflavones and some flavonolignans, were also detected ( Table 1).

Biomarkers of Oxidative Stress in Plasma and Blood Platelets
Three extracts E1-E3 and two fractions A and D inhibited plasma lipid peroxidation stimulated by H 2 O 2 /Fe at the highest tested concentration (50 µg/mL); however, fractions B and C remained inactive ( Figure 2). In addition, extracts E2 and E3 and fractions B, C, and D reduced protein thiol group oxidation in plasma treated with H 2 O 2 /Fe ( Figure 3); extract E3 reduced this process at 10 and 50 µg/mL ( Figure 3).  All tested dandelion fruit preparations inhibited protein carbonylation in plasma treated with H2O2/Fe at the highest tested concentration (50 µg/mL) ( Figure 4). In addition, two extracts (E1 and E3) and two fractions (C and D) were also active at the lower tested concentration (10 µg/mL).  All tested dandelion fruit preparations inhibited protein carbonylation in plasma treated with H2O2/Fe at the highest tested concentration (50 µg/mL) ( Figure 4). In addition, two extracts (E1 and E3) and two fractions (C and D) were also active at the lower tested concentration (10 µg/mL). All tested dandelion fruit preparations inhibited protein carbonylation in plasma treated with H 2 O 2 /Fe at the highest tested concentration (50 µg/mL) ( Figure 4). In addition, two extracts (E1 and E3) and two fractions (C and D) were also active at the lower tested concentration (10 µg/mL). None of the tested preparations changed the TBARS level in resting platelets at either 10 or 50 µg/mL ( Figure 5A). However, extract E3 and fractions A-D significantly inhibited thrombin-induced platelet lipid peroxidation when administered at 50 µg/mL. For example, 10 and 50 µg/mL luteolin fraction inhibited this process by about 60% compared with positive controls ( Figure 5B). In addition, three flavone fractions (A, B, and D) significantly reduced lipid peroxidation induced by H2O2/Fe when administered at 10 and 50 µg/mL; however, two of the extracts (E1 and E2) and fraction C (10 µg/mL) did not appear to exert any activity ( Figure 5C). H2O2/Fe-induced oxidation of protein thiol functions was ameliorated by preparations E1 and B at 10 µg/mL, and by E3 and A at 50 µg/mL ( Figure 6). Moreover, H2O2/Fe-induced protein carbonylation was inhibited by extracts E1, E2 and fraction A ( Figure 7); for example, phenolic acid extract (E2) demonstrated a protective effect at two used doses (10 and 50 µg/mL) (Figure 7). None of the tested preparations changed the TBARS level in resting platelets at either 10 or 50 µg/mL ( Figure 5A). However, extract E3 and fractions A-D significantly inhibited thrombin-induced platelet lipid peroxidation when administered at 50 µg/mL. For example, 10 and 50 µg/mL luteolin fraction inhibited this process by about 60% compared with positive controls ( Figure 5B). In addition, three flavone fractions (A, B, and D) significantly reduced lipid peroxidation induced by H 2 O 2 /Fe when administered at 10 and 50 µg/mL; however, two of the extracts (E1 and E2) and fraction C (10 µg/mL) did not appear to exert any activity ( Figure 5C). H 2 O 2 /Fe-induced oxidation of protein thiol functions was ameliorated by preparations E1 and B at 10 µg/mL, and by E3 and A at 50 µg/mL ( Figure 6). Moreover, H 2 O 2 /Fe-induced protein carbonylation was inhibited by extracts E1, E2 and fraction A (   Effects of extracts (E1-E3) and flavonoid fractions (A-D) of dandelion fruits (concentrations 10 and 50 µg/mL) on lipid peroxidation in resting blood platelets (A), platelets activated by 5 U/mL thrombin (B), and platelets treated with H2O2/Fe (C). The data are presented as means ± SD (n = 5). Control negative refers to platelets not treated with thrombin or H2O2/Fe, and control positive to platelets treated with thrombin or H2O2/Fe. Kruskal-Wallis test: n.s. p > 0.05, * p < 0.05, ** p < 0.01, *** p < 0.001.

Hemostatic Parameters of Blood Platelets and Plasma
Adhesion of thrombin-activated platelets to fibrinogen was significantly inhibited after preincubation with five tested dandelion fruit preparations (E2 and E3, fractions A, B, and C) ( Figure  8A). Fraction A was highly active at both concentrations tested-10 and 50 µg/mL, while the others (E2, E3, B, and C) at the higher concentration ( Figure 8A). In the case of ADP-activated blood platelets, all four flavonoid fractions demonstrated inhibitory action, fractions A-C at both tested doses (10 and 50 µg/mL), fraction D only at 10 µg/mL ( Figure 8B).

Hemostatic Parameters of Blood Platelets and Plasma
Adhesion of thrombin-activated platelets to fibrinogen was significantly inhibited after preincubation with five tested dandelion fruit preparations (E2 and E3, fractions A, B, and C) ( Figure 8A). Fraction A was highly active at both concentrations tested-10 and 50 µg/mL, while the Molecules 2020, 25, 5402 13 of 33 others (E2, E3, B, and C) at the higher concentration ( Figure 8A). In the case of ADP-activated blood platelets, all four flavonoid fractions demonstrated inhibitory action, fractions A-C at both tested doses (10 and 50 µg/mL), fraction D only at 10 µg/mL ( Figure 8B). are presented as means ± SD (n = 5). Control negative refers to platelets not treated with H2O2/Fe, and control positive to platelets treated with H2O2/Fe. Kruskal-Wallis test: n.s. p > 0.05, * p < 0.05, ** p < 0.01, *** p < 0.001, compared with positive control.

Hemostatic Parameters of Blood Platelets and Plasma
Adhesion of thrombin-activated platelets to fibrinogen was significantly inhibited after preincubation with five tested dandelion fruit preparations (E2 and E3, fractions A, B, and C) ( Figure  8A). Fraction A was highly active at both concentrations tested-10 and 50 µg/mL, while the others (E2, E3, B, and C) at the higher concentration ( Figure 8A). In the case of ADP-activated blood platelets, all four flavonoid fractions demonstrated inhibitory action, fractions A-C at both tested doses (10 and 50 µg/mL), fraction D only at 10 µg/mL ( Figure 8B). Three dandelion fruit preparations with the highest antioxidant and antiplatelet potential (extracts E2 and E3, and fraction A) were selected for subsequent flow cytometry analyses. The samples treated with extract E2 or fraction A demonstrated different blood platelet activation states compared with the untreated control samples (Figures 9 and 10). For both above preparations, the effect was only evident at the higher test concentration (50 µg/mL), for example, E2 significantly reduced PAC-1 binding in platelets activated by 20 µM ADP or 10 µg/mL collagen ( Figure 9C,D); it also significantly reduced CD62P expression on platelets activated by 10 µg/mL collagen ( Figure  10D). In turn, extract E3 did not influence platelet P-selectin expression or GPIIb/IIIa expression for any of the four used models (Figures 9-11). Three dandelion fruit preparations with the highest antioxidant and antiplatelet potential (extracts E2 and E3, and fraction A) were selected for subsequent flow cytometry analyses. The samples treated with extract E2 or fraction A demonstrated different blood platelet activation states compared with the untreated control samples (Figures 9 and 10). For both above preparations, the effect was only evident at the higher test concentration (50 µg/mL), for example, E2 significantly reduced PAC-1 binding in platelets activated by 20 µM ADP or 10 µg/mL collagen ( Figure 9C,D); it also significantly reduced CD62P expression on platelets activated by 10 µg/mL collagen ( Figure 10D). In turn, extract E3 did not influence platelet P-selectin expression or GPIIb/IIIa expression for any of the four used models (Figures 9-11).     None of the dandelion fruit preparations changed the APTT, PT, TT, or AUC10 values measured by T-TAS for either human plasma or whole blood ( Figure 12, Table 4).

Sample Name
Tested Concentration (µg/mL)  None of the dandelion fruit preparations changed the APTT, PT, TT, or AUC 10 values measured by T-TAS for either human plasma or whole blood ( Figure 12, Table 4). None of the dandelion fruit preparations changed the APTT, PT, TT, or AUC10 values measured by T-TAS for either human plasma or whole blood ( Figure 12, Table 4).  15.5 ± 1.6 n.s. 14.9 ± 0.5 n.s. 43.5 ± 4.8 n.s.

Cytotoxicity against Blood Platelets
To examine the toxicity of all dandelion fruit preparations on platelets, the extracellular LDH activity was measured. Compared to a control group, there was no significant difference in the viability of platelets after exposure to dandelion extracts (E1-E3) and fractions (A-D) at 10 and 50 µg/mL ( Figure 13).

Cytotoxicity against Blood Platelets
To examine the toxicity of all dandelion fruit preparations on platelets, the extracellular LDH activity was measured. Compared to a control group, there was no significant difference in the viability of platelets after exposure to dandelion extracts (E1-E3) and fractions (A-D) at 10 and 50 µg/mL ( Figure 13).  Table 5 compares the effects of three extracts (E1-E3) and four fractions (A-D) from dandelion fruits on the biological properties of plasma and blood platelets. It can be seen that extract E2 and fraction A demonstrated stronger antioxidant and antiplatelet properties than the other tested preparations (Table 5).  Table 5 compares the effects of three extracts (E1-E3) and four fractions (A-D) from dandelion fruits on the biological properties of plasma and blood platelets. It can be seen that extract E2 and fraction A demonstrated stronger antioxidant and antiplatelet properties than the other tested preparations (Table 5).

Discussion
The fruits of Taraxacum officinale are not as widely used in phytopharmacology as its other organs, and little is hence understood of their chemical composition and biological properties. In the continuation of our study on the dandelion, the current paper was aimed at the phytochemical analysis of T. officinale fruit extract and estimating its antiradical, antiplatelet, and antioxidant properties related to hemostasis. Efforts were also made to estimate the contribution of individual polyphenols on the reported biological activity of dandelion fruits. The idea behind the separation/fractionation process of methanolic extract of fruits (E1) was, on the one hand, to facilitate and broaden the metabolite identification (preliminary LC-MS analysis showed the presence of some unreported compounds in dandelion), and on the other hand, to test all extracts and fractions for antioxidant and antiplatelet activity, and conclude about the input of individual compounds/groups of compounds into biological potential and exerted effect of dandelion fruits. Previous research on the T. officinale plant identified high numbers of hydroxycinnamic acid derivatives [16][17][18]; therefore, the present study focused more on fractions enriched in flavonoid compounds, which are also an important component of the aerial parts, i.e., the leaves, flowers, and fruits of the dandelion.
With the use of HR-QTOF-MS analysis, about 30 secondary metabolites were identified in T. officinale fruit extract. The phytochemical profile was dominated by flavonoids and hydroxycinnamic acid derivatives ( Table 1). Most of the detected polyphenols, such as caffeic acid esters and flavones, have already been described in dandelion species [13,19,20]; however, several flavone derivatives previously undescribed in the genus Taraxacum and the Asteraceae family were also identified. These included a series of flavone-flavone dimers (biflavones), particularly philonotisflavone (2',8-biluteolin), as well as luteolin 3'-O-glucoside ( Figure S1).
The C-C biflavones are a diverse group of phenolic compounds characterized by a variety of flavone units and linkage sites and have been found in many fruits, vegetables, and plants [21]. However, philonotisflavone has previously been known only from the gametophyte of a few bryophyte species, such as Philonotis Fontana [22] and Bartramia pomiformis [23]. Moreover, biflavonoids have so far been found in only one member of the Asteraceae family: Saussurea eopygmaea Hand.-Mazz. (Carduoideae subfamily) [24]. Several studies on biflavonoids, such as amentoflavone and ginkgetin, as well as on plant extracts enriched with these compounds, such as Gingko biloba leaf extract, have found them to demonstrate a wide range of cytotoxic, antimicrobial, antiviral, antioxidative, and anti-inflammatory activities [25].
To date, several luteolin glycosides have been identified in the Taraxacum genus, including two different isomers of luteolin glucoside (7-O-and 4'-O-glc) [13]; however, the present findings also indicate the presence of a new compound: luteolin 3'-O-glucoside. In addition, several flavonolignans (tricin-lignan conjugates) were observed, such as calquiquelignan D/E and salcolin A/B, which were recently described by Choi et al [26]. in an extract from the whole T. officinale plant.
Studies on the biological activity of dandelion fruit preparations began with the radical scavenging DPPH test. Five preparations (E1-E3, and fractions A,B) demonstrated significant antiradical activities (IC 50 0.06-0.42 mg/mL), while fractions C-D displayed relatively weak activity (IC 50 > 1.3 mg/mL). No simple correlation between polyphenol content (TPC) and antiradical activity (TE and IC 50 values) was observed, indicating a significant influence of qualitative factor. The DPPH test demonstrated the higher free radical scavenging effect of dandelion fruit flavonoids (E3 extract) compared with phenolic acids (E2 extract), even though E3 had significantly lower TPC than E2 (Tables 2 and 3). Similarly, luteolin, the main component of the flavonoid extract (E3), exhibited 2-fold higher TE (2.0 TE, Fraction A, Table 3) than chicoric acid (1.1 TE) [16], the main component of phenolic acid extract. Moreover, the results show that the antiradical potential of the fruit extract is very little influenced by flavonoids lacking an active B-ring catechol moiety, as exemplified by the C-D fractions. The tricin-lignan conjugates and flavone aglycones (chrysoeriol, tricin, and apigenin), the main components of fractions C and D, lack or have modified (methylated or substituted) the catechol system in the B ring of the aglycone, which is known to enhance the antiradical property of phenolics [27].
The anti-oxidative, antiplatelet, and hemostasis related effects of dandelion fruit preparations were investigated using several experiments in vitro. Hemostasis is based on a series of strictly-regulated processes that enable platelet activation, vascular repair, and blood clotting. The ability of blood to clot prevents its excessive loss when the skin or blood vessels are broken. Although a key role in the clotting process is played by fibrinogen, which acts to transform soluble fibrinogen into insoluble fibrin, the process requires a cascade of consecutive biochemical reactions to produce a fibrin clot: each of the factors involved is activated by a previously-activated factor [28]. However, three discrete phases can be distinguished in the formation of a platelet plug: adhesion, activation, and aggregation of platelets. The adhesion phase starts when a blood vessel is damaged, and during this phase, platelets begin to adhere to the exposed elements of the subendothelial layer. This stage is followed by platelet activation, which is stimulated by contact with an agonist such as ADP, collagen, or thrombin. During this stage, the platelet cytoskeleton undergoes reorganization and the platelets become less regular in shape, thus facilitating easier adhesion of successive layers. After activation by thrombin and then together with diacylglycerol lipase, phospholipase A 2, or phospholipase C, the platelets release arachidonic acid (AA) from their cell membrane phospholipids. AA is converted to cyclic prostaglandin peroxides (PGG 2 and PGH 2 ) by cyclooxygenase (COX), and then to thromboxane A 2 (TXA 2 ) by thromboxane A 2 synthase. The prostaglandin peroxides form a range of substances including malondialdehyde (MDA), taking place via a non-enzymatic pathway, whereas pro-aggregation PGF 2a and PGE 2 prostaglandins are formed by S-glutathione transfer. The PGG 2 and PGH 2 peroxides are converted to antiaggregatory PGD 2 prostaglandins by PGD 2 isomerase. The AA metabolic pathway leads to the production of mainly hydroxy acids and hepoxylins, a process catalyzed by 12-lipoxygenase. AA is converted into epoxyicatrienic acids by epoxygenase, interacting with cytochrome P450, and the AA in platelet blood may be converted to isoprostanes via a non-enzymatic route [28]. This phase is followed by aggregation, in which successive platelets adhere to a single layer of cells, resulting in the eventual formation of a platelet plug.
Physiological hemostasis is maintained by an equilibrium between the efficiency of the blood vessel wall, platelet, coagulation, and fibrinolysis system and the inhibitors regulating them. Any imbalance leads to excessive bleeding or clotting [29,30]. To prevent the formation of clots and blockages, antiplatelet drugs are typically introduced. Antiplatelet, otherwise known as anti-aggregation, drugs inhibit specific stages of platelet activation, i.e., adhesion or aggregation. Administration of rhodium, for example, may affect the synthesis of thromboxane A 2 , and influence the cyclic concentration of AMP (cAMP) or the total conversion of AA. Other antiplatelet drugs may block platelet membrane receptors such as acetylsalicylic acid or clopidogrel [31]; however, their use can cause various side effects.
Our previous publications demonstrated that in vitro tests using plasma and platelets, well reflecting the conditions in a living organism, are very helpful in finding substances with antioxidant and antiplatelet properties. In the current study, dandelion fruit preparations (E1-E3, A-D) showed various effects on oxidative stress in human plasma and blood platelets, as well as hemostatic properties of blood platelets.
The antiplatelet potential of the tested preparations, evaluated as an inhibitory effect on the adhesion of washed blood platelets to fibrinogen, was determined colorimetrically. Fractions A, B, and C demonstrated the greatest anti-adhesive activity in the two used models: (1) adhesion of thrombin-activated platelets to fibrinogen and (2) adhesion of ADP-activated platelets to fibrinogen. In addition, changes in hemostasis, such as the degree of platelet activation, occurring following exposure to the tested dandelion fruit preparation were identified in whole blood by T-TAS. T-TAS is a microchip-based flow chamber system that evaluates thrombogenicity in whole blood and may be used to assess the influence of anti-thrombotic preparations on blood platelet activation and coagulation reactions against a collagen or collagen/tissue thromboplastin-coated surface [32]. All selected dandelion fruit preparations (E2, E3, and fraction A) were found to demonstrate anti-coagulant potential against the platelets bound to collagen; however, this activity was not statistically different from untreated blood samples.
Platelet activation was also assessed using flow cytometry analysis of P-selectin expression (CD62P) and activation of GPIIb/IIIa complex (PAC-1 binding) in whole blood samples containing unstimulated platelets, and platelets stimulated by ADP or collagen. Flow cytometry can be used for both diagnoses and basic research, and in this case, it enabled the biological activity of blood platelets to be measured in the natural environment, i.e., immediately after blood collection. GPIIb/IIIa receptors are a good marker of blood platelet activation as the increasing number and change of their conformation during activation, aggregation, and adhesion [33]. Our findings confirm that the surface expression of the active form of GPIIb/IIIa on blood platelets decreases in the presence of extract E2 and fraction A (luteolin); also indicate significantly lower blood platelet adhesion in the presence of the dandelion fruit preparations, especially fraction A against washed blood platelets: we propose, therefore, that inhibition of platelet adhesion to fibrinogen may be associated with low expression of GPIIb/IIIa.
Another marker of blood platelet activation is P-selectin: a glycoprotein present in the alpha granules of unstimulated platelets. During activation, P-selectin is released with the alpha granules and can be seen on the surface of the platelet. Present findings indicate that P-selectin expression was reduced in whole blood following treatment with luteolin fraction (A) and hydroxycinnamic acids extract (E2). Finally, it appeared that luteolin demonstrated the best anti-platelet properties of the two preparations, which was observed not only in washed blood platelets but also in whole blood. Our observations are consistent with the literature: Benavente-Garcia and Castillo [34] reported that luteolin has antithrombotic activity, while Guerrero et al [35]. and Dell'Agli et al [36]. found it to have anti-platelet potential.
Platelet activation is associated with arachidonic acid metabolism. Our study demonstrates that all fruit flavonoid enriched preparations (E3, and A-D) reduced enzymatic lipid peroxidation in thrombin-activated platelets in vitro, as indicated by TBARS measurements. This finding suggests that dandelion fruit flavonoids can modulate blood platelet activation by interfering with the metabolism of arachidonic acid. No such change was observed for the other tested extracts, viz. E1 and E2, even at the highest used concentration, due to different chemical profiles-lack or low level of flavonoids.
In the anti-oxidative experiments, blood platelets and plasma were exposed to a hydroxyl radical donor (H 2 O 2 /Fe 2+ ) and protective effects of dandelion preparations were measured. Four models were used to evaluate this activity: (1) plasma lipid peroxidation, (2) plasma protein carbonylation, (3) platelet lipid peroxidation, and (4) protein thiol oxidation or protein carbonylation in platelets treated with H 2 O 2 /Fe 2+ . It was found that extracts E1-E3 and fraction A (luteolin) demonstrated the best antioxidant properties. The results were therefore largely, but not entirely (an example of fraction B), consistent with the antiradical test with DPPH (E2-E3 extracts and luteolin fraction). In addition, a strong protective effect against the oxidation of plasma lipids and proteins exerted by hydroxycinnamic acids extract (E2), dominated by L-chicoric acid, is consistent with our previous publications [16,17].
The compounds extracted from the dandelion fruits appeared to be safe for use in the blood model, as none of the tested preparations (concentrations 10 and 50 µg/mL) demonstrated the cytotoxicity against platelets, measured as extracellular LDH activity. It is important to note that the concentrations of the tested extracts and fractions correspond with the physiological concentrations of phenolic compounds, including flavonoids, available after oral supplementation [37,38].
Flavonoids are metabolized by a two-phase process: phase I, based on hydroxylation and demethylation by cytochrome P 450 takes place in the liver, and phase II, based on O-methylation and coupling with glucuronic or sulfuric acid, in the intestine. These products are excreted with urine and bile; but must first pass through the intestinal-hepatic circulation, thus prolonging their elimination time and possible activity. The non-absorbent and bile-derived flavonoid metabolites are processed by the intestinal microflora, mainly in the large intestine. Bacterial enzymes can catalyze reactions such as the hydrolysis of glucuronides, sulfates, and glycosides, as well as dehydroxylation, demethylation, double bond reduction, and decomposition of the C-ring with the formation of phenolic acids, followed by their decarboxylation. Phenolic acids can also be absorbed, conjugated, or O-methylated in the liver and then excreted with the urine. Absorption may be relevant to total plasma antioxidant activity, as catecholytic acids show free radical scavenging activity [39]. Interestingly, various metabolites of phenolic compounds have been found as stronger inhibitors of platelet activation than their precursors [40].
In conclusion, the present study provides new information on Taraxacum metabolites, such as biflavonoid-philonotisflavone, and luteolin 3 -O-glucoside. From biological studies, it can be concluded that both the antioxidant and antiplatelet potential of dandelion fruit, demonstrated in several different in vitro experiments, was mainly due to luteolin (the main component of fraction A) and chicoric acid (the main component of E2 extract). We suggest that inhibition of platelet activation by fraction A may be associated with inhibition of receptor's expression (including GPIIb/IIIa) and inhibition of arachidonic acid metabolism. By inhibiting platelet activation and oxidative stress, dandelion fruits and their polyphenols can be recognized as novel and valuable phytopharmacological agents. However, the mechanism of their antiplatelet and antioxidant properties remains unclear and required further studies. Additional studies, including in vivo experiments, are also needed to investigate the overall antioxidant and antiplatelet effects of dandelion fruits.

Plant Material
Dandelion fruits were harvested in May 2017 on a farm located in south-eastern Poland (50 •

Preparation of Phenolic Extracts and Fractions from Dandelion Fruits
The finely-ground dandelion fruits (700 g) were defatted by extraction with n-hexane (4 L) under reflux (eight hours). The obtained defatted material (565 g) was twice extracted with 80% methanol (v/v; 4 L × 2; 12 h × 2) at 30 • C and sonicated (12 × 10 min) to enhance the extraction efficiency. The methanol extracts were filtered and combined, obtaining extract E1, which was then concentrated with a vacuum rotary evaporator (40 • C). The aqueous suspension of E1 (~50 g) was subjected to liquid-liquid extraction with ethyl acetate (0.75 L × 3). After evaporation of the organic solvent, flavonoid extract (E3) was freeze-dried to give 8.1 g of E3. The aqueous residue (~40 g) was found to contain large amounts of water-soluble primary metabolites (carbohydrates and amino acids) in addition to phenolics, as indicated by liquid chromatography-mass spectrometry (LC-MS) analysis. It was further purified by solid-phase extraction (SPE) on a short C18 column (12 × 5 cm, Cosmosil C18-PREP, 140µm, Nacalai Tesque Inc., Kyoto, Japan). The ballast components were washed with 4% methanol (v/v), and the compounds of interest were eluted with 80% methanol (v/v). After evaporation of the organic solvent, phenolic acid extract (E2) was lyophilized to give 9.5 g of E2.
The flavonoid extract (E3) was further fractionated isocratically on a Sephadex LH-20 (Sigma-Aldrich) column (80 × 2.8 cm i.d.) using 95% methanol (v/v) at a flow rate of 2.4 mL/min. The single sample capacity was 800 mg (dissolved in 10 mL of 95% methanol). The separation was monitored by LC-MS analyses. The five pooled fractions (I-V) were collected, concentrated, and freeze-dried to give 1290 mg of fraction I, 115 mg of fraction II, 850 mg of fraction III, and 480 mg of fraction IV. Due to their complex composition, or the presence of impurities, as indicated by LC-MS analysis (data not shown), the four fractions (I-IV) were further purified by high-performance liquid chromatography (HPLC) (Dionex, Sunnyvale, CA, USA) on a system equipped with a photodiode array detector (PDA-100) and FC 204 fraction collector (Gilson, Middleton, WI, USA). Separations were carried out on an Atlantis T3 C18 semi-preparative column (250 × 19 mm, 5 µm, Waters) at 40 • C with aqueous acetonitrile, containing 0.1% formic acid, at a flow rate of 5.5 mL/min as mobile phase. The conditions of the chromatographic run were individually optimized for each fraction (isocratic or gradient mode between 15-55% acetonitrile). The PDA detector was operated at 210 and 345 nm (5 nm bandwidth). The separation was monitored by LC-MS analyses. In total, four fractions (A-D) were collected, concentrated at 40 • C, and freeze-dried: the total yields were 1150 mg for fraction A, 95 mg for fraction B, 275 mg for fraction C, and 375 mg for fraction D. The extraction and fractionation process are illustrated in Figure S1.

Qualitative High-Resolution LC-MS Analysis
The dandelion fruit extracts (E1-E3) and fractions (A-D) were analyzed with a Thermo Ultimate 3000RS (Thermo Fischer Scientific, Waltham, MS, USA) chromatography system equipped with a diode array detector (DAD) and corona-charged aerosol detector (CAD), and coupled with a Bruker Impact II HD (Bruker, Billerica, MA, USA) quadrupole-time of flight (Q-TOF) mass spectrometer (MS). Chromatographic separations were carried out on an HSS C18 column (100 × 2.1 mm, 1.7 µm, Waters, Milford, MA, USA) at 40 • C. The injection volume was 3 µL. Mobile phase A was 0.1% (v/v) formic acid in MilliQ water, and mobile phase B consisted of acetonitrile containing 0.1% (v/v) of formic acid. The sample was separated using a linear gradient from 2 to 50% of solvent B in solvent A (0.4 mL/min, 13 min).
The UV absorbance was measured in the range of 200-600 nm (5 nm bandwidth). The acquisition frequency of both the DAD and CAD detector was set to 10 Hz. The MS analysis was performed in both ESI(-) and ESI(+) ion modes, using the following settings: scanning range 50-1800 m/z; negative ion capillary voltage 3.0 kV; positive ion capillary voltage 4.5 kV; dry gas flow 6 L/min; dry gas temperature 200 • C; nebulizer pressure 0.7 bar; collision RF 750 Vpp; transfer time 100 µs; prepulse storage time 10 µs. The collision energy was set to 20 eV. The results were calibrated internally with sodium formate injected into the ion source at the beginning of separation. Data processing was performed using Bruker DataAnalysis 4.3 software.

Quantitative LC-UV Analysis of Flavonoids and Phenolic Acids
The flavonoid and phenolic acid contents in the dandelion extracts and fractions were determined using an ACQUITY UPLC system (Waters) equipped with a photodiode array detector (PDA) and a tandem quadrupole (TQD) mass spectrometer. Chromatographic separations were carried out on an HSS C18 column (100 × 2.1 mm, 1.7 µm, Waters), using parameters identical to those described in Section 4.4.1. The injection volume was 2.5 µL. The UV spectra were recorded within the range of 190-490 nm (3.6 nm resolution).
The lyophilized samples were dissolved in 80% methanol at a concentration of 2 mg DW/mL (two replicates were prepared for each sample) and appropriately diluted before chromatographic analyses.
Flavonoids were detected at UV 345nm and phenolic acids at UV 320nm . Quantitative determinations were based on an external standard method, using two group standards: luteolin and L-chicoric acid (the main flavonoid and phenolic acid of dandelion fruits, respectively). Both reference compounds had been previously isolated by our laboratory and their purity was determined by LC-MS analysis: L-chicoric acid (95%) [16], and luteolin (99%).
Both the luteolin and L-chicoric acid calibration curves were prepared in six concentrations, ranging from 0.5 to 150 µg/mL, and showed good linearity (R 2 ≥ 0.999). Three injections were performed for each sample/standard working solution. Quantitative results were expressed as mg standard (L-chicoric acid or luteolin) equivalents (eq)/g of extract/fraction.

Determination of Total Phenolic Content (TPC)
The total phenolic content in all samples was determined with the Folin-Ciocalteu assay, as described previously [16]. Briefly, 100 µL of F-C reagent was added to 1600 µL of an appropriately diluted sample (6-25 µg/mL) or gallic acid standard solution (six concentrations in the range between 0.5-8 µg/mL). After adding 300 µL of Na 2 CO 3 (20% w/v), the mixture was incubated in a water bath (40 • C) for 30 min. The absorbance was measured at 765 nm against a blank sample using an Evolution 260 Bio spectrophotometer (Thermofisher Scientific). The TPC of samples was read from the linear curve for gallic acid (R 2 > 0.999) and expressed as milligrams of gallic acid equivalents (mg GAE/g DW).

DPPH Free Radical Scavenging Activity
The radical scavenging activity of the extracts and fractions against the DPPH free radical was determined according to Brand-Williams [42], with slight modifications as described by Lis et al [17]. Briefly, 1900 µL of DPPH methanol solution (100 µM) was mixed with 100 µL of the sample (four different concentrations in the range between 10-500 µg/mL) or Trolox solution (Sigma-Aldrich; five different concentrations in the range between 10-200 µg/mL) in a cuvette. After 30 min, the absorbance was measured against methanol (blank) at 517 nm using the Evolution 260 Bio spectrophotometer (Thermofisher Scientific). The percentage of absorbance inhibition was calculated from the equation: Inhibition (%) = 100 × [(A blank − A sample )/A blank ], where A blank and A sample are the absorbance values of the blank and test samples at t = 30 min, respectively.
To calculate the Trolox Equivalent (TE) of samples on DPPH, the slope of the sample linear curve, i.e., the absorbance inhibition (%) vs. concentration (µg/mL), was divided by the slope of the standard linear curve.
The IC 50 value of extracts and fractions, defined as the concentration of sample necessary to cause 50% inhibition was determined from the sample linear curves as absorbance inhibition (%) vs. concentration (µg/mL), with Trolox used as a positive control.

Blood Platelets and Plasma Isolation
Human blood was obtained from a Medical Center in Lodz (Poland), and the biological material came from regular, non-smoking/non-drinking alcohol and medication-free donors. The blood was collected into tubes with citrate/phosphate/dextrose/adenine (CPDA) anticoagulant. All experiments were approved by the University of Lodz Committee for Research on Human Subjects and carried out under permission number 2/KBBN-UŁ/II/2016.
Plasma and blood platelets were isolated from fresh human blood by differential centrifugation as described previously [16]. The platelet pellet was suspended in modified Tyrode's buffer (pH 7.4). The number of platelets in suspensions used in the experiments was determined at 800 nm using a UV-Visible Helios α spectrophotometer (Unicam) according to Walkowiak et al [43].; the amount was found to be 1.5-2.5 × 10 8 /mL. The protein concentration was calculated by measuring the absorbance of tested samples at 280 nm according to Whitaker and Granum [44]. Each sample (both plasma and blood platelets) taken for testing was obtained from different subjects and could be regarded as an independent trial.

Incubation of Plasma, Blood Platelets, and Whole Blood with Plant Extracts and Fractions
Stock solutions of the T. officinale L. fruit extracts and fractions were prepared with 50% DMSO (v/v). The final concentration of DMSO in test samples was lower than 0.05%, and its effects were determined in all experiments. The extracts and fractions were then taken for use in different models to evaluate the antithrombotic properties. Briefly: The plasma was pre-incubated for five minutes at 37 • C with dandelion extracts (E1-E3) and fractions (A-D) at two concentrations, 10 and 50 µg/mL, and then treated with 4.7 mM H 2 O 2 /3.8 mM Fe 2 SO 4 /2.5 mM EDTA (25 min, at 37 • C).
The blood platelets were incubated for 30 min at 37 • C with dandelion extracts (E1-E3) and fractions (A-D) at final concentrations of 10 and 50 µg/mL.
The blood platelets were pre-incubated (25 min at 37 • C) with dandelion extracts (E1-E3) and fractions (A-D) at two concentrations, 10 and 50 µg/mL, and then treated with thrombin at a final concentration of 5 U/mL (5 min at 37 • C).
Whole blood was incubated (30 min at 37 • C) with dandelion extracts (E2 and E3) and fraction A at final concentrations of 10 and 50 µg/mL.

Platelet Adhesion to Fibrinogen
Adhesion of blood platelets to fibrinogen was determined colorimetrically by measuring acid phosphatase activity [45]. Blood platelets were incubated in microtiter plates that had been precoated with fibrinogen, then nonadherent platelets were washed out and dissolved with Triton X-100. The details of the procedure were described previously [16,17]. Finally, 2M NaOH was added and the p-nitrophenol produced was measured at 405 nm, using SPECTROstar Nano Microplate Reader (BMG LABTECH, Germany). A control group was platelets without the addition of plant preparations; and its absorbance was expressed as 100%.

Flow Cytometry
Changes in the activation and reactivity of resting and stimulated blood platelets were studied in the whole blood model [46]. Fresh blood samples (150 µL) were incubated with dandelion fruit preparations (30 min., 25 • C), and then stimulated with 10 and 20 µM ADP for 15 min at room temperature (RT), or collagen (10 µg/mL, 15 min, RT). After incubation, the samples were diluted tenfold (1:9) in sterile PBS with Mg 2+ and stained with anti-CD61/PerCP, anti-CD62/PE, or PAC-1/FITC antibodies (30 min, RT, in the dark). Appropriate isotype controls were then prepared: the resting blood samples were stained with anti-CD61/PE and isotype control antibodies marked with FITC or PE. Finally, all samples were fixed with 1% CellFix (60 min, 37 • C).
The platelets were counted using an LSR II Flow Cytometer (Becton Dickinson, San Diego, CA, USA) based on the fluorescence of 10,000 platelets (CD61/PerCP positive objects). The platelets were distinguished from other blood cells by a forward light scatter (FCS) vs. side light scatter (SSC) plot on a log/log scale (first gate) and by positive staining with monoclonal anti-CD61/PerCP antibodies (second gate). The percentages of CD62P-positive and PAC-1-positive platelets were calculated relative to the total number of platelets (CD61-positive cells) presented in each sample. Non-specific antibody binding was determined using the isotype control antibodies IgG1/PE and IgM/FITC. All results were analyzed using BD FACSDiva software (Becton Dickinson, San Diego, CA, USA).