Key words
aggregation - arachidonic acid - cyclooxygenase - flavonoid - platelet - thromboxane
Abbreviations
AA:
arachidonic acid
ASA:
acetylsalicylic acid
COX-1:
cyclooxygenase-1
EDTA:
ethylenediaminetetraacetic acid
PRP:
platelet rich plasma
Introduction
Platelets are essential components of the human blood responsible for rapid blood
coagulation during injuries. On the other hand, excessive platelet aggregation is
associated with cardiovascular diseases, in particular with the risk of serious or
fatal coronary heart disease. Because both decreased and increased platelet aggregation
are associated with pathological states, the process of aggregation has to be tightly
controlled. The homeostasis is very complex and scientists have not been able to precisely
define the process in all aspects up to date [1]. There are several pro-aggregatory factors which enhance thrombus formation. The
central role in this cascade appears to be associated with the release of AA from
the cytoplasmatic membrane and its transformation into prostaglandin H2 by platelet COX-1, with further metabolism into thromboxane A2 by platelet thromboxane synthase. This process is stimulated in particular by collagen,
but other inducers like ADP and thrombin play a role as well [2], [3]. Despite this fact, most of the current clinically used drugs are irreversible inhibitors
of COX-1, like ASA, and antagonists on ADP receptors (clopidogrel, ticagrelor, prasugrel)
[4]. There are no other drugs acting on other levels of the AA cascade in clinical settings.
Such drugs blocking thromboxane synthase or acting as antagonists on thromboxane receptors
[5] might enrich the palette of clinically useful drugs in the future. Since they act
specifically on the described steps of platelet aggregation, they may have fewer side
effects in comparison to inhibitors of COX-1, which is an enzyme with many essential
physiological roles.
Flavonoids are promising candidates to be both, natural modulators of the disruption
of platelet aggregation, and antiplatelet drugs. There is a vast amount of literature
on the subject showing that many flavonoids are potent inhibitors of platelet aggregation
induced by molecules such as collagen, AA, thromboxane receptor agonist U-46619, ADP,
and epinephrine [6], [7], [8], [9], [10]. The data on thrombin-induced aggregation are controversial, some claim that flavonoids
have no effect [9], [11], other report positive effects [12]. Additionally, the inhibition of platelet aggregation by flavonoids is a reversible
process [9], which is another important factor when considering possible side effects of the
current antiplatelet therapy. Although many mechanisms responsible for antiplatelet
effects have been proposed [13], only a few are documented by published studies. In particular, their effects on
the AA-based aggregation cascade are well documented [10], [14], [15] but there are several discrepancies ensuing from the use of different protocols
(e.g., animal or human platelets, different analytical procedures, use of washed platelets,
platelets in plasma, or in the whole blood). Nevertheless, flavonoids may potentiate
the effects of ASA via different mechanisms of action [15].
A comprehensive investigation between the flavonoid structure and the effects on the
AA cascade including COX-1, thromboxane synthase and thromboxane receptors is still
missing. In addition, the effects on thromboxane synthase have been analyzed only
indirectly so far. Therefore, this study was aimed at delivering a detailed analysis
of 29 flavonoids ([Fig. 1]), representing the most commonly found natural flavonoids, and some relevant synthetic
congeners in order to establish structure-activity relationship on three consecutive
steps of the AA-based platelet aggregation using human platelets.
Fig. 1 Chemical structure of flavonoids tested in this study. Glc: glucose, Rha: rhamnose,
Glu: glucuronic acid.
Results
Initially, all flavonoids were tested for their possible effects on the inhibition
of ovine COX-1 in a concentration of 100 µM and compared to ASA. ASA was moderately
active at this concentration, the isoflavones genistein and daidzein were more potent
inhibitors (p < 0.01), while all other flavonoids were essentially inactive ([Fig. 2 A]). Considering these surprising results, we retested ASA and the active isoflavonoids
for their concentration-dependent effects. Herein, genistein was more potent at lower
concentration again but comparable to ASA in higher concentrations. In the case of
daidzein, a threshold effect at about 40 % was found ([Fig. 2 B]). To ascertain if this could be valid for humans, human platelet suspension in plasma
was used. In these physiological conditions, ASA showed an excellent effect in units
of µM and completely inhibited COX-1 in higher concentrations. Both isoflavonoids
were significantly less active, they did not reach full inhibition even at high, pharmacologically
unachievable, concentrations but their activity in units of µM could have a real clinical
relevance ([Fig. 2 C]). The activities of daidzein and genistein were similar in this set of experiments.
Fig. 2 Effects of flavonoids and ASA on COX-1. A Comparison of the effects of all tested flavonoids on ovine COX-1. Grey area shows
the error of the method. Compounds were tested at a concentration of 100 µM, several
flavonoids were retested at higher concentrations, too (7-hydroxyflavone, taxifolin,
luteolin, hesperetin, luteolin and taxifolin at 200 µM and troxerutin at 400 µM).
B Concentration-effect curves for effective isoflavonoids and ASA on inhibition of
ovine COX-1. C Concentration-effect curves for effective isoflavonoids and ASA on inhibition of
human plaletet COX-1 in plasma.
The second step of the AA cascade is the transformation of prostaglandin H2 into thromboxane A2 via thromboxane synthase. All flavonoids were tested for their effect on this enzyme
and compared to the known inhibitor 1-benzylimidazol. At the concentration of 100 µM,
only three flavonoids showed more than 25 % inhibition, but all of them were clearly
less potent than 25 µM of 1-benzylimidazole ([Fig. 3 A]). These partly active flavonoids, 7-hydroxyflavone, apigenin and epicatechin, were
additionally tested for a concentration-dependent effect. From these data ([Fig. 3 B]) it was clear that all three flavonoids were similarly active, and were able to
markedly block the enzyme at concentrations which are pharmacologically not relevant,
and were more than about 1 order less efficient (IC50 in the range of 151 to 226 µM) compared to 1-benzylimidazol with an IC50 of 8 ± 1 µM.
Fig. 3 Effects of flavonoids and 1-benzylimidazol on thromboxane synthase. A Comparison of all compounds at the concentration of 100 µM. For comparison, 25 µM
of 1-benzylimidazol (BI 25) is shown. Grey area means the error of the method. All
partly active flavonoids were less potent than 1-benzyimidazol but there were no significant
differences among them. B Concentration-effect curves of partly active flavonoids.
The last well-known part of the aggregation cascade is the activation of thromboxane
A2 receptors. In this assay, we used the stable thromboxane A2 receptor agonist U-46619. Again, all flavonoids were tested. Notwithstanding the
equal platelet concentration used in this study, the effect of U-46619 on platelet
aggregation was highly variable, and therefore, we used a two-step calibration in
order to ascertain reproducibility of the results: first, the used concentration of
U-46619 in the range of 0.75–1.5 µM had to produce more than 90 % aggregation, and
second, epicatechin at the concentration of 300 µM had to decrease the aggregation
to 15–30 %. The need of the final concentration of U-46619 in the mentioned range
suggests that this analogue has a lower affinity to thromboxane receptors in comparison
to the endogenous substrate thromboxane A2, but it is well known that it has a much better stability. Due to these reasons,
it is not easy to determine the clinical relevance of the inhibitory concentrations.
Herein, we tested all compounds up to a concentration of 300 µM. If no inhibition
of platelet aggregation was achieved at this high concentration, the compounds were
considered inactive. According to their activity, we divided the flavonoids in 3 classes
([Table 1]). The active flavonoids were then compared according to two criteria, i.e., IC50 and the effect at a concentration of 100 µM ([Fig. 4 B, C]), because their anti-aggregatory curves had different shapes ([Fig. 4 A]). Although there were some differences between active flavonoids in terms of IC50 ([Fig. 4 B], daidzein and apigenin-7-glucoside were the most potent), there were insignificant
differences among all five active flavonoids concerning their effect at a concentration
of 100 µM.
Fig. 4 Inhibition of human platelet aggregation induced by thromboxane agonist U-46619 by
flavonoids. A Examples of antiaggregatory curves of three active (iso)flavonoids (7-hydroxyflavone,
daidzein and genistein), moderately active epicatechin and slightly active myricetin.
Comparison of IC50 values (B) and the percentage of the anti-aggregatory effect at a concentration of 100 µM (C) of flavonoids which had some effect on U-46619 induced aggregation (according to
[Table 1]). IC50 of slightly active flavonoids (kaempferol and quercetin) were only assessed, naringenin
is not shown since the calculation of an IC50 failed.
Table 1 Classification of flavonoids according to their potential to inhibit aggregation
induced by thromboxane receptor agonist U-46619.
|
Class
|
Effect range
|
Flavonoids
|
|
* genistin activity was on the border between active and moderately active
|
|
Inefficient
|
no effect at 300 µM
|
baicalein, baicalin, flavone, hesperidin, hesperetin, 3-hydroxyflavone, 5-hydroxyflavone,
mosloflavone, negletein, luteolin, rutin, naringin, taxifolin and troxerutin
|
|
Slightly efficient
|
effect at 300 µM but no effect at 150 µM
|
kaempferol, myricetin, naringenin and quercetin
|
|
Moderately active
|
effect at 150 µM but no effect at 100 µM
|
catechin, diosmin, epicatechin, chrysin, genistin* and morin
|
|
Active
|
effect bellow 100 µM
|
apigenin, apigenin-7-glucoside, daidzein, genistein and 7-hydroxyflavone
|
Discussion
Several epidemiological studies have suggested protective effects of flavonoids against
cardiovascular diseases. These effects may be linked to their influence on platelets
as can be implied from the lower incidence of ischaemic stroke observed with increasing
flavonoid consumption [16], [17], [18], [19], [20], [21]. Although positive effects of flavonoids in experimental models of thrombosis have
been documented in animal studies [22], [23], clinical data are still missing and the few available small human studies have
reported no effect of flavonoids on platelet aggregation in healthy volunteers [12], [24], [25]. Thus the current knowledge does not enable an unambiguous conclusion. In particular,
the following issues should be considered: 1) epidemiological data are based on a
food questionnaire, such estimates of flavonoid consumption represent a very imprecise
indicator of flavonoid plasma levels and in addition, patient medication cannot be
analyzed in detail, 2) small clinical evaluations have been performed on healthy volunteers
and the fact that flavonoid consumption has not modified normal human platelet aggregation
can be considered as a positive fact, 3) there is high variability in flavonoids pharmacokinetics
and, as well, pharmacodynamic effects [3]. The variation in response to the thromboxane receptor agonists was observed in
our study too, notwithstanding PRP from different donors was normalized to the same
platelet concentration. Moreover, artificial animal models of platelet thrombus injuries
may not sufficiently mimic the clinical situation.
Data from ex vivo and in vitro studies are much more convincing. A number of known flavonoids has been shown to
possess antiplatelet effects [3], [6], [8], [9]. Although many mechanisms, including inhibition of phospholipase A2, phosphodiesterases and/or protein kinases, have been reported [12], [26], [27], [28], [29], the mechanism of action, which seems to be common for the majority of flavonoids
and in some of them within the range of achievable plasma concentrations, appears
to be the inhibition of the AA-based pathway of platelet aggregation [10], [14]. Indeed, flavonoids are able to decrease aggregation stimulated by different inducers.
Particularly, flavonoids seem to be potent inhibitors of aggregation caused not only
by AA, but also by collagen, which is known to play the key role in phase 1 of (patho)physiological
platelet aggregation [30]. In addition, certain flavonoids, rather at higher concentrations, may block aggregation
induced by ADP or thrombin [8], because the AA-based pathway is only one part of the complex mechanism of platelet
activation evoked by endogenous signal molecules mentioned above. However, as described
earlier, it is reasonable to think that several flavonoids may possess additional
effects on platelet aggregation beyond their influence on the AA-based pathway due
to the structural diversity of this group.
In this study we concentrated on the AA-based pathway which plays likely the key role
in the effect of a majority of flavonoids. AA is, through a two-enzyme mediated reaction
(COX-1 and thromboxane A2 synthase), transformed into a potent platelet aggregation inducer, thromboxane A2, which stimulates its own receptors, resulting in platelet aggregation. We have critically
assessed the effects of flavonoids on all three levels of the AA-based aggregation
cascade. Similarly to our recent study, where we have shown that some coumarins may
affect two steps in this cascade [31], we documented that the isoflavonoids genistein and daidzein acted both as inhibitors
of COX-1 and as functional antagonists at thromboxane A2 receptors. In addition, a number of flavonoids are able to act as thromboxane receptor
antagonists ([Fig. 5]). On the contrary, although three flavonoids were able to block thromboxane A2 synthase, the concentration necessary for this effect seems to be too high and thus
it is very likely that inhibition of thromboxane A2 synthase by flavonoids has no clinical importance. One may speculate that the functional
antagonism at the thromboxane A2 receptors was achieved in this study with rather high concentrations of flavonoids.
This is true, but that assay uses a high concentration of the thromboxane A2 receptor agonist U-46619 as well, which is a stable compound when comparing to very
unstable endogenous thromboxane A2. The stability aspect might be important. Even a transient weak blockade of thromboxane
A2 receptors by flavonoids might hinder the aggregation caused by thromboxane A2 because of its rapid metabolism. In contrast, in the case of U-46619, a competition
between this stable agonist and flavonoids may take place for the receptors. In the
literature, there are few reports showing that flavonoids may influence the formation
of thromboxane A2 and act as functional antagonists at the thromboxane A2 receptors [8], [9]. Interestingly, in washed platelets, some flavonoids were able to block thromboxane
A2 formation from AA in µM concentrations [8]. This result is not easily explainable since other studies have found substantial
inhibitory effect of flavonoids only at higher concentrations [9], [11], [32]. It is known that the antiplatelet effect in washed platelets can be achieved in
lower concentrations than in PRP [11] and interindividual differences cannot be excluded either. The latter can be supported
by this study where genistein and daidzein reached partial inhibition of human platelet
COX-1 at lower concentrations than in the case of the recombinant ovine COX-1 enzyme,
and also by another study, where quercetin blocked COX-1 from bovine platelets with
an IC50 of 8 µM [33]. It should also be mentioned that the activation of isolated rat peritoneal leucocytes
with ionophore A23187 leads to thromboxane A2 formation, which may be inhibited by several flavonoids at very low concentrations
as well [34], [35], but here, flavonoids seemed to act upstream of the AA metabolism.
Fig. 5 AA-based platelet aggregation and sites of effects of active flavonoids. Thromboxane
A2 formed by this AA-based pathway may stimulate its receptors both on the same platelet
or of other platelets. Proposed mechanisms of action of isoflavonoids are shown. PGH2: prostaglandin H2; TxA2: thromboxane A2. (The figure was redrawn from our previous article [31].) (Color figure available online only.)
As far as we know, no results on the direct inhibition of thromboxane A2 synthase by flavonoids have been published. One possible reason is the lack of an
effect and ensuing difficulties in the publication of negative data. Older reports
suggested that flavonoids decrease thromboxane A2 levels indirectly mainly by inhibition of COX-1 in line with our data [32]. The available data on COX-1 inhibition are more in accordance with this study and
with previous reports. In a study comparing the effect of different flavonoids and
isoflavonoids in microsomal suspension of COX-1, genistein was the most active. Interestingly,
daidzein was markedly less active, but its IC50 was still lower than that of ASA [36].
The essential structural features for both activities are reported here for the first
time ([Fig. 5]). For COX-1 inhibition, an isoflavone ring with a 7-hydroxyl group was necessary,
while for being an antagonist at thromboxane A2 receptors, the free 7-hydroxyl group is not absolutely needed, since its blockade
in apigenin-7-glucoside did not abolish the effect. On the other hand, the presence
of glucose at position 7 in isoflavones (genistin vs. genistein) decreased the effect.
The position of ring B (isoflavones vs. corresponding flavones) is important for inhibition
of COX-1 but not for antagonism at thromboxane receptors, where there was insignificant
difference between isoflavone genistein and the corresponding flavone apigenin. The
catechol ring B, absence of the 4-keto group or the 2,3-double bond and the presence
of the 3- or 6-hydroxyl group were factors decreasing the activity. It is not known
if effects on both levels of the AA-aggregation pathway might be an advantage, because
apigenin and genistein were similarly active in a majority of studies. The superiority
of the antiplatelet effect of apigenin and genistein over other flavonoids has been
reported previously [9], [15], [25], [26], [37]. On the other hand, one study did not find any effect of apigenin on platelet aggregation
[11].
Although in vitro and ex vivo data appear to support epidemiological studies, the way to the final confirmation
of a positive effect of flavonoids on platelet aggregation and its effects in pathological
conditions needs additional studies. In particular, when analyzing the influence of
flavonoids on humans in ex vivo or in vitro conditions, one of the most important sources of errors is to overlook the pharmacokinetics
of flavonoids. There are three major kinetic factors: firstly, the oral bioavailability
of flavonoids is low; secondly, the absorption of non-cleaved flavonoids leads to
a conjugation mainly with glucuronic acid and sulphate; and third, non-absorbed flavonoids
are cleaved by intestinal bacteria into phenolic acids which may have a relevant effect
on the human being as well [37]. The outcome is that the oral administration of flavonoids results in a very low
concentration of total non-cleaved flavonoids in the plasma in the µM range at maximum,
and the concentration of unconjugated or non-metabolized flavonoids may be negligible.
This clearly does not mean that flavonoids cannot have clinical effects, but that
the transfer of in vitro and ex vivo data into the clinical setting has to be critically evaluated. Interpretation may
be additionally complicated by human deconjugation enzymes in tissues. For quercetin,
the concentrations of free aglycone may be higher in tissue than in plasma [38], [39].
Another interesting finding is that the derivation of the free 7-hydroxyl group with
a sugar moiety did not decrease the antagonistic effect on thromboxane A2 receptors in the case of apigenin. This result was surprising, therefore we repeated
the experiments with platelets from two additional donors but the results were essentially
similar. This fact is quite important from two aspects. First, it suggests that flavonoids
were acting directly on the receptor because the duration of the sample incubation
for the thromboxane receptor assay was only 2 minutes. Hence, it is not highly probable
that a glucoside easily penetrates into platelets or is rapidly and very efficiently
metabolized into the aglycone. Secondly, this may implicate that human metabolism
phase II (conjugation of flavonoids) could not abolish the effect of the flavonoids
because the chemical difference between glucuronic acid and glucose is not substantial.
At this moment, we do not have in vivo data supporting this hypothesis, in particular due to only limited number of flavonoids
human metabolites available commercially as conjugates with a glucuronic acid and/or
a sulphate. This hypothesis needs to be tested in future. Interestingly, the presence
of a glucose unit at C-7 of genistein reduced the effect, but did not completely abolish
it. Therefore, the interaction of the receptor with flavonoids appeared to be very
specific. This is emphasized by the fact that 7-hydroxyflavone and apigenin were very
active, but chrysin, possessing a similar structure to apigenin but lacking the 4′-hydroxyl
group, was only moderately active. As expected, the addition of a hydroxyl group in
position 6 fully abolished the effect. The same is true for the introduction of another
hydroxyl group in ring B (luteolin vs. apigenin).
In conclusion, this study confirmed the previous finding that flavonoids can affect
platelet aggregation through the AA-based pathway. Their effect seems to be mediated
mainly by antagonism on thromboxane A2 receptors. No clinically relevant inhibition of thromboxane A2 synthase is suggested. Genistein and daidzein from the isoflavone subgroup blocked
COX-1 as well and their effect was partly comparable to that of ASA. Since both above
mentioned isoflavones are common components of the human diet, in particular in people
consuming soy products, they may have a positive impact on human platelet aggregation
and thus on cardiovascular diseases associated with enhanced platelet activity.
Materials and Methods
Materials
AA was purchased from Chrono-Log Co. and sodium citrate solution from Biotika. Thromboxane
B2 EIA kit, prostaglandin H2, U-46619 and the COX inhibitor screening assay kit were purchased from Cayman Chemical
Company. Genistin and apigenin-7-glucoside were purchased from Extrasynthese. Mosloflavone
and negletein were synthesized by a convergent synthesis starting from chrysin according
to the previous report [40] at the Sapienza University of Rome. All other flavonoids (minimal purity of 95 %,
[Fig. 1]), DMSO, EDTA, 1-benzylimidazole (99 % purity), indomethacin (> 99 % purity) and
ASA (> 99 % purity) were purchased from Sigma-Aldrich. 96 % ethanol was purchased
from Penta.
Blood samples from 31 healthy non-smoking volunteers were collected by venipuncture
into plastic disposable syringes containing 3.8 % sodium citrate (1 : 9, v/v). For
mechanistic experiments, the COX inhibitor indomethacin, or the thromboxane synthase
inhibitor 1-benzylimidazol, were immediately added to the collected blood at a final
concentration of 10 µM. All volunteers were health-workers who had not taken any drug
for at least 14 days before the blood collection and who had given informed consent
for the study. The study was performed under the supervision of the Ethical Committee
of Charles University in Prague, Faculty of Pharmacy in Hradec Králové (approval date:
November 12, 2012) and conforms to the Declaration of Helsinki.
PRP was obtained as a supernatant by centrifugation of the collected blood for 10 min
at 500 g (centrifuge MPW-360, MPW Med. Instruments). Platelet poor plasma was prepared by
centrifugation of the remaining blood for 10 min at 2,500 g. The platelet count was determined using a BD Accuri C6 flow cytometer equipped with
BD CFlow Software and adjusted to 2.5 or 3.5 × 108 platelets/mL according to the planned protocol with the use of autologous plasma.
Cyclooxygenase-1 inhibition
A commercial set from Cayman Chemicals [41], which does not give false positive results for antioxidants, was used for the evaluation
of COX-1 inhibition.
Shortly, ASA or flavonoids dissolved in DMSO (final concentration of DMSO was 2 %
v/v) were incubated with ovine COX-1 at 37 °C and AA (final concentration of 100 µM)
was added to the mixture to start the reaction. The formed prostaglandin H2 was measured following its reduction to prostaglandin F
2α
by stannous chloride and assessed by enzyme immunoassay. The percentage of inhibition
was related to the positive control with DMSO. Analogously, in additional experiments,
PRP with a platelet concentration of 3.5 × 108 per mL pretreated with 1-benzylimidazol, to block further metabolism of prostaglandin
H2, was used instead of ovine COX-1 for testing of inhibition of human COX-1.
Thromboxane A2 synthase inhibition
Thromboxane A2 synthase inhibition was evaluated according to the method of Chang et al. [42], with minor modifications. PRP containing indomethacin with a platelet concentration
of 3.5 × 108 per mL was incubated with the tested compounds for 3 min at 37 °C. After addition
of prostaglandin H2 (50 ng), the mixture was incubated for 5 min. The incubation was immediately terminated
by addition of chilled EDTA (2 mM) and the solution was centrifuged at 10,500 g for 2 min (centrifuge MPW-52, MPW Med. Instruments). The thromboxane B2 levels in the supernatant were measured using a thromboxane B2 EIA kit according to the instructions of the manufacturer.
Antagonism at the thromboxane receptors
Antagonism at the thromboxane A2 receptors was performed by turbidimetry using a Chrono-log 500-Ca aggregometer connected
to a computer (Aggro/Link software, Chrono-Log Co) according to a previously reported
method [31]. The turbidities of PRP and platelet poor plasma were measured as the controls.
In brief, PRP (500 µl, 2.5 × 108 per mL) was pipetted into a siliconized glass cuvette and stirred at 1,000 rpm with
a magnetic stirrer at 37 °C for 2 min in the aggregometer. The tested flavonoids,
as well as the standard drugs, were dissolved in DMSO to obtain a 10 % (w/v) concentration.
5 µL of a tested compound were then added to the reaction mixture at various concentrations
to obtain the concentration-response curves. Individual samples were incubated at
37 °C for 2 min. After the incubation period, platelet aggregation was induced by
the addition of U-46619, a stable agonist of thromboxane A2 receptors. The aggregation process was monitored for 5 min.
Statistical analysis
The differences between compounds were assessed by one-way ANOVA followed by Tukey
multiple comparison test. The differences between concentration-effect curves were
analyzed by use of 95 % confidence intervals. The percent inhibition of aggregation
at a concentration of 100 µM (Y100 μM) was calculated according to the following formula:
where max is the maximal inhibition of platelet aggregation expressed as per cent of platelet
aggregation, slope is the slope of the curve. IC50 has the common meaning.
Acknowledgements
This study was supported by the grant of the Czech Science Foundation (project No.
P303/12/G163). M.Ř. would like to thank Charles University in Prague (No. SVV 260
064). Special thanks to all volunteers donating their blood for this study. We would
like to thank Dr. Ilaria Proietti Silvestri and Dr. Paolo Bovicelli for synthetizing
mosloflavone and negletein as well.
