Blood flow controls coagulation onset via the positive feedback of factor VII activation by factor Xa
© Shibeko et al. 2010
Received: 17 August 2009
Accepted: 26 January 2010
Published: 26 January 2010
Skip to main content
© Shibeko et al. 2010
Received: 17 August 2009
Accepted: 26 January 2010
Published: 26 January 2010
Blood coagulation is a complex network of biochemical reactions, which is peculiar in that it is time- and space-dependent, and has to function in the presence of rapid flow. Recent experimental reports suggest that flow plays a significant role in its regulation. The objective of this study was to use systems biology techniques to investigate this regulation and to identify mechanisms creating a flow-dependent switch in the coagulation onset.
Using a detailed mechanism-driven model of tissue factor (TF)-initiated thrombus formation in a two-dimensional channel we demonstrate that blood flow can regulate clotting onset in the model in a threshold-like manner, in agreement with existing experimental evidence. Sensitivity analysis reveals that this is achieved due to a combination of the positive feedback of TF-bound factor VII activation by activated factor X (Xa) and effective removal of factor Xa by flow from the activating patch depriving the feedback of "ignition". The level of this trigger (i.e. coagulation sensitivity to flow) is controlled by the activity of tissue factor pathway inhibitor.
This mechanism explains the difference between red and white thrombi observed in vivo at different shear rates. It can be speculated that this is a special switch protecting vascular system from uncontrolled formation and spreading of active coagulation factors in vessels with rapidly flowing blood.
Blood coagulation is a complex reaction network that functions to form a fibrin clot that covers damaged vessel wall and prevents blood loss . The clotting process is initiated by tissue factor (TF), a transmembrane protein exposed in the damaged parts of the wall. This protein forms a complex called extrinsic tenase with plasma protein activated factor VII (VIIa). Extrinsic tenase activates factor X, which activates thrombin, the main protein of blood coagulation. Activated factor X (factor Xa) activates factor VII in complex VII-TF (inactive extrinsic tenase), thus forming a positive feedback. Extrinsic tenase is inhibited by tissue factor pathway inhibitor (TFPI) in a complex factor Xa-dependent manner . Thrombin forms fibrin, which polymerizes to create a clot.
Although reactions of the coagulation cascade are well known, and no new essential components of this system have been discovered over the last fifteen years , the present understanding of the functioning of this system is limited. The incredible biochemical complexity of coagulation additionally complicated by protein diffusion and blood flow makes it extremely difficult to establish a correlation between the roles of individual reactions and the functioning of the clotting system in vivo as a whole. For diffusion, a number of recent studies brought attention to the essential role of spatial non-uniformity and rate-limiting diffusion of specific coagulation factors thus proposing new concepts of clotting regulation, alternative to the classic "cascade" paradigm [1, 3–6]. This is not the case for blood flow.
A major role of flow in hemostasis and thrombosis was recognized as early as 19th century, flow being one of the components of the famous Virchow's triad . Two primary hemostatic mechanisms, platelet plug formation and blood coagulation, are known to differently depend on the flow conditions: platelet adhesion and aggregation require high blood flow velocities, while fibrin deposition occurs better in slowly flowing blood [8, 9]. Moreover, a recent report suggests that fibrin clot formation is inhibited by flow in a threshold-like manner . This can be illustrated in vivo by formation of fibrin-rich red thrombi containing erythrocytes in veins (where shear rate is low) and of platelet-rich white thrombi in arteries . However, the phenomenon of blood clotting inhibition by flow has not been studied in detail and was assumed to be a self-evident consequence of the removal of active coagulation factors from the site of vascular damage by flow.
In order to gain insight into this phenomenon, a modular decomposition strategy was used. We created a detailed quantitative mechanism-driven mathematical model of thrombus formation in flowing plasma. In agreement with experimental reports, clot formation in the model depended on blood flow shear rate in a threshold-like manner. Sensitivity analysis of this model was performed to identify reactions forming a module (or subsystem) within this biochemical network that was responsible for this effect. We demonstrated that a specially designed positive feedback of factor VII activation combined with chemical inhibition of extrinsic tenase and flow-induced removal of factor Xa becomes a "switcher" that can "decide": whether to start clotting or to abstain from it. It can be speculated that this switching can serve to prevent uncontrolled spreading of coagulation factor by flow throughout the vasculature.
This following study was concentrated on the most obvious and immediate effect of the flow, the increase of time between TF exposure and the clot appearance. In order to characterize this quantitatively, we introduced the parameter of lagtime, i.e. time required to cover half of the activator with high density fibrin clot (3,800 nM).
While the vast majority of model parameters are based on experimental data and have been thoroughly validated by comparison with experiments without flow , two assumptions were made in this study: specific values for fibrin gelation concentration and the fibrin level to calculate lagtime. In order to estimate the effect of these assumptions, we performed calculations at different values. For this, we introduced two numerical parameters characterizing control of clotting by flow: (overall inhibition of clotting by flow in the interval of shear rates i-j) and R' (non-linearity of this inhibition) as defined under Methods.
Effect of system perturbation on sensitivity to flow
i-j, s -1
Control, Fn gel concentration 380 nM
Control, Fn gel concentration 760 nM
Control, Fn concentration of solid clot 450 nM
Control, Fn concentration of solid clot 7600 nM
Control (activator size 500 μm)
No convection of zymogens
No convection of enzymes
No convection of enzymes except for IIa & Xa
No convection of IIa & Xa
No Xa-induced VII-TF activation
No IIa-induced activation of V, VII, VIII, XI
Addition of TFPI at 5 nM
Addition of AT-III at 6800 nM
Addition of VIIa at 10 nM
This was not the case with enzymes (blue line in Fig. 4a). When convection of all enzymes (i.e. factors IIa, Va, VIIa, VIIIa, IXa, Xa, XIa, activated protein C) was "switched off", clot formation within the total range of flow rates was exactly the same as for the zero shear rate. Neither lagtime (Fig. 4a) nor clot shape (data not shown) depended on wall shear rate. = 0 and R' could not be calculated at all, as the lagtime dependence on wall shear rate could be approximated with a line y = 2.5. The system became completely insensitive to flow. Thus, it is the removal of active factors by flow that changes the coagulation system behavior under flow conditions.
Further analysis carried out by analogy revealed specific enzymes responsible for the effect. "Switching off" convection of factors IIa and Xa only had the same effect as that of all enzymes taken together (magenta line in Fig. 4a, table 1). R' could not be calculated, and was very low (0.01). Additional analysis showed that individual contributions of factors IIa and Xa to the phenomenon were comparable (data not shown). When convection of all enzymes except for factors IIa and Xa (i.e. Va, VIIa, VIIIa, IXa, XIa, activated protein C) was turned off (dark cyan line in Fig. 4a, Table 1), the influence of flow on coagulation system was reduced approximately 2.6-fold ( = 0.39), while sensitivity to flow rate variation was reduced by a factor of 1.6 (R' = 0.63). This remaining difference was due to factor Va convection: turning on factor Va convection in this system made the difference with normal case minimal (data not shown). Summarizing, these simulations provide evidence that it is convection of factors IIa and Xa that determines the effect of system sensitivity to flow rate variation.
After identification of the specific factors, removal of which by flow is most important for the regulation of coagulation system behavior under flow conditions, we analyzed the reactions, wherein these factors participate by their removal from the system. Thrombin-mediated cleavage of fibrinogen and prothrombin activation by factor Xa are essential and critical components of the coagulation system, and their removal from the reaction scheme could not be attempted. Other important reactions involving these enzymes are positive feedback loops of factors V, VII, VIII, XI activation by thrombin, and of the factor VII-TF complex activation by factor Xa . We found out that "switching off" all positive feedbacks catalyzed by thrombin caused moderate increase of and almost did not change R' (blue line in Fig. 4b). Thus, although system was slightly better suppressed by flow, its sensitivity to flow rate variation remained on the same level (R' = 1.12). This was not the case for factor Xa-catalysed reactions: "turning off" the factor VII-TF complex activation (red line in Fig. 4b) made the system 3-fold less sensitive to flow rate variation (R' = 0.33), and it was suppressed by flow 7.7-fold better.
We propose the following explanation of observed phenomena: the onset of blood coagulation under flow conditions is governed by the positive feedback of extrinsic tenase activation by factor Xa. When flow does not influence this feedback (as it was in the case of no convection of factors Xa and IIa), the duration of clotting initiation phase becomes independent of flow rate. When the feedback does not function (factor Xa does not activate complex VII-TF to its active form, extrinsic tenase) flow suppresses coagulation, although this influence varies little as flow rate is increased. But in normal case, flow gradually switches off feedback extrinsic tenase activation. This not only causes lagtime increase, but also makes this increase non-linear.
Fig. 5b represents the kinetics of TF inhibition at different shear rates. It illustrates that flow actually prolongs the lifetime of extrinsic tenase. Summarizing, the calculations in Fig. 5 show that flow plays a dual role in the regulation of extrinsic tenase. On the one hand, extrinsic tenase formation is slowed down, and clot is slower to form. On the other hand, this protects TF from inhibition (it can be inhibited only in complex with VIIa) and thus flow can prolong the life time of extrinsic tenase and allow clot initiation later.
Although the "flow-dependent trigger" obtained in our model was in agreement with experimental evidence [10, 14, 15] and did not require any hypotheses about unknown reactions, its mechanism remains to be experimentally confirmed. The most direct way to do this would be to study coagulation in a flow chamber using mutant molecules of factors X and VII that would be unable to participate in some of the feedback reactions. However, there are simpler experiments that can be undertaken to test tis mechanism, and we attempted to predict them with the help of the model.
On the other hand, addition of 5 nM TFPI (3-fold increase of the normal plasma concentration) should result in a dramatic increase of suppression by flow ( = 5.55) (Fig. 6), and the system sensitivity to flow rate variation should also increase (R' = 4.5). TFPI would have very little effect in the absence of flow, similar to the lack of effect of factor VII activation without flow. Increase of TFPI concentration would lead to better inhibition of the feedback, thus system would be more sensitive either to flow rate variation and suppression by flow. We propose that TFPI concentration is a determining factor for the range of flow rate where the coagulation system can form a clot.
As a possible control, addition of 6800 nM antithrombin III (3-fold increase of the normal plasma concentration) should result in a very moderate increase of suppression by flow ( = 1.38), definitely much less than in the case of TFPI addition (Fig. 6), and should not change the sensitivity to flow rate variation sufficiently (R' = 1.22). So we propose that antithrombin III-induced regulation of the early stages of coagulation in flow is moderate.
In the presence of flow with low shear rates, more extrinsic tenase is needed to clot plasma because factor Xa is removed by flow. However, additional factor VIIa-TF complex produced by factor Xa from factor VII-TF complex can meet these requirements. Because of this positive feedback, blood coagulation is almost not sensitive to flow at low wall shear rates: with shear rate increase, lagtime increases very slowly. The region of this low sensitivity is determined by the balance of positive feedback of factor VII activation and TFPI-mediated inhibition: the lagtime is sensitive to these reactions in the presence of flow (Fig. 4). This result disagrees with the theoretical prediction of , where TFPI was found not to influence coagulation in flow at all. We suppose that this disagreement was caused by different ways of modeling TFPI inhibition .
Finally, at higher shear rates this mechanism is no more able to support clotting: factor Xa is removed very rapidly and is not able to activate factor VII-TF complex. Lag time becomes very long and is rapidly increasing as an exponential function of wall shear rate.
In other words, this concept can be summarized as follows: positive feedback activation of factor VII-TF complex by factor Xa, inhibition of factor VIIa-TF complex by TFPI and factor Xa removal by flow from the activation region together constitute a special controlling mechanism, which makes blood coagulation insensitive to flow at low shear rates, and almost completely switches clotting off at higher shear rates.
This concept agrees well with the theoretical prediction of Beltrami and Jesty that a biochemical system composed of two enzymes reciprocally activating each other in the presence of flow, one enzyme being membrane-bound and another being soluble, will depend on the wall shear rate in a threshold manner. A marked difference between that prediction and the result obtained here for coagulation system is that our, more complex system does not actually have a clear threshold: it is just a steep exponential dependence, which can, however, serve the same regulatory purposes.
Generally speaking, flow can exert its effects by either drag-in or drag-out of the components of coagulation. Previous studies somewhat disagree in the relative contribution of these phenomena. Experiments performed using a reconstituted system and a continuous flow reactor have shown that drag-in of rapidly consumable coagulation factor predecessors (such as factor VIII) can be very important and enhance coagulation . Computer simulations of fibrin polymerization  and enzyme catalysis on the channel wall  in flow suggested that it is the drag-out of these components which is the most important mechanism of flow action. Our calculations performed in a detailed model of coagulation system agree with the latter conclusion: removal of coagulation factors by flow is the predominant mechanism. Moreover, we were able to identify two most important factors that are removed, factor Xa and thrombin. Convection of other factors contributes negligibly.
Previous studies do not completely agree in their conclusions on flow effect in coagulation. Although there are reports that blood flow can have some beneficial effect on blood coagulation, at least in purified systems (for example, the rate of factor X activation by extrinsic tenase under high shear rate conditions is increased [18, 20]), it is more or less generally accepted that the overall effect of flow on clotting is inhibitory as can be seen from both flow chamber perfusion experiments and in vivo studies [10, 11, 15, 21, 22]. The mechanisms of this effect are not clear either. A recent mathematical modeling study has shown that flow can inhibit fibrin polymerization  and that this inhibition can give results similar to those obtained experimentally . The same group showed that flow-mediated removal of factors IXa and Xa inhibits the initiation of coagulation .
We studied the behavior of plasma coagulation depending on the wall shear rate in the range of 0-28 s-1, and found out that at rates higher than 17-28 s-1 plasma coagulation is almost inhibited by flow. It should be considered that this range of wall shear rates significantly depends on the activator size and the chosen mechanism of gelation. If we take into consideration that most researchers estimate physiological range of wall shear rates as 25-6500 s-1 according to  then we cannot escape the conclusion that the clotting system function is actually designed to minimize fibrin thrombi formation under physiological conditions. The problem of fibrin polymerization in flow requires further exploration, because it can essentially affect the range of shear rates where clotting is possible; the qualitative phenomenon, however, should not change because it is determined by flow effect on factors IIa and Xa.
Summarizing, our results suggest that roles of the initial reactions of the extrinsic pathway might be somewhat reconsidered. In particular, this is relevant for factor VII-TF complex activation by factor Xa and factor VIIa-TF complex inhibition by TFPI. In the absence of flow, these reactions do not have clear roles. Although they can influence clotting activated with diluted TF in vitro [13, 24], their contribution in the presence of TF highly concentrated on fibroblasts is negligible in the conditions of this study: formation of extrinsic tenase is sufficient without any need of additional factor VIIa, and inhibition by TFPI cannot significantly delay clotting . On the other hand, in the presence of flow, which removes factor Xa (both the product and the activator of extrinsic tenase) these reactions become critically important and completely shape the response of the system.
Removal of factor Xa by flow is the predominant mechanism of flow influence on the initial stage of blood coagulation. Together with TFPI-induced inhibition of extrinsic tenase complex, these mechanisms provide strongly non-linear response making coagulation insensitive to flow at low shear rates and almost preventing coagulation at higher rates. Although elucidation of these purposes is beyond the scope of this study, it might be speculated that this mechanism can protect the vascular system from formation of fibrin thrombi at high shear rates. However, when blood flow is reduced (e.g., as a result of platelet plug formation), the clotting in the area without flow can proceed rapidly.
A detailed mathematical model of blood coagulation cascade previously developed and verified by our group  for clotting of plasma without flow was used with modifications that included addition of a second dimension, convection terms, Navier-Stokes equations, and the process of fibrin polymerization as described below. The model consisted of 28 partial differential equations describing biochemical reactions, diffusion and convection of the reactants. All initial concentrations and kinetic constants used in the model were obtained from experimental reports, and no adjustment was performed.
The model simulated coagulation in a two-dimensional rectangle region (1,000 × 6,000 μm) with laminar flow. Activation patch (1,000 μm) covered with tissue factor (TF) was located on one of the walls of the simulated channel. When coagulation was activated, it led to fibrin network formation. We assumed the rate of fibrin polymerization to be sufficiently high so that each fibrin molecule was rapidly incorporated into the network; thus, we neglected the effect of flow on fibrin fibers formation. The permeability of this network was assumed to depend on fibrin concentration. While fibrin concentration was below the point of gelation, the permeability was 1, and it reduced to 0 when the fibrin concentration exceeded the critical concentration. Although impermeable to flow, fibrin gel in the model did not hinder diffusion  and was assumed to have no effect on the reactions of coagulation except for thrombin adsorption onto fibrin. The detailed description of the model is available in Additional file 3 and program realization of mathematical simulation based on this model is presented in Additional file 4.
was used as relative flow rate variation sensitivity coefficient.
The first coefficient indicates the integral flow induced lagtime growth within the certain range of flow rates, and the second coefficient indicates the increase of the rate of lagtime growing in the dependence lagtime versus flow rate. Based on values of these parameters, we could decide whether a certain part of coagulation system was sufficient for forming its behavior under flow conditions.
We thank Prof. A.I. Lobanov (Moscow Institute of Physics and Technology) for his kind help and suggestions with regard to numerical integration of Navier-Stokes equation. The study was supported by Russian Academy of Sciences Presidium Basic Research Programs "Molecular and Cellular Biology" (MCB RAS Russia) and "Basic Sciences for Medicine", by Russian Foundation for Basic Research grants 07-04-00146, 09-04-00232, 09-02-00018 and 09-04-92427.
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.