(20a) A Microfluidic Model of Hemostasis Sensitive to Platelet Function and Coagulation

Hemostasis is the physiological process by which a vascular injury is sealed by the formation of a blood clot, or thrombus, while maintaining vessel patency. Several in vitro and in vivo models of intravascular thrombus formation have been developed. These models are relevant to thrombotic events such as those following atherosclerotic plaque rupture. However, relatively few models of extravascular thrombus formation exist, and these are relevant to bleeding diatheses and trauma.

Objective

The objective of our study is to develop an in vitro model of extravascular thrombus formation. The model should show the dependence of hemostasis on platelet function and coagulation in order to plug a hole in a model vessel under physiologic pressure gradients.

Methods

Microfluidic Device Design and Fabrication

The master template for the device was prepared with photoresists (KMPR 1010 and KMPR 1050) to define a two-layer device with heights of 20 µm and 50 µm. Polydimethylsiloxane (PDMS) was molded off of these masters and covalently bonded to glass using standard soft lithography procedures. The device was designed in the shape of the letter ‘H’ where the outer two vertical channels represent the vascular and extravascular compartments (10 mm long X 100 µm wide X 50 µm high), respectively. The vertical channels are connected by a horizontal channel (150 µm long X 50 µm wide X 20 µm high) representing a hole in the vessel wall, which we refer to hereafter as the injury channel. Type I fibrillar collagen, recombinant human tissue factor (TF), or a combination of the two were adsorbed to the walls of the injury channel. Bovine serum albumin (2%, BSA) in HEPES buffered saline (HBS), collagen, TF or collagen-TF solutions were allowed to adsorb on the walls of the injury channel at 4 °C for 12 h.

Blood Collection and Preparation

Blood was collected from healthy donors by venipuncture into vacutainer tubes containing 3.2% sodium citrate. Platelet counts and hematocrit were measured and recorded for each donor. 960 µL aliquots of blood collected in sodium citrate was combined in tubes with 40 µL of Alexa-555 labeled fibrinogen (final concentration 56 µg/mL) which was added to visualize fibrin deposition. Platelets were labeled with the lipophilic dye DiOC6 (1 µM final concentration). Labeled platelets were incubated at 37 °C in the presence or absence of 12 nM anti-FVIII antibody or 100 µM the P2Y12 antagonist 2-MeSAMP for 15 minutes prior to the assay.

Device Operation

Pressure was controlled for the blood, recalcification buffer, and wash buffer (3.2% sodium citrate in HBS) independently by applying a pressure to the headspace of their respective reservoirs using a pressure-based flow controller. Blood was recalcified in the ratio of 9:1 (citrated whole blood:recalcification buffer) using a herringbone mixer. The output from the herringbone mixer was connected to the blood channel of the extravascular injury device. The pressures of the blood and recalcification reservoirs were held constant at 10 kPa. The wash buffer reservoir was connected to the extravascular channel by. The pressure in the wash buffer was initially set to 3.5 kPa to drive a small amount of wash buffer through the injury channel while the blood channel was filled with recalcified blood driven by 10 kPa of pressure in the blood and recalcification reservoirs. The pressure in the wash reservoir was then reduced to 1.75 kPa so that the blood passes from the blood channel into the injury channel and out into the extravascular channel. Thrombus formation was monitored through an inverted microscope equipped with a 16 bit CCD camera using relief contrast and epifluorescence microscopy with a 20X objective for up to 45 minutes. Time to closure was defined as the first time that no red blood cells pass through the injury channel for 5 seconds.

Results

Type I collagen and TF and their effect on the time to closure of the injury channel are considered. Channels coated with BSA did not support platelet adhesion or significant fibrin deposition. This suggests that velocity gradients at the entrance and exit of the injury channel do not promote thrombus formation in the absence of agonist. Type I collagen alone supports platelet adhesion and fibrin accumulation in the near-wall-region underneath platelet aggregates. The initial thrombus formation is dictated by the position of collagen fibers in the injury channel and not by fluid dynamics in this model. There was no consistent location for initial platelet adhesion in the channel for collagen or collagen-TF surfaces. However, full closure is not achieved over 45 minutes, as the thrombus approaches closure, platelet aggregates embolize. TF alone leads to initial fibrin formation in the corners of the injury channel followed by platelet adhesion to the deposited fibrin. The time to closure of the horizontal channel is 15.8 ± 2.0 minutes. Type I collagen and TF together support platelet accumulation and fibrin formation with a time to occlusion of 7.5 ± 1.6 minutes. On the combined collagen-TF surface, platelets first adhere and aggregate to form a plug in the injury channel that stops the loss of blood into the extravascular compartment. Fibrin was observed within the dense platelet plug, particularly in the near-wall region. Following cessation of flow, fibrin rapidly polymerizes across the entire horizontal channel. These acquired results demonstrate that collagen and TF act synergistically to form a stable thrombus. Moreover, both primary hemostasis (formation of a platelet plug) and secondary hemostasis (stabilization by fibrin) are necessary.

Normal whole blood treated with a function blocking anti-human FVIII antibody did not form stable thrombi. Initial platelet accumulation was similar to that for the collagen-TF surface with control blood. However, as the thrombus approached closure large aggregates of platelets were shed. This process was repeated several times over the duration of the experiment. After 30 minutes there was small increase in platelet accumulation and blood continued to leak through the channel for up to 45 minutes. Fibrin accumulation was significantly reduced compared to controls. Moreover, fibrin accumulation was limited to the near-wall region where large platelet aggregates were attached likely protecting coagulation products from dilution by flow.

Normal whole blood treated with the P2Y12 antagonist 2-MeSAMP resulted in a prolonged time to closure (15.2 ± 3.4 minutes) compared to control blood on collagen-TF surfaces. The platelet plug was larger in area than in controls, typically filling the entire horizontal channel, and took longer to accumulate than for the untreated blood. Thrombi appeared less compact and contained less fibrin, potentially due to reduced platelet activation, and were less mechanically stable than in control blood as indicated by the displacement of large platelet aggregates in the direction of flow. Importantly, following closure as defined the arrest of red blood cells moving across the injury channel, we continued to observe leakage of plasma indicating a porous thrombus. No such leakage was observed in controls.

Conclusions

Hemostasis is a function of blood flow, platelet function, and coagulation. The described microfluidic model of hemostasis depends on all three of these phenomena. A thrombus forms in a small side channel to stop blood loss from one channel into another in a process that better represent hemostasis than similar models of intravascular thrombus formation. The presented geometry provides two unique features compared to conventional flow models. First, there are two paths for blood flow, namely; through the injury channel or the blood channel. Both are controlled independently. When the injury channel occludes, the flow rate through the blood channel remains unchanged. Moreover, plasma proteins and blood cells are continuously delivered to the periphery of the thrombus. Second the normal force is smaller in an extravascular injury compared intravascular injury for comparable pressure gradients and channel geometries. The sizes of the channels are similar to the microvasculature in humans. Most bleeds that occur in genetic bleeding disorders such as hemophilia, von Willebrand disease, or platelet disorders take place in the microvasculature. The wall shear rate in the blood channel and the pressure drop across the injury channel are comparable to those found in the arterioles. Normal whole blood in a collagen-TF coated injury channel recreates primary and secondary hemostasis with the initial formation of a platelet plug that is then stabilized with fibrin. The time to closure of the injury depends on both FVIII and autocrine signaling through the ADP receptor P2Y12, with deficits in either resulting in prolonged closure times and unstable or leaky thrombi. We described an experimental platform for studying hemostasis with a focus on collagen-TF induced thrombus formation within a model vascular wall. In the current configuration, the platform could be extended to study the effect of extravascular coagulation and the geometry of the extravascular space.

References

R. M. Schoeman, K. Rana, N. Danes, et al.. A Microfluidic Model of Hemostasis Sensitive to Platelet Function and Coagulation. Cellular and Molecular Bioengineering. 10:3-15, 2017. DOI: 10.1007/s12195-016-0469-0

M. Lehmann, A. M. Wallbank, K. A. Dennis, A. R. Wufsus, K. M. Davis, K. Rana, and K. B. Neeves. On-chip recalcification of citrated whole blood using a microfluidic herringbone mixer. Biomicrofluidics. 9, 064106:1-13, 2015. DOI: 10.1063/1.4935863