Flow-dependent channel formation in clots by an erythrocyte-bound fibrinolytic agent

Postsurgical patients at increased risk for both bleeding and thrombosis require safe short-term thromboprophylaxis.¹  Anticoagulants and antiplatelet agents are only partially effective, pose a risk of bleeding (eg, heparin),²  or have a delayed onset (eg, warfarin). Thrombolytic therapy using plasminogen activation is confined to emergencies and is not used for short-term postsurgical thromboprophylaxis because of rapid drug clearance and unacceptable postoperative hemorrhage risk.^1,3,4 

Coupling to a red blood cell (RBC) converts tissue-type plasminogen activator and urokinase-type plasminogen activator from short-acting therapeutics to durable thromboprophylactic agents by prolonging circulation times and preventing penetration into tissues and existing hemostatic clots while delivering PA to the interior of nascent thrombi. Prophylactic administration of RBC-tissue-type plasminogen activator in animals restores blood flow within minutes after formation of occlusive arterial and venous thrombi⁵  and has been used for cerebrovascular thromboprophylaxis⁶  without increased propensity for bleeding.⁷  A single injection of a PA fused to an antibody fragment binding to mouse RBCs precluded the need for infusion of modified RBCs and conferred immediate and lasting thromboprophylaxis, thereby safely preventing arterial and venous occlusion in mouse models of vascular injury.^8,9  Thus, short-term prophylaxis with RBC-PAs might prevent thrombotic occlusion while limiting the risk of bleeding.¹⁰ 

The mechanism by which RBC-bound PA mediates reperfusion is not intuitive because the enzyme is restricted to the cell surface. Previously, we used confocal microscopy of fluorescently labeled fibrin and RBCs to demonstrate that therapeutic lysis by soluble PA progressed gradually and uniformly, whereas prophylactic lysis by RBC-PAs embedded in clots occurred focally.¹¹  Pores formed and enlarged within the fibrin network around RBC-PAs. The RBC-PAs then became highly mobile within these pores and moved quickly around the edges of dissolving fibers, propagating fibrinolysis by gradually enlarging the pores. This process continued until merging of the pores resulted in complete dissolution of the bulk clot.¹¹ 

We posited that, in clots exposed to directional flow in blood vessels, these focal pores enlarge in the direction of flow and merge to form patent channels that permit blood components to traverse the clot before dissolution, thereby shortening ischemia. To test this hypothesis, we used confocal microscopy to monitor lysis and RBC passage through plasma clots containing RBC-PAs perfused with buffer at flow rates typical of venous and arterial vasculature.

All human and mouse blood samples were acquired and used according to accepted institutional review board and Institutional Animal Care and Use Committee protocols approved by review boards at the University of Pennsylvania. Clots were formed in Ibidi μ-slide VI^0.1 flow chambers by mixing fluorescently labeled human platelet-poor plasma with Vybrant (R) DiD-labeled mouse RBCs carrying 70 000 molecules of urokinase-type plasminogen activator per RBC (RBC-PAs, 2% final hematocrit), 20mM CaCl₂, and 0.1 U/mL human α-thrombin.^8,11  Clot formation was complete 16 minutes after addition of thrombin. The clots were then perfused with Dulbecco phosphate-buffered saline containing 0.01% bystander RBCs (no PA attached) and 5% platelet-poor plasma at venous (160 s⁻¹) or arterial (900 s⁻¹) shear rates. Images were acquired by laser scanning and spinning disk confocal microscopy (image acquisition information in supplemental Methods, available on the Blood Web site; see the Supplemental Materials link at the top of the online article).

Perfusion of buffer through a clot with incorporated RBC-PAs changed the mechanism of clot dissolution. Lysis of a static clot by incorporated RBC-PAs proceeded as previously described¹¹  with multilateral pore enlargement around RBC-PAs (Figure 1A-B; supplemental Movie 1). Bystander RBCs in the buffer placed at the clot edges did not enter the network during lysis. We plotted percent clot lysis, measured by the amount of fibrin network present at each time point, versus time (Figure 1C).

Exposing clots to 160 s⁻¹ of shear (venous) led to enlargement over time of network pores that formed predominantly along the direction of flow (Figure 1D-E; supplemental Movie 2). Although individual fibrin fiber degradation occurred at a similar rate as in the absence of flow, not all fibers dissolved (Figure 1C). However, the holes that formed between the remaining fibrin strands merged into channels aligned along the direction of flow, through which bystander RBCs from the perfusion buffer passed unimpeded (Figure 1F). Although the perfusion buffer was supplemented with 5% platelet-poor plasma to provide limited plasminogen replenishment, the bulk of the original plasminogen and some RBC-PAs were eventually washed out, leading to an experimental artifact of lysis arrest over time.¹²  At arterial shear (900 s⁻¹), patent channels formed even more rapidly and were quickly traversed by bystander RBCs (Figure 1G-I).

Therefore, within the short time interval tested (30-60 minutes), the final mass of fibrin lost under flow never exceeded 60% of the original mass. The initial clot dissolution rate was faster at higher shear but dropped off significantly once approximately 25% of the fibrin bulk was dissolved (Figure 1C). It was at this early time point when bystander RBCs began to perfuse the clot extensively (supplemental Movie 3).

We then studied the fibrin network structural changes triggered by RBC-PAs using a piezo stage controller to image at 3 successive locations along the direction of flow (Figure 2; supplemental Movie 4). In accordance with the permeability data shown in Figure 1, channel formation under shear stress occurred before clot dissolution (Figure 2C). A few bystander RBCs were trapped in the residual fibrin, but most passed through the patent channels so rapidly they could not be captured in the Z-projections shown, although they are evident in individual slices. Pores enlarged and merged in the downstream direction (Figure 2B). As individual fibers were cut leaving one end free, these free ends oriented in the direction of flow, leading to alignment of the remaining fibrin mass. This response to lysis under flow triggered by incorporated RBC-PAs is a potential mechanism for the channel formation we observed. In our previous study, we observed pore enlargement triggered by RBC-PAs¹¹  but without the directionality evident here.

These data indicate that a simple analysis of the loss of total fibrin network mass does not reveal some key, clinically relevant aspects of thrombolysis by RBC-PAs under flow. For example, although shear forces promote the formation of patent channels from initial holes surrounding individual PA-containing RBCs in all settings, the rate of complete clot dissolution by incorporated RBC-PA appears to be inversely related to shear (Figure 1C).

Under flow, rapid formation of channels in RBC-PA-containing clots allowed bystander RBCs to pass through a substantial mass of residual clot. Arterial shear provided faster channel formation versus venous shear, despite slower total clot dissolution. The channels observed resemble the fingerlike lysis patterns described by Blinc et al, who perfused free PAs through preformed whole blood clots,^13,14  although there are differences between the experimental systems in PA size, stability, diffusion, impact on flow, and origin (effect of RBC-PAs originates from within the clot rather than the perfusion buffer). The current outcome correlates well with fast reperfusion of venous and arterial clots facilitated by RBC-PAs in vivo⁸  and indicates that RBC-PAs rapidly form patent channels in otherwise occlusive clots at both arterial and venous shear. This feature of thromboprophylaxis by RBC-PAs may further accelerate reperfusion and salvage of ischemic tissue even before clot dissolution is complete.

The authors thank Ms Samira Ismail and Dr Ronald Carnemolla for mouse blood.

This work was supported by the National Institutes of Health: HL30954 and HL090774 (J.W.W.), AHA SDG-0535258N (S.Z.), HL-076406, HL-077760, HD-057355, and HL-077760 (D.B.C.), and RO1-HL090697 (V.M.).

National Institutes of Health

Contribution: K.C.G. performed and designed research, analyzed the results, and wrote the manuscript; and S.Z., D.B.C., V.M., and J.W.W. designed the research protocols, analyzed the data, and wrote the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: John W. Weisel, Department of Cell & Developmental Biology, University of Pennsylvania School of Medicine, 1054 BRB II/III, 421 Curie Blvd, Philadelphia, PA 19104-6058; e-mail: weisel@mail.med.upenn.edu.